1: \documentclass[prl,twocolumn,superscriptaddress,showpacs]{revtex4}
2: \usepackage{graphicx}
3: \usepackage{amssymb,amsmath}
4: \usepackage[T1]{fontenc}
5: \begin{document}
6:
7: \title{Stripe phases - possible ground state of
8: the high-$T_c$ superconductors}
9:
10: \author{Marcin Raczkowski}
11: \affiliation{Marian Smoluchowski Institute of Physics, Jagellonian
12: University, Reymonta 4, PL-30059 Krak\'ow, Poland}
13: \affiliation{Laboratoire CRISMAT, UMR CNRS--ENSICAEN(ISMRA) 6508,
14: 6, Bld. du Mar\'echal Juin, F-14050 Caen, France}
15:
16: \author{Andrzej M. Ole\'s}
17: \affiliation{Marian Smoluchowski Institute of Physics, Jagellonian
18: University, Reymonta 4, PL-30059 Krak\'ow, Poland}
19:
20: \author{Raymond Fr\'esard}
21: \affiliation{Laboratoire CRISMAT, UMR CNRS--ENSICAEN(ISMRA) 6508,
22: 6, Bld. du Mar\'echal Juin, F-14050 Caen, France}
23:
24: \date{\today}
25:
26: \begin{abstract}
27: Based on the mean-field method applied either to the extended single-band
28: Hubbard model or to the single-band Peierls-Hubbard Hamiltonian we
29: study the stability of both site-centered and bond-centered charge domain
30: walls. The difference in energy between these phases is found to be
31: small. Therefore, moderate perturbations to the pure Hubbard model, such
32: as next nearest hopping, lattice anisotropy, or coupling to the lattice,
33: induce phase transitions, shown in the corresponding phase diagrams.
34: In addition, we determine for stable phases charge and magnetization
35: densities, double occupancy, kinetic and magnetic energies, and
36: investigate the role of a finite electron-lattice coupling. We also
37: review experimental signatures of stripes in the superconducting copper
38: oxides.
39: \end{abstract}
40:
41: \pacs{71.10.Fd, 71.27.+a, 74.25.-q, 74.72.-h}
42: \maketitle
43:
44: \section{\label{sec:1} Introduction}
45:
46:
47: Since the discovery of high-temperature superconductivity by Bednorz and
48: M\"uller \cite{Bed86}, the unusual physical properties of the copper
49: oxides have stimulated theorists and have led to the appearance of many
50: new ideas \cite{Car02}. One of the especially appealing new pictures
51: that has emerged is the instability towards a novel type of coexisting
52: incommensurate (IC) charge and magnetic order, i.e., stripe phase. As a
53: rare event in the theory of high temperature superconductivity, the
54: theory preceeded here the experiment and the existence of stripe
55: phases was predicted on the basis of Hartree-Fock (HF) calculations in
56: the two-band model for CuO$_2$ planes of layered La$_{2-x}$Sr$_x$CuO$_4$
57: (LSCO) \cite{Zaa89}, before their experimental confirmation. This
58: instability persists as well in the effective single-band Hubbard model
59: \cite{Poi89,Sch89,Kat90,Inu91}. All
60: these calculations yielded solutions with a phase separation manifested
61: in formation of nonmagnetic lines of holes, one-dimensional (1D) domain
62: walls or stripes, which separate antiferromagnetic (AF) domains of
63: opposite phases. Such states result from the competition between the
64: superexchange interaction, which stabilize the AF long-range order in
65: the parent Mott insulator, and the kinetic energy of doped holes. Indeed,
66: the magnetic energy is gained when electrons occupy the neighboring sites
67: and their spins order as in the N\'eel state, whereas the kinetic energy
68: is gained when the holes can move and the AF order is locally suppressed
69: along a domain wall (DW). Thus, a stripe phase provides the best
70: compromise between the superexchange promoting the AF order and the
71: kinetic energy of doped holes.
72:
73: However, the debate on the microscopic origin of the stripe instability
74: is far from closed. Two main scenarios, based on a Ginzburg-Landau free
75: energy, for the driving mechanism of the stripe phase have been discussed
76: \cite{Zac98, Kiv96}. In the first one, stripes are charge-density waves
77: with large periodicity arising from the Fermi surface (FS) instability
78: with the transition being spin driven \cite{Zaa89}. A general feature of
79: such an instability is a gap/pseudogap which opens up precisely
80: on the FS. Hence, the spacing between DWs is equal to $1/x$,
81: with $x$ denoting doping level so as to maintain a gap/pseudogap on the
82: FS. In this scenario spin and charge order occur at the same
83: temperature or charge stripe order sets in only after spin order
84: has developed.
85:
86: An alternative scenario comes from the Coulomb-frustrated phase
87: separation suggesting that stripe formation is charge driven. Indeed,
88: using the Ising model, it has been shown that the competition between
89: long range Coulomb interactions and short range attraction
90: between holes leads to formation of stripes \cite{Low94}.
91: In this case Ginzburg-Landau considerations lead to an onset of charge
92: order prior to spin order as the temperature is lowered.
93: However, the above analysis does not take into account spin fluctuations
94: which might be crucial for the nature of the phase transition by
95: precluding the spins from ordering at the charge-order temperature
96: \cite{Dui98}. Moreover, the conjecture that long range Coulomb forces
97: are required to stabilize stripe phases has been challenged by the studies
98: of the $t$-$J$ model, in which the DW structures were obtained without
99: such interactions \cite{Whi98}.
100:
101: In order to investigate the influence of strong electron correlations
102: due to large on-site Coulomb repulsion $U$ at Cu ions, several methods
103: have been employed to study the stripe phases which go beyond the HF
104: approximation, such as:
105: Density Matrix Renormalization Group (DMRG) \cite{Whi98,Whi99},
106: Slave-Boson Approximation (SBA) \cite{Sei98,Sei98V,Sei04},
107: variational local ansatz approximation \cite{Gor99},
108: Exact Diagonalization (ED) of finite clusters \cite{Toh99},
109: analytical approach based on variational trial wave
110: function within the string picture \cite{Wro00},
111: Dynamical Mean Field Theory (DMFT) \cite{Fle00,Fle01},
112: Cluster Perturbation Theory (CPT) \cite{Zac00},
113: and Quantum Monte Carlo (QMC) \cite{Bec01,Rie01}.
114: In spite of this huge effort, it remains unclear whether DWs are centered
115: on rows of metal atoms, hereafter named site-centered (SC) stripes, or if
116: they are centered on rows of oxygen atoms bridging the two neighboring
117: metal sites, the so-called bond-centered (BC) stripes, and even
118: calculations performed on larger clusters did not yield a definite answer
119: \cite{Rac05}. Therefore, the purpose of this paper is to study the
120: stability of both structures based on the mean-field
121: method applied either to the extended single-band Hubbard model or the
122: single-band Peierls-Hubbard Hamiltonian which includes the so-called
123: static phonons \cite{Zaa96}. For stable phases we
124: determine charge and magnetization densities, double occupancy,
125: kinetic and magnetic energies, and investigate the role of a finite
126: electron-lattice coupling.
127:
128:
129: \section{\label{sec:2} Experimental signatures of stripes }
130:
131: Experimentally, stripe phases are most clearly detected in insulating
132: compounds with a static stripe order, but there is growing evidence
133: of fluctuating stripe correlations in metallic and superconducting
134: materials. The most direct evidence for stripe phases in doped
135: antiferromagnets has come from neutron scattering studies in which charge
136: and spin modulations are identified by the appearance of some IC Bragg
137: peaks, in addition to those which correspond to the crystal structure.
138: However, sometimes sufficiently large crystals are not available for such
139: experiments, and one has to resort to other methods capable of probing
140: local order. These methods include nuclear magnetic resonance (NMR),
141: nuclear quadruple resonance (NQR), muon spin rotation ($\mu$SR),
142: scanning tunneling microscopy (STM),
143: and transmission electron microscopy (TEM).
144: Furthermore, angle-resolved photoemission spectroscopy (ARPES),
145: angle-integrated photoemission spectroscopy (AIPES),
146: as well as x-ray photoemission (XPS) and
147: ultraviolet photoemission (UPS) spectroscopies all provide
148: essential information about conspicuous changes in the
149: electronic structure when stripe structure sets in.
150: Finally, a distinct imprint of the 1D spin-charge modulation
151: on transport properties should be detectable as the in-plane anisotropy
152: of the resistivity and the Hall coefficient $R_H$.
153:
154: The abundance of the current evidence on various types of stripe order
155: as well as the recent ARPES results on the spectral weight of the cuprate
156: superconductors is contained in the review articles by Kivelson
157: {\it et al.} \cite{Kiv03}, and by Damascelli {\it et al.} \cite{Dam03}.
158: Historically, the first compelling evidence for both magnetic and charge
159: order in the cuprates was accomplished in a neodymium codoped
160: compound La$_{2-x-y}$Nd$_{y}$Sr$_x$CuO$_4$ (Nd-LSCO).
161: For $y=0.4$ and $x=0.12$, Tranquada {\it et~al.} \cite{Tra95Nd,Tra96Nd}
162: found that the magnetic scattering is not characterized by the
163: two-dimensional (2D) AF wave vector ($1/2,1/2$), but by IC peaks
164: at the wave vectors $(1/2\pm\epsilon,1/2)$ with $\epsilon = 0.118$.
165: Moreover, inspired by the pioneering works demonstrating that the
166: staggered magnetization undergoes a phase shift of $\pi$ at the charge
167: DWs \cite{Zaa89,Poi89,Sch89,Kat90,Inu91}, the authors found
168: additional charge order peaks $(\pm 2\epsilon,0)$,
169: precisely at the expected position $2\epsilon=0.236$. Interestingly,
170: this doping corresponds to a local minimum in the doping dependence of
171: the superconducting temperature $T_c$ in Nd-LSCO \cite{Ich00}, suggesting
172: that the static stripes are responsible for this anomalous depression of
173: superconductivity. However, it may well be that the apparent correlation
174: is entirely accidental and therefore the role of stripes in
175: superconductivity remains an open question \cite{Car02}.
176:
177: Unfortunately, in early studies Tranquada {\it et~al.} \cite{Tra97Nd}
178: detected only magnetic IC peaks at higher doping levels $x=0.15$ and
179: $x=0.2$. Nevertheless, systematic NQR studies of Nd-LSCO revealed the
180: presence of robust charge stripe order throughout the entire
181: superconducting regime of doping $0.07\le x\le 0.25$ \cite{Sin99}.
182: Also in a more recent study, both charge and spin superlattice peaks
183: at $x=0.15$ were found recently in the neutron diffraction experiments
184: by Wakimoto {\it et~al.} \cite{Wak03}.
185:
186: In fact, the reason why static stripes could be detected in this compound
187: is a structural transition from the low temperature orthorhombic (LTO) to
188: the low temperature tetragonal (LTT) phase, induced by the substitution
189: for La ions by isovalent Nd ions. This, in turn, provides a pinning
190: potential for dynamic stripes and stabilizes the charge order.
191: Evidence of a similar pinning potential has also been found both in the
192: $\mu$SR and NQR studies of La$_{2-x-y}$Eu$_{y}$Sr$_x$CuO$_4$ (Eu-LSCO)
193: with $y\simeq 0.2$ \cite{Kla00,Tei00}.
194: Moreover, the connection between the LTT phase and the
195: appearance of charge and spin stripe order has been clearly demonstrated
196: both in the neutron scattering and $x$-ray diffraction studies on
197: La$_{2-x-y}$Ba$_{y}$Sr$_x$CuO$_4$ (Ba-LSCO) with $y=1/8$ \cite{Fuj02Ba,Kim04}.
198: Finally, static IC charge $(2\pm 2\epsilon,0)$ and magnetic
199: $(1/2\pm\epsilon,1/2)$ peaks have been detected within the LTT phase
200: of La$_{2-x}$Ba$_{x}$CuO$_4$ (LBCO) with $x=1/8$ \cite{Fuj04}.
201: The position of the peaks and the established incommensurability
202: $\epsilon = 0.118$ are exactly the same as those obtained by Tranquada
203: {\it et~al.} \cite{Tra96Nd} for Nd-LSCO. Notably, the peaks that correspond
204: to charge order appear always at somewhat higher temperature than the
205: magnetic ones, indicating that the stripe order is driven by the charge
206: instability.
207:
208: Let us now discuss the experimental evidence of slowly fluctuating stripes
209: in La$_{2-x}$Sr$_{x}$CuO$_4$. The main difference between the Ba and Sr
210: codoped system is the fact that the latter undergoes a structural phase
211: transition from the high-temperature tetragonal (HTT) phase to the LTO
212: phase. As a consequence, in the superconducting regime $x\ge0.06$,
213: the LSCO system exhibits purely \emph{dynamic} magnetic correlations
214: which give rise to IC peaks at the wave vector $(1/2\pm\epsilon,1/2)$
215: specified in tetragonal lattice units $2\pi/a_{tetra}$. In their seminal
216: inelastic neutron scattering studies, Yamada {\it et~al.} \cite{Yam98}
217: established a remarkably simple relation $\epsilon\simeq x$ for
218: $0.06\le x\le 0.12$, followed by a lock-in effect at $\epsilon\simeq 1/8$
219: for larger $x$.
220:
221:
222: %
223: \begin{figure}[t!]
224: \begin{center}
225: \includegraphics[width=0.47\textwidth ]{fig1.eps}
226: \end{center}
227: \caption
228: {Summary of experimental data illustrating the doping dependence of
229: incommensurability $\epsilon$ in the cuprates.
230: Results have been obtained by different groups:
231: Nd-LSCO (Refs. \cite{Tra95Nd,Tra96Nd,Tra97Nd,Ich00});
232: LSCO (Refs.~\cite{Yam98,Wak99,Wak00,Fuj02,Mat00,Mat00a,Mat02});
233: LCO (Ref.~\cite{Lee99});
234: Zn-LSCO (Refs.~\cite{Hir98,Kim99});
235: YBCO (Refs.~\cite{Dai01,Moo02}). In
236: LSCO, $\epsilon$ has been defined as a distance from the IC peak
237: position to the AF wave vector ($1/2,1/2$) either in the orthorhombic
238: ($x<0.06$) or tetragonal ($x>0.06$) notation (see insets), whereas at
239: $x=0.06$, both definitions are used due to the coexistence of diagonal
240: and parallel to the Cu-O bonds spin modulations.
241: }
242: \label{fig:epsilon}
243: \end{figure}
244: %
245:
246: In contrast, in the insulating spin-glass regime of LSCO $x\le 0.06$,
247: quasielastic neutron scattering experiments with the main weight at zero
248: frequency demonstrate that IC magnetic peaks are located at the wave
249: vectors $(1/2\pm\epsilon/\sqrt{2},1/2\pm\epsilon/\sqrt{2})$
250: \cite{Wak99,Wak00,Fuj02}. This phenomenon has often been interpreted
251: as the existence of static diagonal stripes, even though no signatures of
252: a charge modulation were observed. Another possible explanation is the
253: formation of a short ranged spiral order as its chirality also breaks
254: the translational symmetry of the square lattice by a clockwise or
255: anticlockwise twist \cite{Has04}. Remarkably, even though the spin
256: modulation changes from a diagonal to vertical/horizontal one, i.e.,
257: along Cu-O bonds, at $x$ around 0.06, $\epsilon$ follows the doping $x$
258: reasonably well over the entire range $0.03\le x\le 0.12$, as shown in
259: Fig.~\ref{fig:epsilon}. In fact, just for $x=0.06$, both diagonal
260: ($\epsilon=0.053$) and vertical/horizontal ($\epsilon=0.049$) IC spin
261: modulations have been found to coexist \cite{Fuj02}. In a stripe model
262: this corresponds to a constant density of 0.5 (0.7) holes per Cu atom in
263: the DWs in the vertical/horizontal (diagonal) stripe phases, respectively,
264: because of the difference in Cu spacings in the two geometries, i.e.,
265: $a_{ortho}=\sqrt{2}a_{tetra}$. In contrast, in the narrow region
266: $0.02\le x\le 0.024$, IC magnetic peaks are located at the wave vector
267: $(1/2\pm\epsilon/2,1/2\pm\epsilon/2)$ with $\epsilon\simeq x$
268: corresponding to a constant charge of one hole/Cu ion along a diagonal DW
269: \cite{Mat00,Mat00a,Mat02}. However, below $x=0.02$, this does not hold
270: anymore and the incommensurability gets locked with the value
271: $\epsilon\simeq0.014$.
272:
273: Unfortunately, any concomitant charge ordering has not yet been detected
274: in LSCO. Nevertheless, by comparing the data based on the wipeout effect
275: of $^{63}$Cu NQR charge order parameter in LSCO with the ones obtained
276: from charge stripe compounds as (Nd,Eu,Ba)-LSCO, Hunt {\it et~al.}
277: \cite{Hun99} concluded that a similar stripe instability exists in LSCO
278: over the whole underdoped superconducting region $1/16\le x\le 1/8$.
279: It is also worth mentioning that a very compiling evidence for its
280: existence has been established in the measurements of the in-plane
281: resistivity and the dynamical infrared conductivity anisotropy
282: \cite{And02,Dum03}.
283:
284: Experimental detection of IC magnetic peaks in the LTO phase of LSCO
285: suggests that the LTT structure is not essential for the appearance of
286: stripes. This conjecture has been confirmed in experiments on the
287: oxygen doped La$_{2}$CuO$_{4+\delta}$ (LCO) with the orthorhombic
288: crystal structure \cite{Lee99}.
289: It is also supported by the evidence for \emph{static}
290: IC magnetic peaks in another orthorhombic compound
291: La$_{2-x}$Sr$_{x}$Cu$_{1-y}$Zn$_y$O$_4$ (Zn-LSCO)
292: with $y$ up to $0.03$, even though attempts to observe the charge order
293: peaks were unsuccessful \cite{Hir98,Kim99}.
294: In fact, Zn substitution pins the stripe fluctuations similarly to the
295: rare-earth elements. However, in contrast to the latter, it does not
296: induce a structural transition to the LTT phase, but provides randomly
297: distributed pinning centers that promote meandering of stripes and
298: correspondingly broadens IC peaks.
299:
300: An important question is whether charge stripes appear solely in
301: monolayered lanthanum compounds or if they are a generic feature of all
302: the cuprates. The latter conjecture seems to be supported by inelastic
303: neutron scattering experiments on bilayered
304: YBa$_{2}$Cu$_{3}$O$_{6+\delta}$ (YBCO) compounds that have identified the
305: presence of IC spin fluctuations throughout its entire superconducting
306: regime \cite{Dai01}. In fact, as the doped charge is nontrivially
307: distributed between the CuO$_2$ planes and CuO chains, it is very
308: difficult to determine the precise doping level $x$ in the CuO$_2$ sheet
309: of YBCO. Nevertheless, systematic studies by Dai {\it et al.} \cite{Dai01}
310: have shown that the incommensurability in YBCO increases initially with
311: doping but it saturates faster than in LSCO, i.e., already at
312: $x\simeq 0.1$ with the value $\epsilon\simeq 0.1$. Unfortunately, there
313: is no any compelling explanation that would account for such a different
314: behavior of $\epsilon$ in both systems.
315: Eventually, charge order peaks have been observed in YBCO$_{6.35}$
316: but in spite of several attempts, no static charge order could be
317: detected in YBCO$_{6.5}$ and YBCO$_{6.6}$ so far \cite{Moo02}.
318:
319: Furthermore, although some neutron scattering experiments have been
320: performed on Bi$_2$Sr$_2$CaCu$_2$O$_{8+\delta}$ (BSCCO) sample, the
321: sample has only produced weak evidence of the IC structure \cite{Moo97}.
322: In contrast, Fourier transform of the recent STM data has revealed some
323: IC peaks corresponding to a four-period modulation of the local density
324: of states along the Cu-O bond direction, which may imply the existence of
325: stripes \cite{How03}. Nevertheless, definite answer pertinent to the
326: appearance of stripes in all the cuprates remains still unsettled and
327: further experiments are required to reach an unambiguous conclusion, even
328: though the summary of the experimental data illustrating the doping
329: dependence of the incommensurability $\epsilon$ in cuprates, depicted in
330: Fig.~\ref{fig:epsilon}, includes an array of compounds.
331:
332: Tendency towards phase separation is also a starting point to understand
333: the doping evolution of the electronic structure in LSCO and Nd-LSCO.
334: For example, ARPES spectra measured at the $X=(\pi,0)$ point in LSCO
335: show that even though the data are solely characterized by a single high
336: binding energy feature in the insulating regime, upon increasing
337: doping one observes a systematic transfer of spectral weight from the
338: high- to the low binding energy part \cite{Ino00}. Consequently, a
339: well-defined quasiparticle (QP) peak develops near the optimal doping.
340: In contrast, the intensity near the $S=(\pi/2,\pi/2)$ point remains
341: suppressed for the entire underdoped regime so that a QP peak is
342: observed only for $x\ge 0.15$.
343:
344: Another peculiar feature of the ARPES band dispersion is extensively
345: discussed in the literature saddle point at the $X$ point, the so-called
346: flat band \cite{Ino02}. As hole doping increases, the flat band moves
347: monotonically upwards and crosses the Fermi level $E_F$ at $x\simeq 0.2$.
348: This is reflected in the enhancement of the DOS at the chemical potential
349: $N(\mu)$ observed by AIPES \cite{Ino98}.
350:
351: The experimental distribution of the photoemission spectral weight near
352: the $X$ and $S$ points in doped LSCO has been nicely reproduced using the
353: DMFT approach for vertical SC stripes obtained within the Hubbard model
354: \cite{Fle00}. As a consequence of the stripe order, the obtained spectra
355: along the $\Gamma-X-M$ path were not equivalent to those along the
356: $\Gamma-Y-M$ one, with $\Gamma=(0,0)$ and $Y=(0,\pi)$. Moreover, as in
357: the experiment, the spectral weight along the $\Gamma-X$ direction was
358: suppressed close to the $\Gamma$ point and simultaneously enhanced at the
359: $X$ point. Furthermore, in the framework of stripes, the flat QP band
360: near the $X$ point with a large intensity at the maximum below the
361: chemical potential $\mu$ follows from a superposition of the
362: dispersionless 1D metallic band along the $x$ direction, formed by holes
363: propagating along the vertical domain walls, and an insulating band that
364: stems from the AF domains. In contrast, an AF band at the $Y$ point is
365: characterized by a high binding energy well below $\mu$ and consequently
366: the spectral weight at $\omega=\mu$ almost vanishes. Finally, a distinct
367: gap for charge excitations should open at $\mu$ near the $S$ point. This
368: gap follows indeed from the stripe structure --- while the system may be
369: metallic along the stripes, i.e., in the antinodal directions $\Gamma-X$
370: or $\Gamma-Y$, the low-energy excitations should be noticeably
371: suppressed along the nodal direction $\Gamma-S$ crossing all the stripes.
372: This conjecture is also supported either by the ED studies \cite{Toh99}
373: or by the analytical approach based on variational trial wave function
374: within the string picture \cite{Wro00}, both applied to the
375: $t$-$t'$-$t''$-$J$ model, or by the CPT for the $t$-$J$ model \cite{Zac00}.
376:
377: In fact, the low-energy spectral weight of Nd-LSCO at $x=0.12$, a model
378: compound for which the evidence of spin and charge stripe order is
379: the strongest, is also mostly concentrated in flat regions along the
380: $\Gamma-X$ and $\Gamma-Y$ directions, while there is only little spectral
381: weight along the $\Gamma-S$ direction \cite{Zho99}. On the other hand,
382: ARPES spectra of both LSCO and Nd-LSCO at $x=0.15$ have revealed not only
383: the presence of flat bands around the $X$ and $Y$ points, but also the
384: existence of appreciable spectral weight at $E_F$ in the nodal region
385: \cite{Zho01}. While the observation of flat segments might be directly
386: ascribed to 1D domain walls \cite{Sal96}, detection of nodal spectral
387: weight poses a formidable task to develop a theory that would describe
388: the electronic structure resembling the FS of a fully 2D system because,
389: as it was already stressed out, the nodal spectral weight is expected to
390: be suppressed in a static SC stripe picture
391: \cite{Fle00,Toh99,Wro00,Zac00}. Indeed, the experimentally established
392: FS looks rather like the one arising from disorder or from dynamically
393: fluctuating stripes \cite{Sal96}.
394:
395: Alternatively, guided by the CPT results showing that while the SC
396: stripes yield little spectral weight near the nodal region, the BC ones
397: reproduce quite well the nodal segments \cite{Zac00}, Zhou {\it et~al.}
398: \cite{Zho01} have conjectured that the experimental FS may result from
399: the coexistence of the SC and BC stripes. Within this framework, upon
400: increasing doping the BC stripes are formed at the expense of the SC
401: ones. This scenario is particularly interesting because it has been
402: shown that the BC stripe, in contrast to its SC counterpart, enhances
403: superconducting pairing correlations \cite{Eme97}.
404: The relevance of a bond order at the doping level $x=0.15$ is supported
405: by recent studies of the ARPES spectra in a system with the BC stripes
406: \cite{Wro05}. These studies have yielded pronounced spectral weight both
407: in the nodal and antinodal directions, reproducing quite well the
408: experimental results in Nd-LSCO and LSCO \cite{Zho01}. Furthermore, the
409: stripe scenario would also explain the origin of the already discussed
410: two components seen in the ARPES spectra at the $X$ point near $x=0.05$
411: \cite{Ino00}. Indeed, the response from the AF insulating regions would
412: be pushed to the high binding energies due to the Mott gap, whereas the
413: charge stripes would be responsible for the other component near $E_F$.
414:
415: Existence of DWs should also give rise to the appearance of new states
416: inside the charge-transfer gap that would suppress the shift of the
417: chemical potential $\mu$ in the underdoped regime $x<1/8$ where
418: $\epsilon$ increases linearly. Such pinning of $\mu$ in LSCO was indeed
419: deduced from XPS experiments \cite{Ino97}. In contrast, in the overdoped
420: region with a lock-in effect of $\epsilon$, the number of stripes per
421: unit cell saturates, doped holes penetrate into the AF domains, and
422: consequently $\mu$ would move fast with doping in agreement with the
423: experimental data. The picture of broadened stripes and holes spreading
424: out all over the AF domains above $x=1/8$ is also indicated by the doping
425: dependence of the resistivity and the Hall coefficient $R_H$ in Nd-LSCO.
426: Namely, a rapid decrease in the magnitude of $R_H$ for doping level
427: $x\le 1/8$ at low temperature provides evidence for the 1D charge
428: transport, whereas for $x>1/8$, relatively large $R_H$ suggests a
429: crossover from the 1D to 2D charge transport \cite{Nod99}. Altogether,
430: it appears that the metallic stripe picture does capture the essence of
431: the low-lying physics for Nd-LSCO and LSCO systems.
432:
433: Conversely, it is important to note that so far no evidence of IC peaks
434: has been detected in any electron-doped cuprates superconductors.
435: Instead, the neutron scattering experiments have established only
436: \emph{commensurate} spin fluctuations as in Nd$_{2-x}$Ce$_x$CuO$_{4}$
437: (NCCO), both in the superconducting and in normal state \cite{Yam03}.
438: Moreover, observation of such peaks is consistent with the XPS
439: measurements in NCCO showing that the chemical potential increases
440: monotonously with electron doping \cite{Har01}.
441:
442: \section{\label{sec:3} Numerical results}
443:
444: In this Section we attempt a systematic investigation of the properties
445: and relative stability of filled vertical and diagonal stripes. We shall
446: see that in spite of the difficulty to stabilize the ground state with
447: half-filled stripes (one hole per every two atoms in a DW), the
448: mean-field framework is useful as providing a generic microscopic
449: description of filled inhomogeneous reference structures with the
450: filling of one doped hole per stripe unit cell. Their special stability
451: rests on a gap that opens in the symmetry broken state between the
452: highest occupied state of the lower Hubbard band and the bottom of the
453: so-called mid-gap bands, i.e., some additional unoccupied bands lying
454: within the Mott-Hubbard gap that are formed due to holes propagating
455: along DWs \cite{Zaa96}.
456:
457: Here, we extend early HF studies of the filled DWs
458: \cite{Poi89,Sch89,Kat90,Inu91} and determine a phase diagram of the
459: Hubbard model with an anisotropic nearest-neighbor hopping $t$ by
460: varying the on-site Coulomb repulsion $U$ and investigating locally
461: stable structures for representative hole doping levels $x=1/8$ and
462: $x=1/6$. We also report the changes in stability of the stripe structures
463: in the extended Hubbard model due to the next-neighbor hopping $t'$ and
464: to the nearest-neighbor Coulomb interaction $V$. Finally, in order to
465: gain a comprehensive understanding of the competition between different
466: types of stripes in a realistic model, we include lattice degrees of
467: freedom induced by a static Peierls electron-lattice coupling.
468:
469: \subsection{\label{sec:3a} Extended single-band Hubbard model}
470:
471: The starting point for the analysis of stripe structures is the extended
472: single-band Hubbard model, which is widely accepted as the generic model
473: for a microscopic description of the cuprate superconductors \cite{Fei96},
474: %
475: \begin{equation}
476: H=-\sum_{ij\sigma}t^{}_{ij}c^{\dag}_{i\sigma}c^{}_{j\sigma} +
477: U\sum_{i}n^{}_{i\uparrow}n^{}_{i\downarrow} +
478: V\sum_{\langle ij\rangle}n^{}_in^{}_j,
479: \label{eq:Hubb}
480: \end{equation}
481: %
482: where the operator $c^{\dag}_{i\sigma}$ $(c^{}_{j\sigma})$ creates
483: (annihilates) an electron with spin $\sigma$ on lattice site $i$ ($j$),
484: and $n_i=c^{\dag}_{i \uparrow}c^{}_{i \uparrow}
485: + c^{\dag}_{i\downarrow}c^{}_{i\downarrow}$
486: gives the electron density. The hopping $t_{ij}$ is $t$ on the bonds
487: connecting nearest neighbors sites $\langle i,j\rangle$ and $t'$ for
488: second-neighbor sites, while the on-site and nearest-neighbor Coulomb
489: interactions are, respectively, $U$ and $V$.
490:
491: The model can be solved self-consistently in real space within the HF,
492: where the interactions are decoupled into products of
493: one-particle terms becoming effective mean fields that act on each
494: electron with the same strength. This approximation basically involves
495: solving an eigenvalue problem. The obtained wavefunctions form a new
496: potential and hence the Hamiltonian for a new eigenvalue problem. Typically,
497: the new potential is chosen as some linear combination of the current
498: and preceding potential. The iterations are continued until the input and
499: output charge density and energy do not change within some prescribed
500: accuracy. The most significant drawback of this method is that it neglects
501: correlations. Electron correlation changes the system properties and
502: manifests itself in the decrease of the ground state energy. The difference
503: between the energy of the exact ground state and the energy obtained within
504: the HF is thus called the correlation energy. It arises from the fact that
505: an electron's movement is correlated with the electrons around it,
506: and accounting for this effect lowers further the energy, beyond the
507: independent electron approximation.
508:
509: We do not consider noncollinear spin configurations, and use the most
510: straightforward version of the HF with a product of two
511: separate Slater determinants for up and down spins, whence,
512: %
513: \begin{equation}
514: n_{i\uparrow}n_{i\downarrow}\simeq
515: n_{i\uparrow}\langle n_{i\downarrow}\rangle
516: +\langle n_{i\uparrow}\rangle n_{i\downarrow}
517: -\langle n_{i\uparrow}\rangle\langle n_{i\downarrow}\rangle.
518: \label{eq:MF}
519: \end{equation}
520: %
521: A similar decoupling is performed for the nearest-neighbor Coulomb
522: interaction. Calculations were performed on $12\times 12$ ($16\times 16$)
523: clusters for $x=1/6$ ($x=1/8$) with periodic boundary conditions, and we
524: obtain stable stripe structures with AF domains of width five atoms for
525: $x=1/6$ and seven atoms for $x=1/8$. Typical solutions at $x=1/8$ are
526: shown in Fig.~\ref{fig:16SC} with the local hole density,
527: %
528: \begin{equation}
529: \langle n_{{\rm h}i}\rangle =
530: 1-\langle n_{i\uparrow} + n_{i\downarrow}\rangle,
531: \label{eq:nhi}
532: \end{equation}
533: %
534: scaled by the diameter of the black circles and the length of the
535: arrows being proportional to the amplitude of local magnetization density,
536: \begin{equation}
537: \langle S_i^z\rangle =
538: \tfrac{1}{2}|\langle n_{i\uparrow} - n_{i\downarrow}\rangle|.
539: \label{eq:Szi}
540: \end{equation}
541: %
542:
543: %
544: \begin{figure}[!t]
545: \begin{center}
546: \unitlength=0.01\textwidth
547: \begin{picture}(100,46)
548: \put(3,46){\includegraphics[scale=0.25,angle=270]{fig2a.eps}}
549: \put(25,46){\includegraphics[scale=0.25,angle=270]{fig2b.eps}}
550: \put(3,24){\includegraphics[scale=0.25,angle=270]{fig2c.eps}}
551: \put(25,24){\includegraphics[scale=0.25,angle=270]{fig2d.eps}}
552: \put(2.5,0){\vector(1,0){10}}
553: \put(2.5,0){\vector(0,1){10}}
554: \put(5,1){$l_x$}
555: \put(0.5,4){$l_y$}
556: \end{picture}
557: \end{center}
558: \caption{ Vertical site-centered (VSC) and
559: diagonal site-centered (DSC)
560: stripe phases as found for $U/t=5$ at hole doping $x=1/8$.
561: The length of arrows is proportional to the magnetization
562: $\langle S_i^z\rangle$ and the hole density $\langle n_{{\rm h}i}\rangle$
563: is scaled by the diameter of black circles.}
564: \label{fig:16SC}
565: \end{figure}
566: %
567:
568: These structures possess nonmagnetic DWs with enhanced hole
569: density which separate AF domains having hole density almost unchanged
570: with respect to the undoped case. Note that the AF sites on each side of
571: the DWs have a phase shift of $\pi$.
572:
573: In order to appreciate better the microscopic reasons
574: of such arrangement let us consider a small cluster consisting of
575: three atoms filled by two electrons and one hole (with respect to
576: half-filling with the electron density $n=1$ per site). For simplicity
577: we assume that the electrons are confined to the considered cluster
578: owing to large Coulomb interaction $U\gg t$, and we do not
579: take into account any interactions with the AF background. There are two
580: possible candidates for the ground state. The first one corresponds to a
581: hole added to three atoms of a single AF domain in which, if we suppose
582: that a $\downarrow$-spin electron is replaced by a hole, the two remaining
583: $\uparrow$-spin electrons can be found in one of three allowed
584: configurations: $\{\uparrow, 0, \uparrow\}$, $\{\uparrow, \uparrow, 0\}$,
585: and $\{0,\uparrow,\uparrow\}$ (the other configurations are excluded by
586: the Pauli principle). Hence, this polaronic state gives the total energy,
587: %
588: \begin{equation}
589: E_P=-\sqrt{2}t,
590: \label{eq:pol}
591: \end{equation}
592: %
593: and the Coulomb interaction $U$ does not contribute.
594:
595: %
596: \begin{figure}[!t]
597: \begin{center}
598: \unitlength=0.01\textwidth
599: \begin{picture}(100,46)
600: \put(3,46) {\includegraphics[scale=0.25,angle=270]{fig3a.eps}}
601: \put(25,46) {\includegraphics[scale=0.25,angle=270]{fig3b.eps}}
602: \put(3,24){\includegraphics[scale=0.25,angle=270]{fig3c.eps}}
603: \put(25,24){\includegraphics[scale=0.25,angle=270]{fig3d.eps}}
604: \put(2.5,0){\vector(1,0){10}}
605: \put(2.5,0){\vector(0,1){10}}
606: \put(5,1){$l_x$}
607: \put(0.5,4){$l_y$}
608: \end{picture}
609: \end{center}
610: \caption{ Vertical bond-centered (VBC)
611: and diagonal bond-centered (DBC)
612: stripe phases as found for $U/t=5$ at hole doping $x=1/8$.
613: The meaning of the arrows and black circles as in Fig.~\ref{fig:16SC}.}
614: \label{fig:16BC}
615: \end{figure}
616: %
617:
618: A different situation is obtained when a hole occupies instead a DW
619: separating two AF domains. Delocalization leads then to similar three
620: configurations to those obtained above:
621: $\{\uparrow, 0 ,\downarrow\}$, $\{\uparrow, \downarrow, 0\}$, and
622: $\{0, \uparrow, \downarrow\}$, but in addition, three configurations with
623: one doubly occupied site $\{\uparrow\downarrow, 0, 0\}$,
624: $\{0, \uparrow\downarrow, 0\}$, and $\{0, 0, \uparrow\downarrow\}$, can be
625: reached as excited states which cost Coulomb energy $U$. Moreover, three
626: other configurations with interchanged $\uparrow$- and $\downarrow$-
627: spins are then also accessible via the decay of double occupancies:
628: $\{\downarrow, 0, \uparrow\}$, $\{\downarrow, \uparrow, 0\}$, and
629: $\{0, \downarrow, \uparrow\}$. In the regime of large $U$, the total
630: energy in the ground state can be found in a perturbative way, and as
631: a result one obtains,
632: %
633: \begin{equation}
634: E_S=-\sqrt{2}t - \frac{4t^2}{U}.
635: \label{eq:sol}
636: \end{equation}
637: %
638: Therefore, the Hilbert space for the latter solitonic solution is larger
639: and one finds that this solution is always more stable than the polaronic
640: one \cite{Zaa96}. The argument applies also to 2D systems, where the DWs
641: are more stable than the lines of polarons in an AF background.
642:
643: We compare the stability of such nonmagnetic SC domain walls with
644: the BC stripe phases in which DWs are formed by pairs of magnetic atoms,
645: as obtained by White and Scalapino \cite{Whi98}
646: (\textit{cf}. Fig.~\ref{fig:16BC}). In the three-band model, SC (BC)
647: stripes correspond to DWs centered at metal (oxygen) sites,
648: respectively \cite{Miz97,Yu98,Sad00,Lor02}.
649:
650: \subsection{\label{sec:3b} Effect of hopping anisotropy}
651:
652: We begin by setting $t'=0$ and $V=0$ with the goal of elucidating the
653: effects of hopping anisotropy on the stripes. This is motivated by the
654: fact that the first detection of static stripes in both charge and spin
655: sectors was accomplished in Nd-LSCO \cite{Tra95Nd} indicating that
656: rare-earth elements doping is in some way helpful for pinning the stripe
657: structure. Indeed, it produces a structural transition in the system
658: from the LTO to LTT phase \cite{Buc94}. Both phases involve a distortion
659: of the CuO$_2$ plane by rotation of the CuO$_6$ octahedra. In the
660: LTO phase the tilt axis runs diagonally within the copper plane,
661: such that all the oxygen atoms are displaced out of the plane. Conversely,
662: in the LTT phase this rotation takes place around an axis oriented
663: along the planar Cu$-$O bonds. Therefore, oxygen atoms on the tilt
664: axis remain in the plane, while the ones in the perpendicular direction
665: are displaced out of the plane. This provides a microscopic origin
666: for in-plane anisotropies --- the Cu$-$Cu hopping amplitude $t$ depends
667: on the Cu$-$O bond and it is isotropic in the LTO phase and anisotropic
668: in the LTT one. For a physical tilt angle of order 5$^{\circ}$, the
669: relative anisotropy taking $t_y<t_x$,
670: %
671: \begin{equation}
672: \epsilon_t = \frac{|t_x-t_y|}{t_y},
673: \label{eq:et}
674: \end{equation}
675: %
676: is weak and amounts to $\epsilon_t=1.5\%$ \cite{Nor01,Kam01}. The
677: direction with a larger hopping amplitude coincides with the direction
678: of a stronger superexchange coupling $J$.
679:
680: %
681: \begin{figure}[t!]
682: \begin{center}
683: \includegraphics[width=0.47\textwidth ]{fig4.eps}
684: \end{center}
685: \caption
686: { Local hole $n^{}_{\rm h}(l_x)$ (top) and magnetization $S_{\pi}(l_x)$
687: (second row) density; kinetic energy $E_{t}^{x}(l_x)$ (third row) and
688: $E_{t}^{y}(l_x)$ (bottom) projected on the bonds in the $x$-($y$)-directions,
689: respectively, of the VSC (left) and DSC (right) stripe phases shown in
690: Fig.~\ref{fig:16SC} (open circles) as well as of the ones obtained in the
691: anisotropic model with $t_x/t_y=1.22$ (filled circles). For clarity, the
692: latter are shifted by one lattice constant from the origin of the coordinate
693: system.
694: }
695: \label{fig:SC16}
696: \end{figure}
697: %
698: The possible relationship between this anisotropy and the onset of stripe
699: phases has been intensively studied within anisotropic Hubbard
700: ($t_x\ne t_y$) or $t$-$J$ ($t_x\ne t_y$, $J_x\ne J_y$) models by means of
701: various techniques: unrestricted HF approach \cite{Nor01}, DMRG
702: \cite{Kam01}, and QMC method \cite{Bec01}.
703: The in-plane anisotropies might also be represented theoretically by
704: on-site potentials as in the QMC study by Riera \cite{Rie01}.
705: All these studies have shown a pronounced tendency to forming stripe
706: phases, which manifests itself by the reduction of their energy
707: \cite{Nor01,Kam01}, accompanied by the appearance of IC peaks in the spin
708: and charge structure factor \cite{Bec01,Rie01}. It appears that
709: a finite anisotropy of the next-nearest hopping term $t'$ might play
710: a role in stabilizing diagonal incommensurate peaks observed in the
711: spinglass phase of LSCO ($0.02\leqslant x\leqslant 0.06$)
712: \cite{Wak99,Wak00, Mat00, Mat00a,Fuj02,Mat02}. Indeed, although the LTO
713: phase is usually considered as isotropic, which is the case for
714: nearest-neighbor hopping and interaction, a different length of the
715: orthorhombic axes implies the need for an anisotropic $t'$ parameter.
716: Exact diagonalization studies incorporating such anisotropy have shown
717: that it strongly strengthens hole correlations along one direction and
718: suppresses them along the other, resulting in a 1D pattern of holes
719: \cite{Tip03}.
720:
721:
722: %
723: \begin{figure}[t!]
724: \begin{center}
725: \includegraphics[width=0.47\textwidth ]{fig5.eps}
726: \end{center}
727: \caption
728: {
729: The same as in Fig.~\ref{fig:SC16} but for the BC stripe phases.
730: }
731: \label{fig:BC16}
732: \end{figure}
733:
734: It turns out, however, that the variation of the hopping anisotropy
735: $\epsilon_t$ (\ref{eq:et}) has only a little visible effect on the local
736: hole density,
737: %
738: \begin{equation}
739: n_{\rm h}(l_x) = 1-\langle n_{(l_x,0),\uparrow} +
740: n_{(l_x,0),\downarrow}\rangle,
741: \label{eq:nh}
742: \end{equation}
743: %
744: shown in Fig.~\ref{fig:SC16} as a function of the $x$-direction
745: coordinate $l_x$ for a given $y$-direction coordinate $l_y=0$, even at
746: the unrealistically large anisotropy level $\epsilon_t=22\%$,
747: corresponding to $t_x/t=1.1$ and $t_y/t=0.9$. Similarly, the anisotropy
748: does not modify the modulated magnetization density,
749: %
750: \begin{equation}
751: S_{\pi}(l_x) = (-1)^{l_x}\tfrac{1}{2}
752: \langle n_{(l_x,0),\uparrow} - n_{(l_x,0),\downarrow}\rangle,
753: \label{eq:Spi}
754: \end{equation}
755: %
756: with a site dependent factor $(-1)^{l_x}$ compensating modulation
757: of the staggered magnetization density within a single AF domain.
758:
759: In contrast, the strong effect of finite anisotropy $\epsilon_t$
760: (\ref{eq:et}) is clearly demonstrated by variation of the expectation
761: values of the bond hopping terms along the $x$- and $y$-directions,
762: %
763: \begin{align}
764: E_t^x(l_x) &= -t_x\sum_{\sigma}
765: \bigl\langle c_{(l_x,0),\sigma}^{\dag}c_{(l_x+1,0),\sigma}^{} +
766: h.c\bigr\rangle,
767: \label{eq:Etx}\\
768: %
769: E_t^y(l_x) &= -t_y\sum_{\sigma}
770: \bigl\langle c_{(l_x,0),\sigma}^{\dag}c_{(l_x,1),\sigma}^{} +
771: h.c\bigr\rangle.
772: \label{eq:Ety}
773: \end{align}
774: %
775: These features are seen in Fig.~\ref{fig:SC16}. For the VSC stripes one
776: finds a large anisotropy in the values of the kinetic energies
777: (\ref{eq:Etx}) and (\ref{eq:Ety}), which becomes especially pronounced
778: beside the stripes, and is strongly reinforced by the hopping anisotropy.
779: Therefore, taking into account that the hopping between two different
780: charge densities is favored over motion between equal densities, one
781: should expect that transverse charge fluctuations will always tune the
782: direction of DWs along the weaker hopping direction in the anisotropic
783: model. Analogous conclusion based on Fig.~\ref{fig:BC16} might be drawn
784: concerning the orientation of the VBC stripes.
785:
786: Regarding diagonal stripes, although a finite anisotropy in hopping is
787: also reflected in the kinetic energy anisotropy, a system with either the
788: DSC or DBC stripe pattern becomes topologically frustrated and
789: consequently may gain less kinetic energy compared to a system with
790: vertical stripes, taking a full advantage of the hopping anisotropy
791: (\textit{cf}. Tables~\ref{tab:16E} and \ref{tab:16Ean}).
792:
793: %
794: \begin{table}[!t]
795: \begin{center}
796: \begin{tabular}{ccccccc}
797: \hline\hline
798: & & $E_t^x/t$ & $E_t^y/t$ & $E_U/t$ & $E_{\textrm{tot}}/t$ \\
799: \hline
800: &{\bf VB(S)C}& $-$0.6753 & $-$0.6147 & 0.4900 & $-$0.8000 \\
801: &{\bf DBC} & $-$0.6375 & $-$0.6375 & 0.4726 & $-$0.8024 \\
802: &{\bf DSC} & $-$0.6368 & $-$0.6368 & 0.4696 & $-$0.8040 \\
803: \hline\hline
804: \end{tabular}
805: \end{center}
806: \caption {Site-normalized ground-state energy $E_{\rm tot}$, kinetic
807: energy $(E_t^{x}, E_t^{y})$, and potential energy $E_U$
808: in the isotropic Hubbard model with $U/t=5$ and $x=1/8$ as obtained for
809: different stripe phases: vertical site-centered (VSC), diagonal
810: site-centered (DSC), vertical bond-centered (VBC) and diagonal
811: bond-centered (DBC). In the HF, both types of vertical stripes
812: are degenerate.}
813: \label{tab:16E}
814: \end{table}
815: %
816: %
817: \begin{table}[!b]
818: \begin{center}
819: \begin{tabular}{ccccccc}
820: \hline\hline
821: & & $E_t^x/t$ & $E_t^y/t$ & $E_U/t$ & $E_{\textrm{tot}}/t$ \\
822: \hline
823: &{\bf DBC} & $-$0.8143 & $-$0.4807 & 0.4815 & $-$0.8135 \\
824: &{\bf DSC} & $-$0.8098 & $-$0.4836 & 0.4793 & $-$0.8141 \\
825: &{\bf VB(S)C}& $-$0.8304 & $-$0.4776 & 0.4938 & $-$0.8142 \\
826: \hline\hline
827: \end{tabular}
828: \end{center}
829: \caption { The same as in Table~\ref{tab:16E} but with the hopping
830: anisotropy $\epsilon_t=22\%$.}
831: \label{tab:16Ean}
832: \end{table}
833: %
834:
835: The effect of an increasing anisotropy illustrates the phase diagram shown
836: in Fig.~\ref{fig:txtyU} determined by varying $U$ and the ratio $t_x/t_y$
837: of the nearest-neighbor hoppings in the $x$- and $y$-directions,
838: while maintaining constant $t = {\textstyle \frac{1}{2}} (t_x + t_y)$.
839: We observe the generic crossover from vertical to diagonal stripes with
840: increasing Coulomb interaction reported in early HF studies
841: \cite{Poi89,Sch89,Kat90,Inu91}. The transition from the VSC to DSC
842: stripes appears in the isotropic case at $U/t\simeq 4.1$ for $x=1/8$, and
843: at a higher value $U/t\simeq 4.6$ for $x=1/6$ [\textit{cf}.
844: Fig.~\ref{fig:txtyU}(a)]. The corresponding phase boundary between the VBC
845: and DBC stripes is shifted towards stronger Coulomb interaction and occurs
846: at $U/t\simeq 4.4$ (5.0) for $x=1/8$ ($x=1/6$), respectively
847: [\textit{cf}. Fig.~\ref{fig:txtyU}(b)].
848:
849: The results shown in Fig.~\ref{fig:txtyU} have a simple physical
850: interpretation. Stripe phases occur as a compromise between, on the one
851: hand, the AF interactions between magnetic ions and the local Coulomb
852: interactions which favor charge localization, and the kinetic energy of
853: doped holes which favors charge delocalization on the other hand.
854: The kinetic energies in Table~\ref{tab:16E} show further that the vertical
855: stripes are more favorable for charge dynamics. This result, which is not
856: immediately obvious, has however a straightforward origin. Namely,
857: the HF always leads to a large spin polarization since it is
858: the only way to minimize the on-site Coulomb repulsion. Indeed,
859: removal of a $\downarrow$-spin electron at site $i$ leads to relaxation of
860: the $\uparrow$-spin electron energy level at this site. As a consequence,
861: an alternating on-site level shift develops yielding an energetical
862: motivation for the symmetry breaking and forming the AF order.
863:
864: %
865: \begin{figure}[t!]
866: \begin{center}
867: \includegraphics[width=0.47\textwidth ]{fig6.eps}
868: \end{center}
869: \caption
870: {
871: Phase diagrams for stable:
872: (a) site-centered (SC), and
873: (b) bond-centered (BC)
874: stripe structures obtained in the anisotropic Hubbard model
875: on a 16$\times16$ cluster for doping $x=1/8$ (solid lines)
876: and on a 12$\times12$ cluster for $x=1/6$ (dashed lines)
877: Parameters: $t'=0$, $V=0$.
878: }
879: \label{fig:txtyU}
880: \end{figure}
881: %
882:
883: However, the renormalization of the double-occupancy energy involves a
884: strong reduction of the kinetic energy in the $\downarrow$-spin channel
885: between site $i$ and its neighboring sites, as an electron incoming into
886: this site encounters a high energy potential
887: $U\langle n_{i\uparrow}\rangle$. Therefore, in the HF approximation we
888: shall be able to identify dynamically favorable stripe patterns only by
889: comparing appropriate local magnetization densities. For example, charge
890: fluctuations occur more readily in the VSC stripe geometry presumably
891: due to their greater overall width indicating weaker correlation effects
892: (\textit{cf}. Fig.~\ref{fig:SC16}). This explains their stability at
893: small $U$ where the consequent cost in potential energy $E_U$ becomes
894: insignificant. By contrast, the DSC stripes are narrower having larger
895: hole density along nonmagnetic DWs. Moreover, magnetization density of
896: their nearest neighbor sites is markedly enhanced as compared to the
897: corresponding VSC stripe magnetization, as shown in Fig.~\ref{fig:SC16}
898: and in Table~\ref{tab:16SC}. The former also illustrates that the bonds
899: connecting DWs with their nearest neighboring sites perpendicularly to
900: the walls, have the main contribution to the kinetic energy gain, in fact
901: suppressed here by larger spin polarization. Taken together, the above
902: features are reflected in a more localized character of the DSC stripes,
903: with a lower net double occupancy and hence a more favorable on-site
904: energy $E_U$ (\textit{cf}. Table~\ref{tab:16E}). This clarifies the
905: mechanism of the transition from the VSC to DSC stripes with increasing $U$.
906:
907: %
908: \begin{table}[t!]
909: \begin{center}
910: \begin{tabular}{ccccccccc}
911: \hline\hline
912: & & $i$ & & 1 & 2 & 3 & 4 & 5 \\
913: \hline
914: & & $\langle n^{}_{{\rm h}i}\rangle$
915: & & 0.364 & 0.234 & 0.067 & 0.014 & 0.006 \\ &{\bf VSC} &
916: & & (0.378) &(0.234) & (0.060) &(0.013) &(0.006) \\
917: & & $\langle S_i^z\rangle$
918: & & 0.000 & 0.222 & 0.348 & 0.381 & 0.384 \\ & &
919: & & (0.000) &(0.234) & (0.357) &(0.382) &(0.384) \\
920: \hline
921: & & $\langle n^{}_{{\rm h}i}\rangle$
922: & & 0.388 & 0.193 & 0.070 & 0.032 & 0.020 \\ &{\bf DSC} &
923: & & (0.405) &(0.195) &(0.066) & (0.028) &(0.017) \\
924: & & $\langle S_i^z\rangle$
925: & & 0.000 & 0.262 & 0.352 & 0.373 & 0.380 \\ & &
926: & & (0.000) &(0.272) &(0.360) &(0.377) &(0.382) \\
927: \hline\hline
928: \end{tabular}
929: \end{center}
930: \caption {Local hole $\langle n^{}_{{\rm h}i}\rangle$ and magnetization
931: $\langle S_i^z\rangle$ density of the site-centered stripes shown in
932: Fig.~\ref{fig:16SC}, all labeled by decreasing hole density in the
933: $x$-direction. In parenthesis the values for the extended hopping model
934: with $t'/t=-0.15$ are given. }
935: \label{tab:16SC}
936: \end{table}
937: %
938:
939: Turning now to the analogous crossover between the BC stripes, we shall
940: again compare local hole and magnetization densities on and around
941: their DWs. In contrast to the SC case, a VBC stripe phase possesses
942: larger hole density along DWs, as illustrated in Fig.~\ref{fig:16BC} and
943: Table~\ref{tab:16BC}, suggesting that it is more localized than the DBC
944: one. Nevertheless, a better renormalization of the double occupancy
945: energy $E_U$ by the latter (\textit{cf}. Table~\ref{tab:16E}) follows
946: from a stronger spin polarization not only of the DW atoms but also
947: their nearest neighbors (\textit{cf}. Fig.~\ref{fig:16BC} and
948: Table~\ref{tab:16BC}). This enhancement is directly responsible for a
949: substantial reduction of the kinetic energy along bonds joining these
950: atoms. Correspondingly, it accounts for a crossover from the DBC to VBC
951: stripes in the small $U$ regime when the larger kinetic energy gain
952: becomes crucial.
953:
954:
955: %
956: \begin{table}[b!]
957: \begin{center}
958: \begin{tabular}{cccccccc}
959: \hline\hline
960: & & $i$ & & 1 & 2 & 3 & 4 \\
961: \hline
962: & {\bf VBC} & $\langle n^{}_{{\rm h}i}\rangle$
963: & & 0.326 & 0.136 & 0.030 & 0.007 \\
964: & & $\langle S_i^z\rangle$
965: & & 0.118 & 0.301 & 0.371 & 0.384 \\
966: \hline
967: & & $\langle n^{}_{{\rm h}i}\rangle$
968: & & 0.314 & 0.115 & 0.047 & 0.023 \\ &{\bf DBC} &
969: & & (0.323) &(0.110) &(0.046) & (0.021) \\
970: & & $\langle S_i^z\rangle$
971: & & 0.145 & 0.322 & 0.365 & 0.378 \\ & &
972: & & (0.155) &(0.333) &(0.368) &(0.380) \\
973: \hline\hline
974: \end{tabular}
975: \end{center}
976: \caption {The same as in Table~\ref{tab:16SC} but for the bond-centered
977: stripes. VBC stripe is unstable in the extended hopping model with
978: $t'/t=-0.15$.}
979: \label{tab:16BC}
980: \end{table}
981:
982: We would like to emphasize that the above transition between different
983: types of stripe phases is not an artefact of the HF and occurs
984: also between filled stripes obtained within more realistic approaches
985: including local electron correlations. Indeed, slave-boson studies
986: of the Hubbard model at the doping $x=1/9$ have
987: established that the transition from the filled VSC to DSC stripe phase
988: appears at the value $U/t\simeq 5.7$, being much higher than that
989: predicted by the HF, which yields $U/t\simeq 3.8$ \cite{Sei98}.
990: In this method, enhanced stability of the VSC stripes follows from an
991: additional variational parameter per each site $d_i$, reducing the
992: on-site energy without a strong suppression of the kinetic energy.
993: Remarkably, the total energy difference between the vertical SC and BC
994: stripes at both doping levels is comparable to the accuracy of the
995: present calculation. Such degeneracy was also reported in the HF studies
996: of the charge-transfer model \cite{Miz97}. However, when electron
997: correlations are explicitly included the BC stripes are more stable at
998: and above $x=1/8$ doping \cite{Fle01,Wre04}.
999:
1000: \subsection{\label{sec:3c} Effect of the next-neighbor hopping $t'$}
1001:
1002: We now turn to the effect of a next-neighbor hopping $t'$ on the relative
1003: stability of the stripes. There are numerous experimental and theoretical
1004: results which support the presence of finite $t'$ in the cuprates. For
1005: example, recent slave-boson studies have revealed that the phenomena of
1006: the half-filled vertical stripes in LSCO requires a finite next-neighbor
1007: hopping $t'/t\simeq -0.2$ \cite{Sei04}.
1008:
1009: Let us pause now for a moment to clarify the influence of $t'$ on the
1010: DOS as well as on the FS using the electronic band which follows from
1011: a simple tight-binding model,
1012: %
1013: \begin{equation}
1014: E({\bf k}) = -2t(\cos k_x+\cos k_y) - 4t'\cos k_x\cos k_y.
1015: \label{eq:ek}
1016: \end{equation}
1017: %
1018: By the reduction from the CuO$_2$ multiband model to an effective
1019: single-band model it has been found that $t>0$ and $t'<0$ for hole doped
1020: system, and $t<0$ and $t'>0$ in electron doped system \cite{Fei96}.
1021: Although an accidental cancellation of the various contributions results
1022: in almost perfect electron-hole symmetry of the nearest-neighbor hopping
1023: $t$, the next-neighbor hopping $t'$ asymmetry appears owing to the fact
1024: that the dominant contribution to the latter comes from a direct O-O
1025: hopping $t_{pp}$ in the case of a hole hopping. On the contrary, an
1026: electron hopping follows from a third order
1027: Cu$\rightarrow$O$\rightarrow$O$\rightarrow$Cu process, being therefore
1028: dominated by the Cu-O hopping element $t_{pd}$.
1029:
1030: %
1031: \begin{figure}[t!]
1032: \begin{center}
1033: \includegraphics[width=0.47\textwidth ]{fig7.eps}
1034: \end{center}
1035: \caption
1036: {
1037: Effect of the next-neighbor hopping $t'/t=-0.3$ on the noninteracting 2D
1038: DOS at the doping $x=1/4$:
1039: (a) hole doping ($t=1$); (b) electron doping ($t=-1$).
1040: Dotted line shows the Fermi energy in the undoped case, whereas the gray
1041: area shows the states occupied by either electrons (a) or holes (b).
1042: }
1043: \label{fig:DOS}
1044: \end{figure}
1045: %
1046:
1047: In the noninteracting limit the role of $t'$ is to shift the van Hove
1048: singularity away from the middle of the band, either to higher or to
1049: lower energy depending on its sign \cite{Fle97}. Fig.~\ref{fig:DOS} shows
1050: the tight-binding DOS, centered at $\omega=0$ with the
1051: condition $\int N(\omega)\omega d\omega=0$, and the occupied states at
1052: the doping $x=1/4$. In the hole-doped case, with the vacuum as the zero
1053: electron state, the van Hove singularity lies in the lower part of the
1054: band. Conversely, in the case of electron doping,
1055: with the vacuum as the zero hole state, the van Hove singularity is
1056: shifted towards higher energy part of the band, unoccupied by holes.
1057:
1058: Apart from breaking the electron-hole symmetry, the extra parameter $t'$
1059: modifies the shape of the FS of the free electrons and indeed
1060: it becomes more consistent with the FS topology seen by ARPES
1061: \cite{Kin93,Arm02,Ino02}. In the electron-doped system NCCO, the
1062: low-energy spectral weight at the doping $x=0.04$ is concentrated in
1063: small electron pockets around the ($\pm\pi,0$) and ($0,\pm\pi$) points.
1064: Upon increasing doping, one observes both the modification of the hole
1065: pockets and the emergence of new low-lying spectral weights around
1066: ($\pm\pi/2,\pm\pi/2$). Finally, at $x=0.15$ the FS pieces evolve into a
1067: large holelike curve centered at $M=(\pi,\pi)$. In contrast, it has been
1068: observed that in the lightly doped regime ($x=0.03$) doped holes
1069: in LSCO enter into the hole pockets around ($\pm\pi/2,\pm\pi/2$) points
1070: \cite{Yos03}, implying that the FS is holelike and centered at the $M$
1071: point. However, in the heavy overdoped regime $x=0.3$ it converts into
1072: the electronlike FS around the $\Gamma=(0,0)$ point.
1073:
1074:
1075: %
1076: \begin{figure}[t!]
1077: \begin{center}
1078: \includegraphics[width=0.47\textwidth ]{fig8.eps}
1079: \end{center}
1080: \caption
1081: { FS obtained in the tight binding model at the doping $x=1/4$:
1082: (a) hole doping with $t=1$ and: $t'=-0.3$ (black solid line), $t'=0.3$
1083: (gray solid line), and $t'=0$ (dashed line);
1084: (b) electron doping with $t=-1$ and:
1085: $t'=0.3$ (black solid line), $t'=-0.3$ (gray solid line), and
1086: $t'=0$ (dashed line).
1087: The long-dashed line in both panels corresponds to the undoped case
1088: with $t'=0$. The excessively large value of $|t'|=0.3$ as compared to
1089: LSCO was chosen only for more clarity of the figure.
1090: }
1091: \label{fig:FSth}
1092: \end{figure}
1093: %
1094: %
1095: \begin{figure}[t!]
1096: \begin{center}
1097: \includegraphics[width=0.47\textwidth ]{fig9.eps}
1098: \end{center}
1099: \caption
1100: { Eigenenergy maps of the tight-binding model (\ref{eq:ek}) with
1101: $t'/t=-0.3$ as obtained for:
1102: (a) hole doping ($t=1$); (b) electron doping ($t=-1$).
1103: }
1104: \label{fig:ek}
1105: \end{figure}
1106: %
1107:
1108:
1109: Fig.~\ref{fig:FSth}(a) shows that the model (\ref{eq:ek}) with $t'=0$ has
1110: a nested square FS at half-filling which becomes electronlike and shrinks
1111: around the $\Gamma$ point upon hole doping. However, negative $t'=-0.3$
1112: removes the FS nesting at half filling, and the FS expands in the
1113: $(\pm k,0)$ and $(0,\pm k)$ directions, while contracts along the nodal
1114: $(k,\pm k)$ and $(\pm k,k)$ directions due to a large gradient $dE/dk$
1115: along the latter. Indeed, the eigenenergy map, illustrated in
1116: Fig.~\ref{fig:ek}(a), has in this case a valleylike character with a
1117: minimum at the $\Gamma$ point. Therefore the FS turns into a holelike one
1118: with experimentally observed arcs [\textit{cf}. Fig.~\ref{fig:FSth}(a)].
1119: In contrast, the nearest neighbor hopping $t'$ with the same sign as $t$
1120: interchanges the expansion- and contraction directions
1121: which results in the electronlike FS.
1122:
1123:
1124:
1125: %
1126: \begin{table*}[t!]
1127: \begin{center}
1128: \begin{tabular}{crccrrcc}
1129: \hline\hline
1130: \multicolumn{1}{c} {} &\multicolumn{1}{c} {$t'/t$} &
1131: \multicolumn{1}{c} {$E_t^x/t$} &\multicolumn{1}{c} {$E_t^y/t$} &
1132: \multicolumn{1}{c} {$E_{t'}^{x-y}/t$}&\multicolumn{1}{c} {$E_{t'}^{x+y}/t$} &
1133: \multicolumn{1}{c} {$E_U/t$} &\multicolumn{1}{c} {$E_{\textrm{tot}}/t$}\\
1134: \hline
1135: {\bf VSC} & $-$0.15 & $-$0.6876 & $-$0.5886 & 0.0140
1136: & 0.0140 & 0.4778 & $-$0.7704 \\
1137: {\bf DBC} & $-$0.15 & $-$0.6279 & $-$0.6279 & 0.0000
1138: & 0.0183 & 0.4562 & $-$0.7813 \\
1139: {\bf DSC} & $-$0.15 & $-$0.6275 & $-$0.6275 & 0.0000
1140: & 0.0188 & 0.4533 & $-$0.7829 \\
1141: \hline
1142: {\bf DBC} & 0.15 & $-$0.6442 & $-$0.6442 & 0.0000
1143: & $-$0.0282& 0.4883 & $-$0.8283 \\
1144: {\bf DSC} & 0.15 & $-$0.6437 & $-$0.6437 & 0.0000
1145: & $-$0.0279& 0.4855 & $-$0.8298 \\
1146: {\bf VB(S)C}& 0.15 & $-$0.6612 & $-$0.6372 & $-$0.0169
1147: & $-$0.0169& 0.4997 & $-$0.8325 \\
1148: \hline\hline
1149: \end{tabular}
1150: \end{center}
1151: \caption {Energies per site: ground-state energy $E_{\rm tot}$,
1152: kinetic energy contributions for the bonds along (10) $E_t^{x}$,
1153: (01) $E_t^{y}$, (11) $E_{t'}^{x-y}$ and ($1\bar{1}$) $E_{t'}^{x+y}$
1154: directions, as well as the potential energy $E_U$, all normalized per one
1155: site, in the extended hopping Hubbard model with $U/t=5$ and $x=1/8$.
1156: VBC stripe is unstable at $t'/t=-0.15$.}
1157: \label{tab:16Etp}
1158: \end{table*}
1159: %
1160:
1161: Regarding the electron doped case with $t=-1$, shown in Fig.
1162: \ref{fig:FSth}(b), positive $t'=0.3$ (dark solid line) also leads to the
1163: appearance of arc segments of the FS and makes it closer to experimental
1164: observations. In this case, however, the minimum energy is found at
1165: the $M$ point, as illustrated in Fig.~\ref{fig:ek}(b).
1166: It should be noted in passing that this FS describes the same situation
1167: as the one obtained with $t=1$ and $t'=0.3$, indicated by the gray solid
1168: line in Fig.~\ref{fig:FSth}(a). In fact, the sign of $t$ is less important
1169: and turns out to be equivalent to the $(\pi,\pi)$ shift of the momentum
1170: without changing the corresponding eigenvalues. Consequently, in order to
1171: imitate the effect of hole and electron doping it is sufficient to study
1172: the Hamiltonian (\ref{eq:Hubb}) only below half-filling and the alternation
1173: between two regimes is possible by the particle-hole transform,
1174: %
1175: \begin{equation}
1176: c^{\dag}_{i\sigma}\rightarrow (-1)^{i}c^{}_{i\sigma},
1177: \label{eq:phtr}
1178: \end{equation}
1179: %
1180: mapping the model (\ref{eq:ek}) with $t'<0$ onto the one with $t'>0$.
1181: Therefore, in order to avoid any further confusion concerning the signs
1182: of $t$ and $t'$ in Eq.~(\ref{eq:ek}), we set hereafter
1183: $t$ to be positive; then a negative $t'$ ($t'/t<0$) corresponds to hole
1184: doping, whereas a positive one ($t'/t>0$) indicates electron doping.
1185:
1186: The remarkable differences of the electronic structure due
1187: to the broken hole-electron symmetry by $t'$, result in different phase
1188: diagrams of LSCO and NCCO. In the former the long-range AF order is already
1189: suppressed in the lightly doped regime $x\simeq 0.03$, while in the latter
1190: the antiferromagnetism is known to be quite robust at increasing electron
1191: doping, hence
1192: only {\em commensurate} spin fluctuations are observed at $x=0.15$
1193: \cite{Yam03}. The robustness of the commensurate spin fluctuations in the
1194: electron doped regime is consistent with the ED studies of the $t$-$t'$-$J$
1195: \cite{Toh94,Goo94} and $t$-$t'$-$t''$-$J$ \cite{Toh03,Toh04} models. It is
1196: also supported by the conclusion that a negative $t'$ promotes
1197: incommensuration at a lower doping level than a positive one,
1198: reached using the QMC technique applied to the extended Hubbard model
1199: \cite{Duf95}. Finally, the XPS measurements in NCCO show that the
1200: chemical potential monotonously increases with electron doping
1201: \cite{Har01}, whereas its shift is suppressed in the underdoped region of
1202: LSCO \cite{Ino97}. These data have been nicely reproduced in
1203: Ref. \cite{Toh03} for both compounds, except for the low doping regime of
1204: LSCO where stripes are expected. All these numerical and experimental
1205: results indicate that doped electrons might selforganize in a different
1206: way than holes do --- in the latter case DWs are formed.
1207: Nevertheless, stable diagonal stripes with one doped electron per site in
1208: a DW have been obtained in the slave-boson studies of a more realistic
1209: extended three-band model \cite{Sad00}, so the problem is still open.
1210:
1211: Turning back to the competition between stripes in a doped system,
1212: Fig.~\ref{fig:tpU}(a) shows that negative $t'$ stabilizes the DSC stripes,
1213: whereas positive $t'$ favors the VSC ones, within the parameter range where
1214: $t'$ does not drive a stripe melting. Analogous crossover from vertical
1215: stripes at small $|t'|$ to more complex in shape diagonal ones at
1216: $t'/t=-0.1$ and $t'/t=-0.2$ has been found in other HF studies \cite{Nor02}.
1217: The explanation is contained in Table~\ref{tab:16Etp}: negative $t'$ gives
1218: a positive kinetic energy contribution, which is much more readily minimized
1219: by the diagonal charge configuration. Indeed, despite the solitonic
1220: mechanism yielding a noticeable kinetic energy loss due to the transverse
1221: hopping $t'/t=-0.15$, the overall kinetic energy loss in the case of DSC
1222: stripes along the diagonal $(11)$ and antidiagonal $(1\bar{1})$ directions
1223: is smaller than the corresponding one for the VSC stripe. A more careful
1224: analysis shows that hole propagation along the DSC stripe results in a
1225: contribution having the same sign as $t'$. However, it is entirely
1226: canceled by the ones coming from diagonal bonds of the AF domains so that
1227: $E_{t'}^{x-y}=0$.
1228:
1229: %
1230: \begin{figure}[t!]
1231: \begin{center}
1232: \includegraphics[width=0.47\textwidth ]{fig10.eps}
1233: \end{center}
1234: \caption
1235: {
1236: Phase boundaries for: (a) site-centered, and (b) bond-centered stripes as
1237: obtained in the extended Hubbard model with the next-neighbor
1238: hopping $t'$ for doping $x=1/8$ (solid line) and $x=1/6$ (dashed line).
1239: }
1240: \label{fig:tpU}
1241: \end{figure}
1242: %
1243:
1244: One observes further that positive $t'$ reduces the anisotropy between the
1245: kinetic energy gains in the $x$- and $y$-directions for the VSC stripes, and
1246: makes their sum more favorable, while negative $t'$ has the opposite effect.
1247: For the DSC stripes the total kinetic energy also follows the same trend.
1248: The explanation of these results follows from the reinforcement of stripe
1249: order by a negative $t'$ (\textit{cf}. values in parenthesis in
1250: Table~\ref{tab:16SC}), which suppresses the hopping contributions, and its
1251: smearing out by positive $t'$ where hopping is enhanced.
1252: These trends agree with the earlier finding within
1253: the DMFT that the VSC stripe phase is
1254: destabilized by kink fluctuations \cite{Fle01}. However, this stripe
1255: (dis)ordering tendency also leads to a considerably greater change in the
1256: Coulomb energy $E_U$, listed in Table~\ref{tab:16Etp}, for the DSC than for
1257: VSC stripes, which contributes significantly to the predominance of the
1258: former structure for negative $t'$. In fact, it follows from the increase of
1259: hole density within the nonmagnetic stripes and the magnetization density
1260: enhancement within the AF domains (\textit{cf}. Table~\ref{tab:16SC}).
1261:
1262: Like their SC counterparts, DBC stripes are also stabilized by negative
1263: $t'$ resulting in a phase diagram shown in Fig.~\ref{fig:tpU}(b). In this
1264: case, expelling holes from the AF domains enhances not only magnetization
1265: of their atoms but also increases magnetic moment of the hole rich DWs,
1266: as illustrated in Table \ref{tab:16BC}. This enhancement must, however,
1267: strongly suppress the dominant transverse kinetic energy gain of the
1268: VBC stripes. Therefore, the latter are already unstable at $t'/t=-0.15$.
1269:
1270: It is worth noting that a finite diagonal hopping $t'$ should directly
1271: affect the competition between the $d$-wave pairing correlations and
1272: stripes. Indeed, a systematic comparison of stripe and pairing
1273: instabilities within the DMRG framework has shown that when the stripes
1274: are weakened by positive $t'$, the latter are strongly enhanced due to
1275: increasing pair mobility\cite{Whi99}. This effect is accompanied by a
1276: simultaneous enhancement of the AF correlations \cite{Toh04}. Conversely,
1277: negative $t'$ reinforcing a static stripe order results in the suppression
1278: of pair formation in the underdoped region, as found both in the DMRG
1279: technique and Variational Monte Carlo (VMC) \cite{Him02}. However, the
1280: enhanced pairing correlation, attributed to the change of the FS topology
1281: in LSCO, has been obtained in the optimally doped and overdoped regimes
1282: \cite{Shi04}.
1283:
1284: \subsection{\label{sec:3d} Effect of the nearest-neighbor Coulomb
1285: interaction $V$}
1286:
1287: %
1288: \begin{figure}[b!]
1289: \begin{center}
1290: \includegraphics[width=0.47\textwidth ]{fig11.eps}
1291: \end{center}
1292: \caption
1293: {
1294: Phase diagrams for the site-centered (a) and bond-centered (b) stripes
1295: obtained in the extended Hubbard model with the nearest-neighbor
1296: Coulomb interaction $V$ for doping $x=1/8$ (solid line) and $x=1/6$
1297: (dashed line).
1298: }
1299: \label{fig:VU}
1300: \end{figure}
1301: %
1302:
1303: We now investigate the changes in the stripe stability due to either
1304: repulsive ($V>0$) or attractive ($V<0$) nearest-neighbor Coulomb
1305: interaction, which give the phase boundaries between the VSC and DSC
1306: stripe phases shown in Fig.~\ref{fig:VU}(a). We have found that realistic
1307: repulsive $V$ favors the latter. The tendency towards the DSC stripe
1308: formation at $V>0$ is primarily due to a large difference between charge
1309: densities at the atoms of the DW itself and at all their nearest-neighbor
1310: sites, a situation which is avoided in the case of VSC stripe phases
1311: (\textit{cf}. Fig.~\ref{fig:16SC}). Consequently, the former
1312: optimize better the repulsive potential energy component $E_V$, as shown
1313: by the data reported in Table~\ref{tab:16EV}. Similarly, the fact that the
1314: nearest-neighbor interaction $V$ is well minimized only by inhomogeneous
1315: charge densities makes the DBC stripe phase more
1316: favorable than the VBC one, as shown in Fig~\ref{fig:VU}(b).
1317: While this is also the leading mechanism for both diagonal stripe
1318: suppression at $V<0$, the asymmetry of the curve in Fig.~\ref{fig:VU}
1319: arises from the fact that the lower $U$ values at the transition favor
1320: the higher kinetic energy contributions available for the vertical stripes.
1321:
1322: %
1323: \begin{table}[t!]
1324: \begin{center}
1325: \begin{tabular}{crcccrc}
1326: \hline\hline
1327: \multicolumn{1}{c} {} &\multicolumn{1}{c} {$V/t$} &
1328: \multicolumn{1}{c} {$E_t^x/t$} &\multicolumn{1}{c} {$E_t^y/t$} &
1329: \multicolumn{1}{c} {$E_U/t$} &\multicolumn{1}{c} {$E_V/t$} &
1330: \multicolumn{1}{c} {$E_{\textrm{tot}}/t$} \\
1331: \hline
1332: {\bf DBC} & $-$0.4 & $-$0.6322 & $-$0.6322
1333: & 0.4626 & $-$0.6194 & $-$1.4212 \\
1334: {\bf DSC} & $-$0.4 & $-$0.6319 & $-$0.6319
1335: & 0.4602 & $-$0.6193 & $-$1.4229 \\
1336: {\bf VB(S)C}& $-$0.4 & $-$0.6655 & $-$0.6083
1337: & 0.4749 & $-$0.6251 & $-$1.4240 \\
1338: \hline
1339: {\bf VB(S)C}& 0.4 & $-$0.6838 & $-$0.6214
1340: & 0.5063 & 0.6207 & $-$0.1782 \\
1341: {\bf DBC} & 0.4 & $-$0.6424 & $-$0.6424
1342: & 0.4829 & 0.6176 & $-$0.1843 \\
1343: {\bf DSC} & 0.4 & $-$0.6412 & $-$0.6412
1344: & 0.4789 & 0.6171 & $-$0.1864 \\
1345: \hline\hline
1346: \end{tabular}
1347: \end{center}
1348: \caption {Energies per site:
1349: ground-state energy $E_{\rm tot}$, kinetic energy
1350: $(E_t^{x}, E_t^{y})$ and potential energy $(E_U, E_V)$ components
1351: in the extended Hubbard model with the nearest-neighbor Coulomb interaction
1352: $V$ for $U/t=5$ and $x=1/8$.}
1353: \label{tab:16EV}
1354: \end{table}
1355:
1356: However, it has been argued based on the results obtained using the SBA
1357: that an increasing repulsive interaction $V$ favors half-filled vertical
1358: stripes, hence the latter take over at $V/t\simeq 0.1$ in the parameter
1359: regime of $x=1/8$ and $U/t=10$ \cite{Sei98V}. This finding could naturally
1360: explain the appearance of filled diagonal stripes in the nickelates,
1361: provided that they were characterized by a small $V$ term, and the
1362: stability of the half-filled vertical ones in the Nd-codoped cuprates due
1363: to possibly larger value of $V$. It is also worth mentioning other HF
1364: \cite{Kat00} and variational \cite{Ros03} studies in which a variety of
1365: intriguing stripe phases, coexisting at $V/t\simeq 1.5$ with charge order,
1366: has been found in a broad doping region.
1367:
1368: \subsection{\label{sec:3e} Effect of the lattice deformations}
1369:
1370: %
1371: \begin{figure}[t!]
1372: \begin{center}
1373: \includegraphics[width=0.47\textwidth ]{fig12.eps}
1374: \end{center}
1375: \caption
1376: { Local hole $n^{}_{\rm h}(l_x)$ (top) and magnetization $S_{\pi}(l_x)$
1377: (second row) density; fractional change of the length for the bonds to
1378: the right nearest-neighbor along the $x$-direction $u_{x}^{(0)}$ (circles)
1379: and double occupancy $D(l_x)$ (squares) (third row), as well as the
1380: kinetic energy $E_{t}^{x}(l_x)$ projected on the
1381: bonds in the $x$-direction (bottom) of
1382: the VSC (left) and DSC (right) stripe phases, as obtained in
1383: the Peierls-Hubbard model (\ref{eq:hpho}) with $U/t=5$, $\lambda=0.5$
1384: and $x=1/8$ (filled symbols). For comparison the results obtained with
1385: $\lambda=0$ are shown by open symbols.
1386: }
1387: \label{fig:SC16pho}
1388: \end{figure}
1389: %
1390:
1391:
1392: So far, we have demonstrated that a finite anisotropy of
1393: the transfer integral $t$ can tip the balance between vertical and
1394: diagonal stripes. Here we will show that such anisotropy naturally emerges
1395: in a doped system with DWs, described by a single-band Peierls-Hubbard
1396: Hamiltonian,
1397: %
1398: \begin{equation}
1399: H= -\sum_{ij\sigma} t^{}_{ij}(u^{}_{ij})
1400: c^{\dag}_{i\sigma}c^{}_{j\sigma}
1401: + U\sum_{i}n^{}_{i\uparrow}n^{}_{i\downarrow}
1402: + \tfrac{1}{2}K\sum_{\langle ij\rangle}u^2_{ij}.
1403: \label{eq:hpho}
1404: \end{equation}
1405: %
1406: In this model we keep only the leading term and assume a linear dependence
1407: of the nearest neighbor hopping element $t^{}_{ij}$ on the
1408: lattice displacements $u^{}_{ij}$,
1409: %
1410: \begin{equation}
1411: t^{}_{ij}(u^{}_{ij}) = t^{}_0(1+\alpha u^{}_{ij}).
1412: \label{eq:tuij}
1413: \end{equation}
1414: %
1415: Furthermore, we include the elastic energy $\propto K$ which allows to
1416: investigate the stability of the system with respect to a given lattice
1417: deformation and to determine the equilibrium configuration.
1418: For convenience, we parametrize the electron-lattice coupling with a single
1419: quantity, $\lambda = {\alpha^2t_0/K}$, with the parameter values
1420: $K/t_0=18$\AA$^{-2}$ and $\alpha=3$\AA$^{-1}$ assumed following the
1421: earlier HF studies \cite{Zaa96}. As previously, we focus on the doping
1422: $x=1/8$ ($x=1/6$) and present the results of calculations performed on
1423: $16\times 16$ ($12\times 12$) clusters, respectively, with periodic
1424: boundary conditions. These calculations have shown that such clusters
1425: give the most stable filled stripe solutions for the selected doping levels.
1426: The model (\ref{eq:hpho}) was solved self-consistently in real space
1427: within the HF (\ref{eq:MF}).
1428: Thereby, we used an approximate saddle-point formula for the equilibrium
1429: relation between the actual deformation $u_{ij}$ of a given bond and
1430: the bond-charge density $\langle c^{\dag}_{i\sigma}c^{}_{j\sigma}\rangle$,
1431: %
1432: \begin{equation}
1433: u_{ij}^{(0)}\simeq \frac{\alpha t^{}_0}{K}\sum_{\sigma}
1434: \langle c^{\dag}_{i\sigma}c^{}_{j\sigma} + h.c.\rangle,
1435: \label{eq:ph}
1436: \end{equation}
1437: %
1438: being a consequence of the linearity assumption in Eq.~(\ref{eq:tuij}).
1439:
1440: %
1441: \begin{figure}[t!]
1442: \begin{center}
1443: \includegraphics[width=0.47\textwidth ]{fig13.eps}
1444: \end{center}
1445: \caption
1446: {
1447: The same as in Fig.~\ref{fig:SC16pho} but for the bond-centered stripes.
1448: }
1449: \label{fig:BC16pho}
1450: \end{figure}
1451: %
1452:
1453: Quite generally, it is a widely spread out belief that inhomogeneous
1454: states at finite doping are very sensitive to small changes of $\lambda$,
1455: supported both by the HF \cite{Yon92,Yon93} and ED studies \cite{Dob94}.
1456: Further, it has been shown that the electron-lattice interaction
1457: favors DW solutions over other possible phases, such as isolated polarons
1458: or bipolarons \cite{Zaa96}. Therefore, a complete discussion of the stripe
1459: phase stability in correlated oxides has to include the coupling to the
1460: lattice.
1461:
1462: We turn now to the most important aspect of this Section. Figs.
1463: \ref{fig:SC16pho} and \ref{fig:BC16pho} illustrate the effect of the
1464: finite electron-lattice coupling $\lambda=0.5$ on the SC and BC stripes,
1465: respectively. Both figures give a clear demonstration that, in contrast
1466: to the hopping anisotropy $\epsilon_t$ (\ref{eq:et}) discussed above,
1467: finite $\lambda$ markedly modifies both the local hole density
1468: (\ref{eq:nh}) and modulated magnetization (\ref{eq:Spi}) [\textit{cf}.
1469: also Table \ref{tab:16SC} with \ref{tab:16SCpho} (SC stripes) and Table
1470: \ref{tab:16BC} with \ref{tab:16BCpho} (BC stripes)]. Basically, the
1471: influence of $\lambda$ resembles the effect of positive $t'$, smearing out
1472: the stripe order by ejecting holes from the DWs, being however much
1473: stronger. In fact, hole delocalization not only suppresses the
1474: magnetization within the AF domains, but also noticeably quenches magnetic
1475: moments of the BC domain walls. These trends can be understood by
1476: considering energy increments: the kinetic $E_t$, on-site $E_U$,
1477: and elastic energy $E_K$, as explained below.
1478:
1479: %
1480: \begin{table}[!t]
1481: \begin{center}
1482: \begin{tabular}{ccccccccc}
1483: \hline\hline
1484: & & $i$
1485: & & 1 & 2 & 3 & 4 & 5 \\
1486: \hline
1487: &{\bf VSC} & $\langle n^{}_{{\rm h}i}\rangle$
1488: & & 0.270 & 0.212 & 0.103 & 0.038 & 0.022 \\
1489: & & $\langle S_i^z\rangle$
1490: & & 0.000 & 0.146 & 0.259 & 0.310 & 0.321 \\
1491: \hline
1492: &{\bf DSC} & $\langle n^{}_{{\rm h}i}\rangle$
1493: & & 0.292 & 0.179 & 0.094 & 0.058 & 0.046\\
1494: & & $\langle S_i^z\rangle$
1495: & & 0.000 & 0.193 & 0.277 & 0.305 & 0.314\\
1496: \hline\hline
1497: \end{tabular}
1498: \end{center}
1499: \caption {Local hole $\langle n^{}_{{\rm h}i}\rangle$ and magnetization
1500: $\langle S_i^z\rangle$ density at nonequivalent atoms of the SC stripe
1501: phases, all labeled by decreasing hole density in the $x$-direction,
1502: in the Peierls-Hubbard model on a $16\times 16$ cluster with $U/t=5$,
1503: $\lambda=0.5$ and $x=1/8$. }
1504: \label{tab:16SCpho}
1505: \end{table}
1506: %
1507: %
1508: \begin{table}[!b]
1509: \begin{center}
1510: \begin{tabular}{cccccccc}
1511: \hline\hline
1512: & & $i$
1513: & & 1 & 2 & 3 & 4 \\
1514: \hline
1515: &{\bf VBC} & $\langle n^{}_{{\rm h}i}\rangle$
1516: & & 0.255 & 0.156 & 0.063 & 0.026 \\
1517: & & $\langle S_i^z\rangle$
1518: & & 0.074 & 0.209 & 0.291 & 0.319 \\
1519: \hline
1520: &{\bf DBC} & $\langle n^{}_{{\rm h}i}\rangle$
1521: & & 0.248 & 0.130 & 0.073 & 0.049 \\
1522: & & $\langle S_i^z\rangle$
1523: & & 0.103 & 0.243 & 0.294 & 0.312 \\
1524: \hline\hline
1525: \end{tabular}
1526: \end{center}
1527: \caption {The same as in Table~\ref{tab:16SCpho} but for the BC
1528: stripe phases. }
1529: \label{tab:16BCpho}
1530: \end{table}
1531: %
1532:
1533: One should realize that a system described by the Hamiltonian
1534: (\ref{eq:hpho}) might be unstable towards lattice deformations only if
1535: the covalency increase is large enough to compensate both the $E_U$ and
1536: $E_K$ energy cost. Without the electron-lattice coupling, a compromise
1537: solution is mainly reached by developing a strong magnetic order in the
1538: AF domains, where a possible kinetic energy gain is irrelevant, and by
1539: forming nonmagnetic or weakly magnetic DWs with large hole density.
1540: As we have already shown, transverse charge fluctuations around the DWs
1541: yield the leading kinetic energy contribution. However, enhanced covalency
1542: and mixing of the lower $\sim\epsilon_d$ and higher $\sim\epsilon_d+U$
1543: energy states between a DW and the surrounding sites partly delocalize
1544: these states and increase double occupancy,
1545: %
1546: \begin{equation}
1547: D(l_x)=\langle n_{(l_x,0),\uparrow}n_{(l_x,0),\downarrow}\rangle.
1548: \label{eq:D}
1549: \end{equation}
1550: %
1551: Indeed, in the $\lambda=0$ case, double occupancy $D(l_x)$ reaches its
1552: maximum at the DWs, as illustrated in Figs.~\ref{fig:SC16pho} and
1553: \ref{fig:BC16pho}. The only exception is the DSC stripe phase (right
1554: panels of Fig.~\ref{fig:SC16pho}) with the largest $D(l_x)$ in the AF
1555: domains. As a consequence, the latter is the most localized one with the
1556: smallest kinetic energy gain (\textit{cf}. Table~\ref{tab:16E}).
1557:
1558: The situation changes when turning on the electron-lattice coupling.
1559: When the electrons couple to the lattice ($\lambda\neq 0$), the bonds
1560: contract, and the saddle point values of the distortions (\ref{eq:ph}):
1561: $u_{ij}^{(0)}=\langle u_{ij}\rangle$ along (10) and (01) direction,
1562: respectively, are finite. However, a nonuniform charge distribution
1563: results in a different bondlength in the cluster. This is illustrated
1564: in Figs.~\ref{fig:SC16pho} and \ref{fig:BC16pho} showing a fractional
1565: change of the length for the bonds to the right nearest-neighbor along
1566: the $x$-direction $u_{x}^{(0)}$ (third row). Although the values of
1567: $u_{ij}^{(0)}$ in the AF domains are also substantial, the largest lattice
1568: deformations $\sim \langle c^{\dag}_{i\sigma}c^{}_{j\sigma}\rangle$ appear
1569: either on the bonds connecting atoms of the DWs with their nearest
1570: neighbors (\textit{cf}. Fig.~\ref{fig:SC16pho}), or on the bonds which join
1571: two atoms of the bond-centered DWs (\textit{cf}. Fig.~\ref{fig:BC16pho}).
1572: Accordingly, a strengthening nearest neighbor hopping (\ref{eq:tuij})
1573: enables a larger kinetic energy gain on these bonds
1574: (\textit{cf}. bottom of Figs.~\ref{fig:SC16pho} and \ref{fig:BC16pho}).
1575:
1576: %
1577: \begin{table}[!b]
1578: \begin{center}
1579: \begin{tabular}{cccccc}
1580: \hline\hline
1581: \multicolumn{1}{c} {} &
1582: \multicolumn{1}{c} {$E_t^x/t$} &\multicolumn{1}{c} {$E_t^y/t$} &
1583: \multicolumn{1}{c} {$E_U/t$} &\multicolumn{1}{c} {$E_K/t$} &
1584: \multicolumn{1}{c} {$E_{\textrm{tot}}/t$} \\
1585: \hline
1586: {\bf DBC} & $-$0.9679 & $-$0.9679
1587: & 0.6478 & 0.2548 & $-$1.0332 \\
1588: {\bf DSC} & $-$0.9670 & $-$0.9670
1589: & 0.6450 & 0.2544 & $-$1.0346 \\
1590: {\bf VB(S)C}& $-$1.0496 & $-$0.9248
1591: & 0.6719 & 0.2638 & $-$1.0387 \\
1592: \hline\hline
1593: \end{tabular}
1594: \end{center}
1595: \caption {Ground-state energy $E_{\rm tot}$ per site, kinetic energy
1596: $(E_t^{x}, E_t^{y})$ and potential energy $(E_U, E_K)$ components, as
1597: obtained in the Peierls-Hubbard model. Parameters: $U/t=5$, $\lambda=0.5$,
1598: and $x=1/8$.}
1599: \label{tab:16Eal}
1600: \end{table}
1601: %
1602:
1603: As expected, the increasing covalency is accompanied by partial quenching
1604: of magnetic moments. In order to appreciate this tendency, let us
1605: consider a site in the AF domain with larger density of $\uparrow$-spin
1606: electrons (at $A$ sublattice). Once the magnetization is reduced, the
1607: corresponding $\uparrow$-spin energy level which belongs to the lower
1608: Hubbard band is pushed upwards, and the $\downarrow$-spin of the upper
1609: Hubbard band goes down. As a result, the locally raised $\uparrow$-spin
1610: state becomes stronger mixed with $\downarrow$-spin states at the
1611: surrounding sites of $B$ sublattice, and simultaneously bond-charge
1612: density increases. At the same time, electrons, jumping forth and back
1613: between the central site with the $\uparrow$-spin polarization and its
1614: nearest neighbors with the $\downarrow$-spin one, enhance considerably
1615: double occupancy $D(l_x)$, as shown in Figs.~\ref{fig:SC16pho} and
1616: \ref{fig:BC16pho}. This weakens the stripe order and results in a more
1617: uniform distribution of $D(l_x)$.
1618: Of course, the increase of the elastic energy and concomitant enhancement
1619: of the on-site energy, both owing to finite bond contractions (\ref{eq:ph}),
1620: is compensated by the kinetic energy gain and the total energy is lowered
1621: (\textit{cf}. Tables \ref{tab:16E} and \ref{tab:16Eal}).
1622:
1623: %
1624: \begin{figure}[t!]
1625: \begin{center}
1626: \includegraphics[width=0.47\textwidth ]{fig14.eps}
1627: \end{center}
1628: \caption
1629: {
1630: Phase diagrams for site-centered (a) and bond-centered (b) stripe
1631: structures as calculated from the Peierls-Hubbard model
1632: for doping $x=1/8$ (solid line) and $x=1/6$ (dashed line).
1633: }
1634: \label{fig:alU}
1635: \end{figure}
1636: %
1637:
1638: We close this Section with the phase diagrams shown in Fig.~\ref{fig:alU}.
1639: They were obtained by varying $U$ and the coefficient $\alpha$, while
1640: maintaining constant $K/t_0=18$\AA$^{-2}$. The increased stability of
1641: vertical stripes follows from the relative stronger enhancement of the
1642: local hopping elements (\ref{eq:tuij}) (and consequently larger gain of
1643: the kinetic energy), especially on the bonds in the direction
1644: perpendicular to the DWs itself.
1645:
1646:
1647: \section{\label{sec:4} Summary}
1648:
1649: In summary, we have shown that a competition between magnetic energy of
1650: interacting almost localized electrons and the kinetic energy of holes
1651: created by doping leads to the formation of new type of coexisting
1652: charge and spin order --- the stripe phases. We have shown that vertical
1653: (horizontal) and diagonal stripes dominate the behavior of the charge
1654: structures formed by doping the antiferromagnet away from half filling,
1655: using the solutions obtained for the Hubbard model within the HF
1656: approximation in the physically interesting regime of the Coulomb
1657: interaction. The detailed charge distribution and the type of stripe
1658: order depend on the ratio $U/t$, on the value of the next-neighbor hopping
1659: $t'$, and on the nearest-neighbor Coulomb interaction $V$.
1660: We have also shown that a strong electron-lattice coupling might be
1661: responsible for the appearance of the vertical stripes observed in the
1662: superconducting cuprates at $x=1/8$.
1663:
1664: Altogether, although some experimentally observed trends could be
1665: reproduced already in the HF approach, the presented results indicate that
1666: strong electron correlations play a crucial role in the stripe phases and
1667: have to be included for a more quantitative analysis. Further progress
1668: both in the experiment and in the theory is necessary to establish the
1669: possible role of stripes in the phenomenon of high temperature
1670: superconductivity.
1671:
1672:
1673: \section{\label{sec:5} Acknowledgments}
1674: M. Raczkowski was supported by a Marie Curie fellowship of the
1675: European Community program under number HPMT2000-141.
1676: This work was supported by the the Polish
1677: Ministry of Scientific Research and Information Technology,
1678: Project No. 1~P03B~068~26, and by the Minist\`ere
1679: Fran\c{c}ais des Affaires Etrang\`eres under POLONIUM 09294VH.
1680:
1681:
1682: \begin{thebibliography}{00}
1683:
1684: \bibitem{Bed86} J.~G. Bednorz and K.~A. M\"uller,
1685: Z. Phys. B {\bf 64}, 189 (1986).
1686:
1687: \bibitem{Car02} E.~W. Carlson, V.~J. Emery, S.~A. Kivelson, D. Orgad,
1688: \textit{Concepts in High Temperature Superconductivity}
1689: in: K.~H. Bennemann, J.~B. Ketterson (Eds.),
1690: The Physics of Conventional and Unconventional
1691: Superconductors, Springer-Verlag (2002), cond-mat/0206217.
1692:
1693: \bibitem{Zaa89} J.~Zaanen and O.~Gunnarsson,
1694: Phys. Rev. B {\bf 40}, 7391 (1989).
1695:
1696: \bibitem{Poi89} D. Poilblanc and T. M. Rice,
1697: Phys. Rev. B {\bf 39}, 9749 (1989).
1698:
1699: \bibitem{Sch89} H.~J. Schulz, J. Phys. (Paris) {\bf 50}, 2833 (1989);
1700: Phys. Rev. Lett. {\bf 64}, 1445 (1990).
1701:
1702: \bibitem{Kat90} M. Kato, K. Machida, H. Nakanishi, and M. Fujita,
1703: J. Phys. Soc. Jpn. {\bf 59} 1047 (1990).
1704:
1705: \bibitem{Inu91} M.~Inui and P.~B.~Littlewood,
1706: Phys. Rev. B {\bf 44}, 4415 (1991).
1707:
1708: \bibitem{Zac98} O. Zachar, S.~A. Kivelson, and V.~J Emery,
1709: Phys. Rev. B {\bf 57} 1422 (1998);
1710: J. Zaanen, Physica C {\bf 317} 217 (1999).
1711:
1712: \bibitem{Kiv96} S.~A. Kivelson and V.~J. Emery,
1713: Synth. Metals {\bf 80}, 151 (1996).
1714:
1715: \bibitem{Low94} U. L\"ow, V.~J. Emery, K. Fabricius, S.~A. Kivelson,
1716: Phys. Rev. Lett. {\bf 72}, 1918 (1994).
1717:
1718: \bibitem{Dui98} C.~N.~A. van Duin and J. Zaanen,
1719: Phys. Rev. Lett. {\bf 80}, 1513 (1998).
1720:
1721:
1722: \bibitem{Whi98} S.~R. White and D.~J. Scalapino,
1723: Phys. Rev. Lett. {\bf 80}, 1272 (1998).
1724:
1725:
1726: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1727:
1728: \bibitem{Whi99} S.~R. White and D.~J. Scalapino,
1729: Phys. Rev. B {\bf 60}, R753 (1999).
1730:
1731: \bibitem{Sei98} G. Seibold, E. Sigmund, and V. Hizhnyakov,
1732: Phys. Rev. B {\bf 57}, 6937 (1998).
1733:
1734: \bibitem{Sei98V} G. Seibold, C. Castellani, C. Di Castro, and M. Grilli,
1735: Phys. Rev. B {\bf 58}, 13506 (1998).
1736:
1737: \bibitem{Sei04} G. Seibold and J. Lorenzana,
1738: Phys. Rev. B {\bf 69}, 134513 (2004).
1739:
1740: \bibitem{Gor99} D. G\'ora, K. Ro\'sciszewski, and A.~M. Ole\'s,
1741: Phys. Rev. B {\bf 60}, 7429 (1999).
1742:
1743: \bibitem{Toh99} T. Tohyama, S. Nagai, Y. Shibata, and S. Maekawa,
1744: Phys. Rev. Lett. {\bf 82}, 4910 (1999).
1745:
1746: \bibitem{Wro00} P. Wr\'obel and R. Eder,
1747: Phys. Rev. B {\bf 62}, 4048 (2000).
1748:
1749: \bibitem{Fle00} M. Fleck, A.~I. Lichtenstein,
1750: E. Pavarini, and A.~M. Ole\'s,
1751: Phys. Rev. Lett. {\bf 84}, 4962 (2000).
1752:
1753: \bibitem{Fle01} M.~Fleck, A.~I.~Lichtenstein, and A.~M.~Ole\'s,
1754: Phys. Rev. B {\bf 64}, 134528 (2001).
1755:
1756: \bibitem{Zac00} M.~G. Zacher, R. Eder, E. Arrigoni, and W. Hanke,
1757: Phys. Rev. Lett. {\bf 85}, 2585 (2000);
1758: Phys. Rev. B {\bf 65}, 045109 (2002).
1759:
1760: \bibitem{Bec01} F. Becca, L. Capriotti, and S. Sorella,
1761: Phys. Rev. Lett. {\bf 87}, 167005 (2001).
1762:
1763: \bibitem{Rie01} J. Riera,
1764: Phys. Rev. B {\bf 64}, 104520 (2001).
1765:
1766: \bibitem{Rac05} M. Raczkowski, R. Fr\'esard, and A. M. Ole\'s,
1767: Physica B {\bf 359-361}, 780 (2005).
1768:
1769: \bibitem{Zaa96} J. Zaanen and A.~M. Ole\'s,
1770: Ann. Phys. (Leipzig) {\bf 5}, 224 (1996).
1771:
1772: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1773:
1774: \bibitem{Kiv03} S.~A. Kivelson, I.~P. Bindloss, E. Fradkin, V. Oganesyan,
1775: J.~M. Tranquada, A. Kapitulnik, and C. Howald,
1776: Rev. Mod. Phys. {\bf 75}, 1201 (2003).
1777:
1778: \bibitem{Dam03} A. Damascelli, Z. Hussain, and Z.-X. Shen,
1779: Rev. Mod. Phys. {\bf 75}, 473 (2003).
1780:
1781: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1782: %%%%%%%%%%%%%LNdSCO%%%%%%%%%%%
1783:
1784: \bibitem{Tra95Nd} J.~M.~Tranquada, B.~J.~Sternlieb, J.~D.~Axe,
1785: Y.~Nakamura, and S.~Uchida,
1786: Nature {\bf 375}, 561 (1995).
1787:
1788:
1789: \bibitem{Tra96Nd} J.~M. Tranquada, J.~D. Axe, N. Ichikawa, Y. Nakamura,
1790: S. Uchida, and B. Nachumi,
1791: Phys. Rev. B {\bf 54}, 7489 (1996).
1792:
1793: \bibitem{Ich00} N. Ichikawa, S. Uchida, J.~M. Tranquada, T. Niem\"oller,
1794: P.~M. Gehring, S.-H. Lee, and J.~R. Schneider,
1795: Phys. Rev. Lett. {\bf 85}, 1738 (2000).
1796:
1797: \bibitem{Tra97Nd} J.~M. Tranquada, J.~D. Axe, N. Ichikawa, A.~R. Moodenbaugh,
1798: Y. Nakamura, and S. Uchida,
1799: Phys. Rev. Lett. {\bf 78}, 338 (1997).
1800:
1801: \bibitem{Sin99} P.~M. Singer, A.~W. Hunt, A.~F. Cederstr\"om, and T. Imai,
1802: Phys. Rev. B {\bf 60}, 15345 (1999).
1803:
1804: \bibitem{Wak03} S. Wakimoto, R.~J. Birgeneau, Y. Fujimaki, N. Ichikawa,
1805: T. Kasuga, Y.~J. Kim, K.~M. Kojima, S.-H. Lee, H. Niko,
1806: J.~M. Tranquada, S. Uchida, and M. von Zimmermann,
1807: Phys. Rev. B {\bf 67}, 184419 (2003).
1808:
1809: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
1810: %%%%%%%%%%%%%LESCO%%%%%%%%%%
1811: \bibitem{Tei00} G.~B. Teitel'baum, B. B\"uchner, and H. de Gronckel,
1812: Phys. Rev. Lett. {\bf 84}, 2949 (2000).
1813:
1814: \bibitem{Kla00} H.-H. Klauss, W. Wagener, M. Hillberg, W. Kopmann,
1815: H. Walf, F.~J. Litterst, M. H\"ucker, and B. B\"uchner,
1816: Phys. Rev. Lett. {\bf 85}, 4590 (2000).
1817:
1818: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
1819: %%%%%%%%%%%%%LBSCO%%%%%%%%%%
1820:
1821: \bibitem{Fuj02Ba} M. Fujita, H. Goka, K. Yamada, and M. Matsuda,
1822: Phys. Rev. Lett. {\bf 88}, 167008 (2002).
1823:
1824: \bibitem{Kim04} H. Kimura, Y. Noda, H. Goka, M. Fujita, K. Yamada,
1825: M. Mizumaki, N. Ikeda, and H. Ohsumi,
1826: Phys. Rev. B {\bf 70}, 134512 (2004).
1827:
1828: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
1829: %%%%%%%%%%%%%LBCO%%%%%%%%%%%
1830:
1831: \bibitem{Fuj04} M. Fujita, H. Goka, K. Yamada, J.~M. Tranquada, and
1832: L.~P. Regnault,
1833: Phys. Rev. B {\bf 70}, 104517 (2004).
1834:
1835:
1836: %%%%%%%%%%%%%%%%%%%%%%%%%%%%
1837: %%%%%%%%%%%%%LSCO%%%%%%%%%%%
1838:
1839: \bibitem{Yam98} K. Yamada, C.~H. Lee, K. Kurahashi, J. Wada, S. Wakimoto,
1840: S. Ueki, H. Kimura, Y. Endoh, S. Hosoya, G. Shirane,
1841: R.~J. Birgeneau, M. Greven, M.~A. Kastner, and Y.~J. Kim,
1842: Phys. Rev. B {\bf 57}, 6165 (1998).
1843:
1844:
1845: \bibitem{Wak99} S. Wakimoto, G. Shirane, Y. Endoh, K. Hirota, S. Ueki,
1846: K. Yamada, R.~J. Birgeneau, M.~A. Kastner, Y.~S. Lee,
1847: P.~M. Gehring, and H.~S. Lee,
1848: Phys. Rev. B {\bf 60}, R769 (1999).
1849:
1850: \bibitem{Wak00} S. Wakimoto, R.~J. Birgeneau, M.~A. Kastner, Y.~S. Lee,
1851: R. Erwin, P.~M. Gehring, S.~H. Lee, M. Fujita, K. Yamada,
1852: Y. Endoh, K. Hirota, and G. Shirane,
1853: Phys. Rev. B {\bf 61}, 3699 (2000).
1854:
1855: \bibitem{Fuj02} M. Fujita, K. Yamada, H. Hiraka, P.~M. Gehring, S.~H. Lee,
1856: S. Wakimoto, and G. Shirane,
1857: Phys. Rev. B {\bf 65}, 064505 (2002).
1858:
1859: \bibitem{Has04} N.~Hasselmann, A.~H. Castro Neto, and C. Morais Smith,
1860: Phys. Rev. B {\bf 69}, 014424 (2004).
1861:
1862: \bibitem{Mat00} M. Matsuda, Y.~S. Lee, M. Greven, M.~A. Kastner,
1863: R.~J. Birgeneau, K. Yamada, Y. Endoh, P. B\"oni,
1864: S.-H. Lee, S. Wakimoto, and G. Shirane,
1865: Phys. Rev. B {\bf 61}, 4326 (2000).
1866:
1867: \bibitem{Mat00a} M. Matsuda, M. Fujita, K. Yamada, R.~J. Birgeneau,
1868: M.~A. Kastner, H. Hiraka, Y. Endoh, S. Wakimoto, and
1869: G. Shirane,
1870: Phys. Rev. B {\bf 62}, 9148 (2000).
1871:
1872: \bibitem{Mat02} M. Matsuda, M. Fujita, K. Yamada, R.~J. Birgeneau,
1873: Y. Endoh, and G. Shirane,
1874: Phys. Rev. B {\bf 65}, 134515 (2002).
1875:
1876: %%%%%%%%%anizo%%%%%%%%%%%%%%%%
1877:
1878: \bibitem{Hun99} A.~W. Hunt, P.~M. Singer, K.~R. Thurber, and T.~Imai,
1879: Phys. Rev. Lett. {\bf 82}, 4300 (1999).
1880:
1881: \bibitem{And02} Y. Ando, K. Segawa, S. Komiya, and A.~N. Lavrov,
1882: Phys. Rev. Lett. {\bf 88}, 137005 (2002).
1883:
1884: \bibitem{Dum03} M. Dumm, S. Komiya, Y. Ando, and D.~N. Basov,
1885: Phys. Rev. Lett. {\bf 91}, 077004 (2003).
1886:
1887: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1888: %%%%%%%%%%%%%LCOx%%%%%%%%%%%%%
1889:
1890: \bibitem{Lee99} B.~O. Wells, Y.~S. Lee, M.~A. Kastner, R.~J. Christanson,
1891: R.~J. Birgeneau, K. Yamada, Y. Endoh, and G. Shirane,
1892: Science {\bf 277}, 1067 (1997);
1893: Y.~S. Lee, R.~J. Birgeneau, M.~A. Kastner, Y. Endoh,
1894: S. Wakimoto, K. Yamada, R.~W. Erwin, S.-H. Lee, and
1895: G. Shirane,
1896: Phys. Rev. B {\bf 60}, 3643 (1999).
1897:
1898: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1899: %%%%%%%%%%%%%Zn-LSCO%%%%%%%%%%
1900: \bibitem{Hir98} K. Hirota, K. Yamada,
1901: I. Tanaka, and H. Kojima, Physica B {\bf 241}, 817 (1998).
1902:
1903: \bibitem{Kim99} H. Kimura, K. Hirota, H. Matsushita, K. Yamada,
1904: Y. Endoh, S.-H. Lee, C.~F. Majkrzak, R. Erwin,
1905: G. Shirane, M. Greven, Y. S. Lee, M.~A. Kastner, and
1906: R.~J. Birgeneau,
1907: Phys. Rev. B {\bf 59}, 6517 (1999);
1908: J.~M. Tranquada, N. Ichikawa, K. Kakurai, and S. Uchida,
1909: J. Phys. Chem. Solids {\bf 60}, 1019 (1999).
1910:
1911: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1912: %%%%%%%%%%%%%YBCO%%%%%%%%%%%%%
1913:
1914: \bibitem{Dai01} P. Dai, H.~A. Mook, R.~D. Hunt, and F. Do\v{g}an,
1915: Phys. Rev. B {\bf 63}, 054525 (2001).
1916:
1917: \bibitem{Moo02} H.~A. Mook, P. Dai, and F. Do\v{g}an,
1918: Phys. Rev. Lett. {\bf 88}, 097004 (2002).
1919:
1920: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1921: %%%%%%%%%%%%%BSCCO%%%%%%%%%%%%%
1922: \bibitem{Moo97} H.~A. Mook and B.~C. Chakoumakos,
1923: J. Superconductivity {\bf 10}, 389 (1997).
1924:
1925: \bibitem{How03} C. Howald, H. Eisaki, N. Kaneko, M. Greven, and
1926: A. Kapitulnik,
1927: Phys. Rev. B {\bf 67}, 014533 (2003);
1928: A. Fang, C. Howald, N. Kaneko, M. Greven, A. Kapitulnik,
1929: Phys. Rev. B {\bf 70}, 214514 (2004).
1930:
1931: %%%%%%%%%ARPES%%%%%%%%%%%%%%%%%%
1932:
1933: %%%%spectra X,S %%
1934: \bibitem{Ino00} A. Ino, C. Kim, M. Nakamura, T. Yoshida, T. Mizokawa,
1935: Z.-X. Shen, A. Fujimori, T. Kakeshita, H. Eisaki, and
1936: S. Uchida,
1937: Phys. Rev. B {\bf 62}, 4137 (2000).
1938:
1939: %%%%%%flatband&FS%%
1940: \bibitem{Ino02} A. Ino, C. Kim, M. Nakamura, T. Yoshida, T. Mizokawa,
1941: A. Fujimori, Z.-X. Shen, T. Kakeshita,
1942: H. Eisaki, and S. Uchida,
1943: Phys. Rev. B {\bf 65}, 094504 (2002).
1944:
1945: %%%%%%%%DOS%%%%%%%
1946: \bibitem{Ino98} A. Ino, T. Mizokawa, K. Kobayashi, A. Fujimori,
1947: T. Sasagawa, T. Kimura, K. Kishio, K. Tamasaku,
1948: H. Eisaki, and S. Uchida,
1949: Phys. Rev. Lett. {\bf 81}, 2124 (1998).
1950:
1951: %%%%%%x=0.12%%%%%%%
1952: \bibitem{Zho99} X.~J. Zhou, P. Bogdanov, S.~A. Kellar, T. Noda,
1953: H. Eisaki, S. Uchida, Z. Hussain, and Z.-X. Shen,
1954: Science {\bf 286}, 268 (1999).
1955:
1956: %%%%%%x=0.15%%%%%%%
1957: \bibitem{Zho01} X.~J. Zhou, T. Yoshida, S.~A. Kellar, P.~V. Bogdanov,
1958: E.~D. Lu, A. Lanzara, M. Nakamura, T. Noda, T. Kakeshita,
1959: H. Eisaki, S. Uchida, A. Fujimori, Z. Hussain, and
1960: Z.-X. Shen,
1961: Phys. Rev. Lett. {\bf 86}, 5578 (2001).
1962:
1963: \bibitem{Sal96} M.~I. Salkola, V.~J. Emery, and S.~A. Kivelson,
1964: Phys. Rev. Lett. {\bf 77}, 155 (1996).
1965:
1966: \bibitem{Eme97} V. J. Emery, S.~A. Kivelson, and O. Zachar,
1967: Phys. Rev. B {\bf 56}, 6120 (1997).
1968:
1969: \bibitem{Wro05} P. Wr\'obel, A. Maci\k{a}g, and R. Eder,
1970: \texttt{cond-mat/0408703}.
1971:
1972: %%%%%%%%%mu%%%%%%%
1973: \bibitem{Ino97} A. Ino, T. Mizokawa, A. Fujimori, K. Tamasaku, H. Eisaki,
1974: S. Uchida, T. Kimura, T. Sasagawa, and K. Kishio,
1975: Phys. Rev. Lett. {\bf 79}, 2101 (1997).
1976:
1977: \bibitem{Nod99} T. Noda, H. Eisaki, and S. Uchida,
1978: Science {\bf 286}, 265 (1999).
1979:
1980: %%%%%%%%%%%%%%%%%%
1981: \bibitem{Yam03} K. Yamada, K. Kurahashi, T. Uefuji, M. Fujita, S. Park,
1982: S.-H. Lee, and Y. Endoh,
1983: Phys. Rev. Lett. {\bf 90}, 137004 (2003).
1984:
1985: \bibitem{Har01} N. Harima, J. Matsuno, A. Fujimori, Y. Onose,
1986: Y. Taguchi, and Y. Tokura,
1987: Phys. Rev. B {\bf 64}, 220507 (2001).
1988:
1989: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1990:
1991:
1992: \bibitem{Fei96} L. F. Feiner, J. H. Jefferson and R. Raimondi,
1993: Phys. Rev. B {\bf 53}, 8751 (1996).
1994:
1995: \bibitem{Miz97} T. Mizokawa and A. Fujimori,
1996: Phys. Rev. B {\bf 56}, 11920 (1997).
1997:
1998: \bibitem{Yu98} Z.~G. Yu, J. Zang, J.~T. Gammel, and A.~R. Bishop,
1999: Phys. Rev. B {\bf 57}, R3241 (1998).
2000:
2001: \bibitem{Sad00} A. Sadori and M. Grilli,
2002: Phys. Rev. Lett. {\bf 84}, 5375 (2000).
2003:
2004: \bibitem{Lor02} J. Lorenzana and G. Seibold,
2005: Phys. Rev. Lett. {\bf 89}, 136401 (2002).
2006:
2007: \bibitem{Buc94} B. B\"uchner, M. Breuer, and A. Freimuth, and A.~P. Kampf,
2008: Phys. Rev. Lett. {\bf 73}, 1841 (1994).
2009:
2010: \bibitem{Nor01} B. Normand and A.~P. Kampf,
2011: Phys. Rev. B {\bf 64}, 024521 (2001).
2012:
2013: \bibitem{Kam01} A.~P. Kampf, D.~J. Scalapino, and S.~R. White,
2014: Phys. Rev. B {\bf 64}, 052509 (2001).
2015:
2016: \bibitem{Tip03} J.~M. Tipper and K.~J.~E. Vos,
2017: Phys. Rev. B {\bf 67}, 144511 (2003).
2018:
2019: \bibitem{Wre04} P. Wr\'obel, A. Maci\k{a}g, and R. Eder,
2020: \texttt{cond-mat/0408578}.
2021:
2022: \bibitem{Fle97} M. Fleck, A. M. Ole\'s, and L. Hedin,
2023: Phys. Rev. B {\bf 56}, 3159 (1997).
2024:
2025: \bibitem{Kin93} D.~M. King, Z.-X. Shen, D.~S. Dessau, B.~O. Wells,
2026: W.~E. Spicer, A.~J. Arko, D.~S. Marshall, J. DiCarlo,
2027: A.~G. Loeser, C.~H. Park, E.~R. Ratner, J.~L. Peng,
2028: Z.~Y. Li, and R.~L. Greene,
2029: Phys. Rev. Lett. {\bf 70}, 3159 (1993).
2030:
2031: \bibitem{Arm02} N.~P. Armitage, F. Ronning, D.~H. Lu, C. Kim,
2032: A. Damascelli, K.~M. Shen, D.~L. Feng, H. Eisaki,
2033: Z.-X. Shen, P.~K. Mang, N. Kaneko, M. Greven, Y. Onose,
2034: Y. Taguchi, and Y. Tokura,
2035: Phys. Rev. Lett. {\bf 88}, 257001 (2002).
2036:
2037: \bibitem{Yos03} T. Yoshida, X.~J. Zhou, T. Sasagawa, W.~L. Yang,
2038: P.~V. Bogdanov, A. Lanzara, Z. Hussain, T. Mizokawa,
2039: A. Fujimori, H. Eisaki, Z.-X. Shen, T. Kakeshita, and
2040: S. Uchida,
2041: Phys. Rev. Lett. {\bf 91}, 027001 (2003).
2042:
2043: \bibitem{Toh94} T. Tohyama and S. Maekawa,
2044: Phys. Rev. B {\bf 49}, 3596 (1994).
2045:
2046: \bibitem{Goo94} R. J. Gooding, K. J. E. Vos, and P. W. Leung,
2047: Phys. Rev. B {\bf 50}, 12866 (1994).
2048:
2049: \bibitem{Toh03} T. Tohyama and S. Maekawa,
2050: Phys. Rev. B {\bf 67}, 092509 (2003).
2051:
2052: \bibitem{Toh04} T. Tohyama,
2053: Phys. Rev. B {\bf 70}, 174517 (2004).
2054:
2055: \bibitem{Duf95} D. Duffy and A. Moreo,
2056: Phys. Rev. B {\bf 52}, 15607 (1995).
2057:
2058: \bibitem{Nor02} B. Normand and A.~P. Kampf,
2059: Phys. Rev. B {\bf 65}, 020509 (2002).
2060:
2061: \bibitem{Him02} A. Himeda, T. Kato, and M. Ogata,
2062: Phys. Rev. Lett. {\bf 88}, 117001 (2002).
2063:
2064: \bibitem{Shi04} C.~T. Shih, T.~K. Lee, R. Eder, C.-Y. Mou, and Y.~C. Chen,
2065: Phys. Rev. Lett. {\bf 92}, 227002 (2004).
2066:
2067: \bibitem{Kat00} T. Kato and M. Kato,
2068: J. Phys. Soc. Jpn. {\bf 69}, 3972 (2000).
2069:
2070: \bibitem{Ros03} K. Ro\'sciszewski and A.~M. Ole\'s,
2071: J. Phys.: Condens. Matter {\bf 15}, 8363 (2003).
2072:
2073: \bibitem{Yon92} K. Yonemitsu, A.~R. Bishop, and J. Lorenzana,
2074: Phys. Rev. Lett. {\bf 69}, 965 (1992).
2075:
2076: \bibitem{Yon93} K. Yonemitsu, A.~R. Bishop, and J. Lorenzana,
2077: Phys. Rev. B {\bf 47}, 8065 (1993);
2078: {\bf 47}, 12059 (1993).
2079:
2080: \bibitem{Dob94} A. Dobry, A. Greco, J. Lorenzana, and J. Riera,
2081: Phys. Rev. B {\bf 49}, 505 (1994);
2082: J. Lorenzana and A. Dobry
2083: Phys. Rev. B {\bf 50}, 16094 (1994).
2084:
2085: \end{thebibliography}
2086:
2087: \end{document}
2088: