1: %\documentstyle[aps,preprint,psfig]{revtex}
2: \documentstyle[aps,prb,twocolumn,floats,psfig]{revtex}
3:
4: \draft
5: \begin{document}
6: \twocolumn[\hsize\textwidth\columnwidth\hsize\csname@twocolumnfalse%
7: \endcsname
8:
9: \title{Suppression of static stripe formation by next-neighbor hopping}
10:
11: \author{B. Normand and A. P. Kampf}
12:
13: \address{Theoretische Physik III, Elektronische Korrelationen und
14: Magnetismus, Institut f\"ur Physik, \\ Universit\"at Augsburg, D-86135
15: Augsburg, Germany}
16:
17: \date{\today}
18:
19: \maketitle
20:
21: \begin{abstract}
22:
23: We show from real-space Hartree-Fock calculations within the extended
24: Hubbard model that next-nearest neighbor ($t^{\prime}$) hopping processes
25: act to suppress the formation of static charge stripes. This result is
26: confirmed by investigating the evolution of charge-inhomogeneous corral
27: and stripe phases with increasing $t^{\prime}$ of both signs. We propose
28: that large $t^{\prime}$ values in YBCO prevent static stripe formation,
29: while anomalously small $t^{\prime}$ in LSCO provides an additional reason
30: for the appearance of static stripes only in these systems.
31:
32: \end{abstract}
33: \bigskip
34: ]
35:
36: The presence of charge-inhomogeneous and striped phases as candidate
37: ground states for the cuprates continues to be one of the most contentious
38: issues in high-temperature superconductivity (for a review and references
39: see Ref.~\onlinecite{rnk}). Static stripes have been observed
40: experimentally, but to date only in rare-earth-doped La$_{2-x}$Sr$_x$CuO$_4$
41: (LSCO) systems with the low-temperature tetragonal (LTT) structural
42: distortion. Theoretical explanations for stripes fall into three categories:
43: i) they are a true ground state intrinsic to short-ranged models of the
44: CuO$_2$ planes, ii) they are a competing excited state stabilized by
45: anisotropy, or iii) they emerge when long-ranged interactions are
46: invoked (generally to frustrate phase separation). We have recently
47: subscribed to the second viewpoint, motivated by the direct experimental
48: connection of lattice structure with stripe formation and suppression of
49: superconductivity.\cite{rnk}
50:
51: However, the interpretation of striped phases as a consequence of
52: lattice-induced hopping and superexchange anisotropy in the CuO$_2$ plane
53: leaves some open questions. On the qualitative level, these include the
54: absence of static stripes in orthorhombic YBa$_2$Cu$_3$O$_{7-\delta}$
55: (YBCO) systems, while quantitatively they
56: include the value of the relative anisotropies required to stabilize a
57: stripe phase. A key issue long recognized as a source of major differences
58: between cuprate compounds is the shape of the Fermi surface.\cite{rtm}
59: This has been measured by angle-resolved photoemission (ARPES) for
60: Bi$_2$Sr$_2$CaCu$_2$O$_{8+\delta}$ (BSCCO),\cite{rdea} LSCO,\cite{riea}
61: and Nd$_{2-x}$Ce$_x$CuO$_4$ (NCCO)\cite{rka} systems, and is
62: most easily modeled by extended tight-binding band structures.\cite{rtm}
63: The primary influence of an extended band structure may be encapsulated
64: in the single parameter $t^{\prime}$ governing the next-nearest-neighbor
65: hopping on the square lattice. An investigation of $t^{\prime}$ effects
66: is required in context of stripes, and is provided here within the
67: real-space Hartree-Fock (RSHF) technique.\cite{rnk}
68:
69: The effects of next-neighbor hopping on stripes have been considered in
70: the extended $t$-$J$ model by exact diagonalization (ED)\cite{rtgsclmd}
71: and by density-matrix renormalization group (DMRG) calculations.\cite{rws}
72: Both sets of authors reported a suppression of horizontal (or vertical)
73: stripe order with increasing $|t^{\prime}|$ of both signs, but detailed
74: features of the results were not fully consistent. The authors of
75: Ref.~\onlinecite{rws} also made a systematic comparison of stripe and
76: pairing instabilities, finding a strong anticorrelation. A $t^{\prime}$
77: opposite in sign to $t$ was included in a more general study of the
78: extended ($t$-$t^{\prime}$) Hubbard model by RSHF,\cite{rvvg} which
79: confirmed the tendency towards stripe suppression. The influence of
80: $t^{\prime}$ on stripes has been considered in the context of Fermi-surface
81: geometry,\cite{rim} and in terms of its effects on stripe filling.\cite{rmobb}
82: Finally, extended hopping integrals were used in an ED study of doped
83: stripes,\cite{rmgxfd} where their role was primarily to avoid phase
84: separation. We return below to a more detailed discussion of these
85: results.
86:
87: The extended, anisotropic Hubbard model is given by
88: \begin{eqnarray}
89: H & = & - \sum_{i, {\eta = x,y}, \sigma } t_{\eta} ( c_{i \pm \eta
90: \sigma}^{\dag} c_{i\sigma} + H. c.) + U \sum_i n_{i \uparrow} n_{i
91: \downarrow} , \label{eehh} \nonumber \\ & & \;\; - \sum_{i, {\eta = x \pm y},
92: \sigma } t_{\eta}^{\prime} ( c_{i \pm \eta \sigma}^{\dag} c_{i\sigma} +
93: H. c.) ,
94: \end{eqnarray}
95: where $n_{i\sigma} = c_{i\sigma}^{\dag} c_{i\sigma}$. We retain in
96: Eq.~(\ref{eehh}) the possibility of anisotropic nearest-neighbor\cite{rnk}
97: and next-neighbor hopping, $t_x \ne t_y$ and $t_{x+y}^{\prime} \ne
98: t_{x-y}^{\prime}$. However, the symmetry of next-neighbor terms in
99: this one-band model is maintained in the LTT phase of LSCO, and for
100: most of the study to follow. We apply the HF decomposition of
101: the Hubbard term in Eq.~(\ref{eehh}) and seek self-consistent solutions
102: for the static charge and spin configurations.
103:
104: A detailed analysis of the RSHF technique was provided in a recent study
105: of anisotropy effects on stripe formation.\cite{rnk} The solutions were
106: characterized as functions of
107: the ``intrinsic'' system parameters, by which is meant the ratio $U/t$,
108: the (hole) filling $x$, the temperature $T$, and the hopping anisotropy
109: $t_x \ne t_y$, and for their dependence on the ``extrinsic'' parameters
110: system size and geometry, boundary conditions (BCs), and commensuration of
111: $x$ with system size. As a result of this investigation, we have chosen
112: two parameter sets representative of inhomogeneous spin and charge
113: configurations whose evolution we will follow on varying $t^{\prime}$ in
114: the extended Hubbard model. Hereafter we take the energy unit to be $t = 1$.
115:
116: \begin{figure}[t!]
117: \mbox{\psfig{figure=zctpo5018.eps,height=3.0cm,angle=270}}
118: \mbox{\hspace{-0.5cm}\psfig{figure=zctpo5026.eps,height=3.0cm,angle=270}}
119: \smallskip
120: {\centerline{(a) \qquad\qquad\qquad\qquad\qquad$\;\;$ (b)}}
121: \vspace{0.1cm}
122: \caption{Ground-state charge distributions for Hubbard model on a
123: $12\times12$ cluster with $U = 5$, $t^{\prime} = 0$, and open BCs. In (a)
124: $x = 1/8$ and $t_x = t_y = 1$, while in (b) $x = 1/6$, $t_x = 0.9$
125: and $t_y = 1.1$. In these figures the hole density is scaled by radius, and
126: the largest circles correspond to $\langle n_i \rangle = 0.644$, or 35.6\%
127: hole doping of the site. }
128: \end{figure}
129:
130: We focus on 12$\times$12 systems with open BCs, and take $U/t = 5$ to give
131: line-like charge structures. The cases we consider are: I -- isotropic
132: hopping, $t_x = t_y = 1$, with hole doping $x$ = 1/8, where the ground state
133: is a corral, or closed loop of diagonal, antiphase domain walls; II --
134: anisotropic hopping, $t_x = 0.9$, $t_y = 1.1$, with hole doping $x$ = 1/6,
135: where the ground state consists of uniform, filled, antiphase stripes.
136: The charge configurations with $t^{\prime} = 0$ are reproduced\cite{rnk}
137: in Figs.~1(a) and (b) respectively for cases I and II. The site- or
138: bond-centered nature of the charge structures is not important for studying
139: the influence of $t^{\prime}$. With regard to stability against small
140: changes in $t^{\prime}$, both corral and stripe solutions are found to have
141: only minor, quantitative alterations when $|t^{\prime}| \le 0.05$ for
142: both signs.
143:
144: \begin{figure}[t!]
145: \mbox{\psfig{figure=zctpo5108.eps,height=3.0cm,angle=270}}
146: \mbox{\hspace{-0.5cm}\psfig{figure=zctpo5126.eps,height=3.0cm,angle=270}}
147: \smallskip
148: {\centerline{(a) \qquad\qquad\qquad\qquad\qquad$\;\;$ (d)}}
149: \mbox{\psfig{figure=zctpo5208.eps,height=3.0cm,angle=270}}
150: \mbox{\hspace{-0.5cm}\psfig{figure=zctpo5226.eps,height=3.0cm,angle=270}}
151: \smallskip
152: {\centerline{(b) \qquad\qquad\qquad\qquad\qquad$\;\;$ (e)}}
153: \mbox{\psfig{figure=zctpo5308.eps,height=3.0cm,angle=270}}
154: \mbox{\hspace{-0.5cm}\psfig{figure=zctpo5326.eps,height=3.0cm,angle=270}}
155: {\centerline{(c) \qquad\qquad\qquad\qquad\qquad$\;\;$ (f)}}
156: \vspace{0.1cm}
157: \caption{Ground-state charge distributions for Hubbard model on a
158: $12\times12$ cluster with $U = 5$ and open BCs. In (a)-(c) $x = 1/8$ and
159: $t_x = t_y = 1$, while in (d)-(f) $x = 1/6$, $t_x = 0.9$ and $t_y = 1.1$.
160: $t^{\prime} = 0.1$ in (a) and (d), $t^{\prime} = 0.2$ in (b) and (e), and
161: $t^{\prime} = 0.3$ in (c) and (f). Charge scale as in Fig.~1. }
162: \end{figure}
163:
164: Fig.~2 illustrates the effects of increasing a next-neighbor hopping
165: $t^{\prime}$ with the same sign as $t$. This results in a change of shape
166: of the Fermi surface, which remains closed around the $\Gamma$ point but
167: expands in the directions $(k,\pm k)$ while contracting in the $(\pm k,0)$
168: and $(0,\pm k)$ directions. For the corral solution we see at
169: $t^{\prime} = 0.1$ [Fig.~2(a)] a trend towards breaking of the domain line
170: into small clusters, followed before $t^{\prime} = 0.2$ [Fig.~2(b)] by a
171: complete evaporation of the inhomogeneous charge structure. The solution
172: here and for higher $|t^{\prime}|$ is a charge-uniform, ``metallic'' phase.
173: For the stripe solution the behavior is similar, in that the stable stripe
174: phase is lost in the range $0.1 < t^{\prime} < 0.2$ in favor of the
175: uniform phase, which then persists to high values of $t^{\prime}$.
176: The spin configuration (not shown) in the uniform phase is found to be
177: commensurate antiferromagnetism, although with an ordered moment suppressed
178: in comparison to the antiferromagnetic (AF) regions of the corral and
179: stripe solutions at $t^{\prime} = 0$.
180:
181: \begin{figure}[t!]
182: \mbox{\psfig{figure=zctpo5118.eps,height=3.0cm,angle=270}}
183: \mbox{\hspace{-0.5cm}\psfig{figure=zctpo5136.eps,height=3.0cm,angle=270}}
184: \smallskip
185: {\centerline{(a) \qquad\qquad\qquad\qquad\qquad$\;\;$ (d)}}
186: \mbox{\psfig{figure=zctpo5218.eps,height=3.0cm,angle=270}}
187: \mbox{\hspace{-0.5cm}\psfig{figure=zctpo5236.eps,height=3.0cm,angle=270}}
188: \smallskip
189: {\centerline{(b) \qquad\qquad\qquad\qquad\qquad$\;\;$ (e)}}
190: \mbox{\psfig{figure=zctpo5308e.eps,height=3.0cm,angle=270}}
191: \mbox{\hspace{-0.5cm}\psfig{figure=zctpo5336.eps,height=3.0cm,angle=270}}
192: \smallskip
193: {\centerline{(c) \qquad\qquad\qquad\qquad\qquad$\;\;$ (f)}}
194: \vspace{0.1cm}
195: \caption{Ground-state charge distributions for Hubbard model on a
196: $12\times12$ cluster with $U = 5$ and open BCs. In (a)-(c) $x = 1/8$ and
197: $t_x = t_y = 1$, while in (d)-(f) $x = 1/6$, $t_x = 0.9$ and $t_y = 1.1$.
198: $t^{\prime} = - 0.1$ in (a) and (d), $t^{\prime} = - 0.2$ in (b) and (e),
199: and $t^{\prime} = - 0.3$ in (c) and (f). Charge scale as in Fig.~1. }
200: \end{figure}
201:
202: This result has a straightforward interpretation in terms of the metallic
203: state maximizing the hopping kinetic energy. Positive $t^{\prime}$ terms
204: cause direct frustration of local charge and spin alignment by competing
205: with the $t$ term, resulting in a uniform distribution. One may also
206: consider a band picture, where for this sign of $t^{\prime}$ the bandwidth
207: of the lower Hubbard band in the spin-density wave (SDW) state at
208: half-filling is increased, suggesting that doped hole-like carriers favor
209: delocalization.\cite{rvvg} Our result and interpretation are fully consistent
210: with DMRG studies of the $t$-$J$ model,\cite{rws} which for $t^{\prime} > 0$
211: show a systematic loss of stripe charge order. These also find an enhancement
212: of hole pairing, which is further consistent with our result because
213: superconductivity would be expected as the leading low-temperature
214: instability of the isotropic, metallic state. ED calculations\cite{rtgsclmd}
215: in this regime differ in that they report enhanced stripe order for $0 <
216: t^{\prime} < 0.2$, albeit for diagonal stripes at smaller (physical) values
217: of $J/t$. The authors interpret this result in terms of the explicit
218: next-neighbor hopping required to cancel the intra-sublattice hopping
219: generated implicitly by the $t$ term, and in accord with the present
220: findings expect that carrier localization and stripe-formation are
221: maximized when the sum of these two contributions to $t^{\prime}$ is
222: minimized. A DMRG study\cite{rtgsclmd} which tends to support this result
223: remains in minor contradiction to Ref.~\onlinecite{rws}, and the resolution
224: is presumably to be found in commensuration effects on the differing system
225: sizes used.
226:
227: Fig.~3 illustrates the effects of increasing $|t^{\prime}|$ in the physical
228: regime for the cuprates, where $t^{\prime}$ has the opposite sign to $t$.
229: This results in the change from a closed to an ``open'' Fermi surface around
230: the $\Gamma$ point. For the
231: corral solution we see at $t^{\prime} = - 0.1$ and $- 0.2$ [Figs.~3(a,b)]
232: a reinforcement of undisturbed AF order in the center of the system, and
233: retention of the diagonal domain walls, until a more complex behavior
234: sets in at $t^{\prime} = - 0.3$ [Fig.~3(c)]. For the stripe solution there
235: is a definite crossover from horizontal stripes at small $|t^{\prime}|$ to
236: diagonal stripes at $t^{\prime} = - 0.1$ and $- 0.2$ [Figs.~3(d,e)], again
237: preceding a less clear structure at $t^{\prime} = - 0.3$ [Fig.~3(f)]. For
238: both parameter sets, the high-$t^{\prime}$ configurations shown in Fig.~4
239: are rather complex in shape, only weakly inhomogeneous in charge structure
240: and weakly but commensurately AF in spin structure.
241:
242: For $t^{\prime} / t < 0$, the narrowing of the lower SDW Hubbard band
243: suggests reduced carrier mobility and thus an enhanced tendency toward
244: charge inhomogeneity. In a local hopping picture, all next-neighbor processes
245: cost kinetic energy, and so are suppressed\cite{rnk} by i) aligning spins
246: ferromagnetically on diagonal bonds, promoting AF order, and ii) aligning
247: holes on diagonal bonds, promoting diagonal domain walls. Both of these
248: tendencies are clear in Fig.~3. At higher $t^{\prime}$ this trend cannot be
249: sustained, as the kinetic energy cost of failure to delocalize is too great,
250: and for $- 0.4 < t^{\prime} < - 0.3$ the kinetic energy associated with
251: $t^{\prime}$ becomes negative despite the sign, as shown in Fig.~5. This is
252: possible by a change in the coefficients of the HF wave function such that on
253: average the sign of the diagonal overlap is negative. While such an alteration
254: is not readily visualized in terms of charge configurations, its consequences
255: are the weakly inhomogeneous structures in Fig.~4. The tendency to carrier
256: delocalization also for this sign of $t^{\prime}$ is in accord with the
257: results of Ref.~\onlinecite{rmobb} for the hole filling of an isolated stripe.
258:
259: \begin{figure}[t!]
260: \mbox{\psfig{figure=zctpo5618.eps,height=3.0cm,angle=270}}
261: \mbox{\hspace{-0.5cm}\psfig{figure=zctpo5636.eps,height=3.0cm,angle=270}}
262: \smallskip
263: {\centerline{(a) \qquad\qquad\qquad\qquad\qquad$\;\;$ (b)}}
264: \caption{Ground-state charge distributions for Hubbard model on a
265: $12\times12$ cluster with $U = 5$, $t^{\prime} = 0.6$, and open BCs. In
266: (a) $x = 1/8$ and $t_x = t_y = 1$, while in (b) $x = 1/6$, $t_x = 0.9$
267: and $t_y = 1.1$. Charge scale as in Fig.~1. }
268: \end{figure}
269:
270: In the DMRG calculations,\cite{rtgsclmd,rws} increasing $|t^{\prime}|$
271: up to 0.3 introduces a doubling of the charge periodicity. From the above
272: considerations we may offer a change from horizontal to diagonal stripes as
273: a possible consistent explanation of this result, given the small number
274: of holes involved. The observed
275: suppression of pairing\cite{rws} is certainly consistent with the expectation
276: of charge localization accompanying the formation of static, inhomogeneous
277: charge structures. In the ED results,\cite{rtgsclmd} both vertical and
278: diagonal stripe correlations are strongly suppressed by $t^{\prime}$,
279: whereas pairing correlations remain appreciable. Other RSHF studies of
280: the extended Hubbard model\cite{rvvg} confirm a significantly reduced
281: tendency to stripe formation in this parameter regime.
282:
283: All of the above results are for hole-doped systems. In the electron-doped
284: regime, we find by RSHF an exact reproduction
285: of the charge configurations in Figs.~1-4, with the important proviso that
286: the sign of $t^{\prime}$ be reversed. This symmetry emerges from the band
287: and local physics arguments invoked above, and can be seen in Fermi-surface
288: shapes, but most importantly can be proven rigorously from the invariance
289: of the Hamiltonian [Eq.~(1)] under particle-hole transformation combined
290: with sign reversal $c_i \rightarrow - c_i$ on one sublattice.
291:
292: In summary, next-neighbor hopping terms $t^{\prime}$ lead to a rapid
293: suppression of the $t^{\prime} = 0$ inhomogeneous charge structures in
294: the Hubbard model. For hole- (electron-) doped systems the primary influence
295: of positive (negative) $t^{\prime}$ (Fig.~2) on stripes lies in its
296: delocalization effect, while that of negative (positive) $t^{\prime}$
297: (Figs.~3,4) is destruction of the spin registry and antiphase domain wall
298: nature. This latter result suggests the importance of AF order in
299: stabilizing static horizontal stripes.\cite{rcn}
300:
301: \begin{figure}[t!]
302: \centerline{\psfig{figure=ztpke.eps,width=7.5cm,angle=270}}
303: \medskip
304: \caption{ Average kinetic energy $E_{\rm kin}$ per diagonal bond as a
305: function of next-neighbor hopping $t^{\prime}$ opposite in sign to $t$.}
306: \end{figure}
307:
308: In cuprates ($t^{\prime}/t < 0$), the effective $t^{\prime}$ is intrinsically
309: smaller for LSCO and related monolayer systems than for YBCO, BSCCO and the
310: related mono-, bi-, and trilayer systems. This result may be obtained by
311: simple fitting of tight-banding models to the Fermi surfaces measured by
312: ARPES\cite{rtm} and is also confirmed by reduction (downfolding) of band
313: structure calculations.\cite{raljp} In addition to the number of layers,
314: the apical oxygen atoms are seen to be an important factor contributing to
315: the difference. Because deviations of the Cu--O--Cu bond angle
316: from 180$^0$ reduce orbital overlap, the LTT and low-temperature orthorhombic
317: (LTO) distortions of the LSCO structure also act to reduce $|t^{\prime}|$ in
318: comparison with unbuckled cuprate layers. At the tight-binding level, when
319: $t$ and $t^{\prime}$ are the only parameters in the extended hopping band
320: structure, $t^{\prime}$ values of order $- 0.5 t$ are required to model the
321: open, YBCO-like Fermi surface, while the closed LSCO surface is reproduced
322: by values of $|t^{\prime}| < 0.2$.
323:
324: We have considered previously the possible key role of the LTT distortion,
325: which is the ground-state lattice structure only in Nd- and Eu-doped LSCO
326: compounds.\cite{rnk} Only the LTT distortion produces anisotropic $t$ terms,
327: and to the best of our knowledge there remain no reports of static stripes
328: in any non-LTT LSCO structures, or in any other cuprates. A natural question
329: then arises concerning YBCO, where the presence of ${\hat b}$-axis chains
330: leads to a planar structural
331: anisotropy. Our results offer an additional explanation for the absence of
332: static stripes in YBCO: the large effective $t^{\prime}$ values [Fig.~4]
333: suppress the formation of charge-inhomogeneous structures which are
334: candidate ground states at smaller $t^{\prime}$. Further, $t^{\prime}$ may
335: be expected to be an important factor contributing to the absence of static
336: stripes in the majority of hole-doped cuprate materials, where the Fermi
337: surfaces of the CuO$_2$ planes are open. For the electron-doped systems, we
338: have shown these to be in the regime where stripe formation is suppressed by
339: any value of $t^{\prime}$, which for these dopings favors a uniform,
340: metallic phase.
341:
342: We comment briefly that the LTO structure of the CuO$_2$ planes in LSCO
343: and BSCCO has an asymmetrical $t^{\prime}$. In RSHF, and for $t^{\prime}$
344: opposite in sign to $t$, a sufficiently large anisotropy $t_{x+y}^{\prime}
345: > t_{x-y}^{\prime}$ yields diagonal stripes in the direction ${\hat x}+{\hat
346: y}$. However, the required anisotropies are physically unjustified, and no
347: effect of this nature is expected in the real material.
348:
349: In conclusion, we have studied the effects of a next-neighbor hopping term
350: within the extended Hubbard model on the formation of static stripes in
351: cuprate systems. In a RSHF treatment, $t^{\prime}$ terms of both signs
352: cause a rapid suppression of horizontal stripe formation. When $t^{\prime}$
353: has the physical sign for cuprates (opposite to $t$), we find for hole-doped
354: systems a window in which diagonal stripes are favored, and then a broad
355: regime at large $t^{\prime}$ where the compromise charge configuration is
356: only weakly inhomogeneous. For the opposite sign of $t^{\prime}$, a
357: homogeneous, metallic phase is preferred. These properties are exactly
358: reversed in electron-doped systems.
359:
360: In contradiction to observations of superconductivity, the ``physical'' case
361: is dominated by charge localization and suppression of hole pairing, whereas
362: competition of inhomogeneous charge structures with the superconducting
363: instability is reduced for $t^{\prime} / t > 0$. However, these results are
364: consistent with previous studies of extended Hubbard and $t$-$J$ models. We
365: propose that the weak $t^{\prime}$ terms characteristic of the LSCO system
366: in comparison with other high-$T_c$ superconductors is an important
367: contributing factor to the appearance of static stripes only in these
368: materials, and that stripe formation in electron-doped cuprates is
369: suppressed for all $t^{\prime}$.
370:
371: \bigskip
372:
373: We are grateful to F. Guinea and C. Morais Smith for helpful discussions.
374: This work was supported by the Deutsche Forschungsgemeinschaft through SFB
375: 484.
376:
377: \begin{thebibliography}{99}
378:
379: \bibitem{rnk} B. Normand and A. P. Kampf, Phys. Rev. B {\bf 64}, 024521
380: (2001).
381:
382: \bibitem{rtm} See T. Tohyama and S. Maekawa, Supercond. Sci. Technol.
383: {\bf 13}, 17 (2000) and references therein.
384:
385: \bibitem{rdea} H. Ding {\it et al.}, Phys. Rev. Lett. {\bf 76}, 1533 (1996).
386:
387: \bibitem{riea} A. Ino {\it et al.}, J. Phys. Soc. Jpn. {\bf 68}, 1496 (1999).
388:
389: \bibitem{rka} D. M. King {\it et al.}, Phys. Rev. Lett. {\bf 70}, 3159
390: (1993); R. O. Anderson {\it et al.}, {\it ibid.} 3163 (1993).
391:
392: \bibitem{rtgsclmd} T. Tohyama {\it et al.}, Phys. Rev. B {\bf 59},
393: R11649 (1999).
394:
395: \bibitem{rws} S. R. White and D. J. Scalapino, Phys. Rev. B {\bf 60},
396: R753 (1999).
397:
398: \bibitem{rvvg} B. Valenzuela, M. Vozmediano, and F. Guinea, Phys. Rev. B
399: {\bf 62}, 11312 (2000).
400:
401: \bibitem{rim} M. Ichioka and K. Machida, J. Phys. Soc. Jpn. {\bf 68},
402: 4020 (1999).
403:
404: \bibitem{rmobb} I. Martin, G. Ortiz, A. V. Balatsky, and A. R. Bishop,
405: Int. J. of Mod. Phys. {\bf 14}, 3567 (2000).
406:
407: \bibitem{rmgxfd} G. B. Martins, C. Gazza, J. C. Xavier, A. Feiguin, and
408: E. Dagotto, Phys. Rev. Lett. {\bf 84}, 5844 (2000).
409:
410: \bibitem{raljp} O. K. Andersen, A. I. Liechtenstein, O. Jepsen, and
411: F. Paulsen, J. Phys. Chem. Solids {\bf 56}, 1573 (1995).
412:
413: \bibitem{rcn} See A. Castro Neto, Phys. Rev. B {\bf 64}, 104509 (2001)
414: and references therein.
415:
416: \end{thebibliography}
417:
418: \end{document}
419:
420: We also note that
421: for this sign of $t^{\prime}$ the van Hove singularity in the electronic
422: density of states (DOS) is moved to hole fillings in the physical parameter
423: regime: for $t^{\prime} = - 0.1$ it occurs at 9\% hole doping, and for
424: $t^{\prime} = - 0.2$ at 17\% [Fig.~3(e)]. However, because the singularity
425: in the DOS is rather weak, and because singular $(\pi,\pi)$ scattering is
426: effective primarily in destabilizing metallic phases, we do not observe
427: significant effects on the inhomogeneous charge structures favored by RSHF.
428: