1: %%
2: %% This is file `pss-stpl.tex',
3: %% generated with the docstrip utility.
4: %%
5: %% The original source files were:
6: %%
7: %% template.dtx (with options: `pss,small')
8: %%
9: %% $Id: template.dtx,v 1.60 2003/06/29 23:38:16 uwe Exp $
10: %% ====================================================================
11: \documentclass[pss,fleqn]{w-art}
12: \usepackage{times}
13: \usepackage{w-thm}
14: %% By default the equations are consecutively numbered. This may be changed by
15: %% the following command.
16: %% \numberwithin{equation}{section}
17: %%
18: %%x
19: %% The usage of multiple languages is possible.
20: %% \usepackage{ngerman}% or
21: %% \usepackage[english,ngerman]{babel}
22: %% \usepackage[english,french]{babel}
23: \usepackage[]{graphicx}
24: \begin{document}
25: %% The information for the title page will be placed between
26: %% \begin{document} and \maketitle. The order of most entries
27: %% is determined by the class file and can not be changed by
28: %% rearranging them. The maketitle command follows after the
29: %% abstract.
30: %%
31: %% Most of the following commands will be completed by the publisher.
32: %%
33: %% The copyrightyear is defined in the .clo file as the first argument
34: %% of the copyrightinfo command. If the copyrightyear differs from that
35: %% value it might be adjusted by the following definition:
36: %%
37: %% \renewcommand{\copyrightyear}{2003}% uncomment to change the copyrightyear.
38: %%
39: \DOIsuffix{theDOIsuffix}
40: %%
41: %% issueinfo for header and copyright line
42: \Volume{XX}
43: \Issue{1}
44: \Month{01}
45: \Year{2003}
46: %%
47: %% First and last pagenumber of the article. If the option
48: %% 'autolastpage' is set (default) the second argument may be left empty.
49: \pagespan{1}{}
50: %%
51: %% Dates will be filled in by the publisher. The 'reviseddate' and
52: %% 'dateposted' (Published online) entry may be left empty.
53: \Receiveddate{\sf }
54: \Reviseddate{\sf }
55: \Accepteddate{\sf }
56: \Dateposted{\sf }
57: %%
58: \keywords{stripe phase, Hubbard model, doped cuprates.}
59: \subjclass[pacs]{71.10.Fd, 71.27.+a, 74.72.-h}
60:
61: %% \pretitle{Editor's Choice}
62:
63: %% We have a short and a long form for the title. The short form
64: %% (optional argument) goes into the running head.
65:
66: \title[Half-filled stripes in the $t$-$t'$-$U$ Hubbard model]
67: {Half-filled stripes in the $t$-$t'$-$U$ Hubbard model}
68:
69: %% Please do not enter footnotes or \inst{}-notes into the optional
70: %% argument of the author command. The optional argument will go into
71: %% the header. If there is only one address the marker \inst{x} may be
72: %% omitted.
73:
74: \author[M. Raczkowski] {Marcin Raczkowski\footnote{Corresponding
75: author: e-mail: {\sf M.Raczkowski@if.uj.edu.pl},
76: Phone: +48\,12\,663\,5628,
77: Fax: +48\,12\,633\,4079}\inst{1,2}}
78: \address[\inst{1}] {Marian Smoluchowski Institute of Physics, Jagellonian
79: University, Reymonta 4, 30059 Krak\'ow, Poland}
80: \address[\inst{2}] {Laboratoire CRISMAT, UMR CNRS-ENSICAEN (ISMRA) 6508,
81: %6 Bld. du Mar\'echal Juin,
82: 14050 Caen, France}
83:
84: %%
85: %% Information for the second author
86: \author[A.~M. Ole\'s]{Andrzej M. Ole\'s\inst{1}}
87:
88: %%
89: %% Information for the third author
90: \author[R. Fr\'esard]{Raymond Fr\'esard\inst{2}}
91:
92: %%
93: %% \dedicatory{This is a dedicatory.}
94: \begin{abstract}
95: Using a self-consistent Hartree-Fock approximation we investigate
96: the relative stability of various stripe phases in the extended
97: $t$-$t'$-$U$ Hubbard model. One finds that a negative ratio of next- to
98: nearest-neighbor hopping $t'/t<0$ expells holes from antiferromagnetic
99: domains and reinforces the stripe order. Therefore the half-filled
100: stripes not only accommodate holes but also redistribute them so that
101: the kinetic energy is gained, and these stripes take over in the regime
102: of $t'/t\simeq -0.3$ appropriate for YBa$_2$Cu$_3$O$_{6+\delta}$.
103:
104: \end{abstract}
105: %% maketitle must follow the abstract.
106: \maketitle % Produces the title.
107:
108:
109: Two-dimensional (2D) systems with strong electron correlations received
110: a great deal of attention over the past twenty years. Such systems
111: exhibit a subtle competition between rich variety of phases:
112: antiferromagnetic (AF), ferromagnetic (FM), charge density wave (CDW),
113: spin density wave (SDW), and superconducting phase. In particular, strong
114: interplay of different order parameters in doped transition metal oxides
115: leads to the experimentally observed phase separation involving holes
116: ordered into domain walls (DWs) separated by AF domains \cite{Kiv03}.
117: On the theoretical side, stripe phases have been predicted on the basis
118: of Hartree-Fock (HF) calculations of doped AF clusters \cite{Zaa89}.
119: One important feature of these results is that they indicate formation of
120: the so-called {\it filled stripes\/} with density of one doped hole per
121: one atom in a DW. However, neutron experiments on the Nd-codoped
122: La$_{2-x}$Sr$_x$CuO$_4$ (Nd-LSCO) cuprates reveal that the observed
123: stripes are filled by one hole per two DWs, corresponding to
124: {\it half-filled stripes\/} \cite{Tra95Nd}. Unfortunately, the latter are
125: only locally stable within the HF approximation \cite{Zaa96}. Even though
126: some additional Hamiltonian terms like the nearest-neighbor Coulomb
127: repulsion $V$ may slightly enhance their stability, filled stripes remain
128: always better solutions \cite{Zaa96}. Nonetheless, we will show below
129: that under some circumstances, half-filled vertical site-centered (HVSC)
130: stripes may become energetically favored over the filled either diagonal
131: site-centered (DSC) or vertical site-centered (VSC) ones even in the HF
132: approximation.
133:
134: The starting point for the analysis of stripe structures is the effective 2D
135: single-band Hubbard model,
136: %
137: \begin{equation}
138: H=-\sum_{ij\sigma}t^{}_{ij}c^{\dag}_{i\sigma}c^{}_{j\sigma} +
139: U\sum_{i}n^{}_{i\uparrow}n^{}_{i\downarrow},
140: \label{eq:Hubb}
141: \end{equation}
142: %
143: where the hopping $t_{ij}$ is $t$ on the bonds connecting nearest-neighbor
144: sites and $t'$ for next-nearest-neighbor sites, while $U$ is the on-site
145: Coulomb interaction.
146: The model is solved self-consistently within the HF method on
147: a $16\times 16$ cluster with periodic boundary conditions. For a
148: representative doping level $x=1/8$ one obtains stable filled stripe
149: structures with AF domains of width seven atoms. Their stability can be
150: best understood from the band structure shown in Fig.~\ref{fig1}.
151: In general, each filled DW induces the formation of two unoccupied bands,
152: i.e., one for the $\uparrow$-spin and another one for the
153: $\downarrow$-spin, lying within the Mott-Hubbard gap, as illustrated in
154: Fig.~\ref{fig1}(a). Consequently, their special stability rests on a gap
155: that opens in the symmetry broken state between the highest occupied
156: state of the lower Hubbard band and the bottom of the mid-gap band. One
157: may ask what happens now if one decreases the doping level so as to get
158: half-filled stripes without a period quadrupling? Certainly, the twofold
159: degenerate mid-gap states become quarter filled, as there is only half a
160: hole per one DW. Hence, any stringent reason for such a symmetry breaking
161: with the Fermi level in the middle of the lower mid-gap band is absent.
162: Therefore, one needs to lower the symmetry by a period quadrupling so
163: that the period on the stripe becomes four times the lattice constant
164: \cite{Zaa96}. This might be accomplished by introducing a SDW modulation
165: along the DWs which leads to the band structure shown in Fig.~\ref{fig1}(b),
166: with a gap that opens up exactly at the Fermi energy. However, in spite of
167: better optimizing the potential energy $E_U$, the HVSC stripe phase
168: represents only a local minimum of energy, being less stable than both the
169: filled VSC and DSC ones. Thus, guided by the observation that finite
170: negative $t'$ expels holes from the AF domains and reinforces the stripe
171: order \cite{Rac03}, we investigate whether this mechanism suffices to
172: stabilize the former structure in the $t$-$t'$-$U$ model.
173:
174: %
175: \begin{figure}[t!]
176: \begin{center}
177: \includegraphics[width=0.7\textwidth]{fig1.eps}
178: \end{center}
179: \caption
180: {
181: Band structure as a function of momentum $k_y$, for
182: a unit cell of width $N_x=16$ in the Hubbard model with $U/t=5$ for:
183: (a) VSC stripe, doubled Brillouin zone;
184: (b) HVSC stripe, quadrupled Brillouin zone.
185: Black (gray) line corresponds to the bulk (mid-gap) bands,
186: respectively, whereas the dashed line indicates the Fermi level.
187: }
188: \label{fig1}
189: \end{figure}
190: %
191: %
192: \begin{table}[b!]
193: \caption {Energies per site: ground-state energy $E_{\rm tot}$, kinetic
194: energy contributions for the bonds along (10) $E_t^{x}$, (01) $E_t^{y}$,
195: (11) $E_{t'}^{x-y}$ and ($1\bar{1}$) $E_{t'}^{x+y}$ directions,
196: and potential energy $E_U$ of the HVSC, VSC and DSC stripe phases in the
197: $t$-$t'$-$U$ model with $U/t=5$ and $x=1/8$. VSC stripe is unstable at
198: $t'/t=-0.3$.}
199: \label{tab1}
200: \begin{tabular}{rrccrrcc}
201: \hline
202: \multicolumn{1}{c} {} &\multicolumn{1}{c} {$t'/t$} &
203: \multicolumn{1}{c} {$E_t^x/t$} &\multicolumn{1}{c} {$E_t^y/t$} &
204: \multicolumn{1}{c} {$E_{t'}^{x-y}/t$}&\multicolumn{1}{c} {$E_{t'}^{x+y}/t$} &
205: \multicolumn{1}{c} {$E_U/t$} &\multicolumn{1}{c} {$E_{\textrm{tot}}/t$}\\
206: \hline
207: {\bf HVSC} & 0.0 & $-$0.5846 & $-$0.6585 & 0.0000
208: & 0.0000 & 0.4627 & $-$0.7804 \\
209: {\bf VSC} & 0.0 & $-$0.6753 & $-$0.6147 & 0.0000
210: & 0.0000 & 0.4900 & $-$0.8000 \\
211: {\bf DSC} & 0.0 & $-$0.6368 & $-$0.6368 & 0.0000
212: & 0.0000 & 0.4696 & $-$0.8040 \\
213: \hline
214: {\bf DSC} & $-$0.3 & $-$0.6152 & $-$0.6152 & 0.0000
215: & 0.0257 & 0.4379 & $-$0.7668 \\
216: {\bf HVSC} & $-$0.3 & $-$0.5786 & $-$0.6198 & $-$0.0080
217: & $-$0.0080 & 0.4322 & $-$0.7678 \\
218: \hline
219: \end{tabular}
220: \end{table}
221: %
222:
223:
224: \begin{figure}[t!]
225: \begin{minipage}{.57\textwidth}
226: \includegraphics[width=.6\textwidth]{fig2.eps}
227: \caption
228: {
229: Rearrangement of the HVSC stripe order with increasing
230: next-nearest-neighbor hopping:
231: (a) $t'/t=0$;
232: (b) $|t'|/t=0.45$. Circle sizes are proportional to doped hole
233: density. Capital letters label bond kinetic energies along the
234: $(10)$ (A$_x,\dotsc,$D$_x$), $(01)$ (A$_y,\dotsc,$G$_y$),
235: $(11)$ (A,$\dotsc$,H), and $(1{\bar 1})$ (in parenthesis) directions,
236: listed in Table~\ref{tab2}, while $+$ ($-$) correspond to the bond
237: kinetic energy gain (loss) with respect to the $t'=0$ case.
238: Shadow circles indicate two crossing zigzag lines, see text
239: ($U/t$=5, $x=1/8$).
240: }
241: \label{fig2}
242: \end{minipage}
243: \hfil
244: \begin{minipage}{.35\textwidth}
245: \setfloattype{table}
246: \caption
247: {
248: Bond kinetic energies along the $(10)$ (A$_x,\dotsc,$D$_x$), $(01)$
249: (A$_y,\dotsc,$G$_y$), $(11)$ and $(1{\bar 1}$) (A,$\dotsc$,H)
250: directions around the HVSC stripe shown in Fig.~\ref{fig2}.
251: }
252: \label{tab2}
253: \begin{tabular}{crr}
254: \hline
255: \multicolumn{1}{c|}{}& \multicolumn{2}{c}{$t'/t$ } \\
256: \cline{2-3}
257: \multicolumn{1}{c|}{bond} & \multicolumn{1}{c}{0 } &
258: \multicolumn{1}{c}{$-0.45$} \\
259: \hline
260: A$_x$ & $-$0.6081 & $-$0.6141 \\
261: B$_x$ & $-$0.6652 & $-$0.5482 \\
262: C$_x$ & $-$0.6277 & $-$0.6257 \\
263: D$_x$ & $-$0.4377 & $-$0.4885 \\
264: \hline
265: A$_y$ & $-$0.6057 & $-$0.5822 \\
266: B$_y$ & $-$0.6003 & $-$0.5281 \\
267: C$_y$ & $-$0.8505 & $-$0.7234 \\
268: D$_y$ & $-$0.5759 & $-$0.5623 \\
269: E$_y$ & $-$0.6515 & $-$0.5299 \\
270: F$_y$ & $-$0.7206 & $-$0.6364 \\
271: G$_y$ & $-$0.6635 & $-$0.6046 \\
272: \hline
273: A & 0.0000 & $-$0.0053 \\
274: B & 0.0000 & 0.0262 \\
275: C & 0.0000 & $-$0.1300 \\
276: D & 0.0000 & 0.0128 \\
277: E & 0.0000 & 0.0033 \\
278: F & 0.0000 & $-$0.0367 \\
279: G & 0.0000 & 0.0164 \\
280: H & 0.0000 & 0.0020 \\
281: \hline
282: \end{tabular}
283: \end{minipage}
284: \end{figure}
285:
286: In order to establish the role of $t'$, we begin with energies of
287: different phases at $t'=0$ and $t'=-0.3t$ (Table \ref{tab1}). The main
288: effect is the energy gain of HVSC over DSC phase along (11) and
289: $(1\bar{1})$ bonds with increasing $|t'|$. To understand it better we
290: discuss Fig.~\ref{fig2} showing how the charge and spin configuration
291: around the HVSC stripe is altered in the presence of increasing $|t'|$.
292: In Fig. \ref{fig2}(a) we label all inequivalent bond kinetic energies
293: along the $(10)$ (A$_x,\dotsc,$D$_x$), $(01)$ (A$_y,\dotsc,$G$_y$),
294: $(11)$ (A,$\dotsc$,H), and $(1{\bar 1})$ (in parenthesis) directions.
295: The most striking feature of the HVSC stripe phase is that the main
296: kinetic energy gain is released not by transverse charge fluctuations,
297: which is the case for the filled stripes, but by on-wall hopping
298: processes on the bonds $C_y$ ($F_y$) connecting sites with the FM (AF)
299: coupling, respectively (see Table~\ref{tab2}). Another interesting
300: property is that the system tries to regain some of the kinetic energy
301: in the transverse direction by developing a CDW on nearest-neighbor
302: sites to the DW itself. Indeed, although the charge distribution along
303: the stripe is uniform, the on-wall SDW causes the CDW on both sides of
304: the DW. This clearly promotes the hopping between the AF sites (bonds
305: $B_x$) over the FM ones (bonds $D_x$). Furthermore, the CDW is strongly
306: influenced by $t'$, being first almost entirely quenched when
307: $t'\simeq -0.15$ and then it is gradually restored [\textit{cf}. Fig.
308: \ref{fig2}(b)].
309:
310: These trends are fully consistent with the dynamical mean field theory
311: studies of the effect of a single kink along the HVSC stripe \cite{Fle01}.
312: It has been established that the energy cost of forming a kink increases
313: slightly with increasing $|t'|$ up to $|t'|/t<0.1$, whereas further
314: increase leads to the opposite effect, and finally a wall with a kink
315: becomes favored at $|t'|/t\simeq 0.3$. Note, however, that although the
316: overall shape of the brought back CDW is the same as of the initial one,
317: the physical situation is fundamentally different -- in the $t'=0$ limit,
318: it is energetically advantageous for the system to equalize hole density
319: between sites connected by the bonds $B_x$ (AF coupling), whereas in
320: the large $|t'|$ limit, it tries instead to equalize hole density between
321: sites connected by the bonds $D_x$ (FM coupling). Accordingly, such a
322: charge redistribution results in two crossing zigzag lines,
323: which facilitate hole propagation. Indeed, each path consists of
324: a -$C$-$F$- bond pattern along the (11) direction and analogous sequence
325: in the $(1\bar{1})$ direction. Even though a moving hole gains kinetic
326: energy on both type of bonds, a more significant gain is achieved on the
327: FM bond $C$, being the driving force in the formation of zigzag pattern.
328: It is also supported by a systematically growing kinetic energy gain
329: with increasing $|t'|$ on the FM bond $D_x$ connecting the zigzag paths
330: (\textit{cf}. Table~\ref{tab2}).
331:
332: %
333: \begin{figure}[t!]
334: \includegraphics[width=0.9\textwidth]{fig3.eps}
335: \caption
336: {
337: Average kinetic energy $E_{t'}$ per diagonal bond as a function of
338: increasing next-nearest-neighbor hopping $|t'|$ for:
339: (a) half-filled vertical site-centered (HVSC) stripe phase;
340: (b) filled diagonal site-centered (DSC) stripe phase, as obtained for
341: $U/t=4$, 5, 6, 7 (solid, dashed, long-dashed, dot-dashed line).
342: Panel (c) shows the phase boundary between the DSC stripe phase and the
343: HVSC one as obtained for doping $x=1/8$.
344: }
345: \label{fig3}
346: \end{figure}
347:
348: The effect of increasing $|t'|$ on the kinetic energy per diagonal bond
349: $E_t'$ of the HVSC stripe phase illustrates Fig.~\ref{fig3}(a). One
350: observes that in the small $|t'|$ regime next-nearest-neighbor hopping
351: processes cost energy and therefore they are suppressed by quenching
352: the CDW order on sites next to the DW. However, towards larger $|t'|$,
353: the kinetic energy associated with next-nearest-neighbor hopping
354: becomes negative despite the negative sign of $t'$.
355: Consequently, the system develops a new CDW so as
356: to optimize the $E_{t'}$ gain. In contrast, $E_{t'}$ of the filled DSC
357: stripe phase remains positive even in the large $|t'|=0.4t$ regime, as
358: shown in Fig.~\ref{fig3}(b). Moreover, increasing Coulomb repulsion $U$
359: yields in this case only a small suppression of $E_{t'}$, whereas the
360: same increase of $U$ is reflected in much larger $E_{t'}$ gain of the
361: HVSC stripe phases. This, together with a better on-site energy $E_U$,
362: explains a broader region of stability of the HVSC stripe phases in the
363: strongly correlated regime, as illustrated in Fig.~\ref{fig3}(c).
364: We interprete their enhanced stability in the small $U$ regime
365: as following from the melting of both stripe phases.
366: Indeed, increased mobility of the holes released
367: from the DSC stripes results in a faster, as compared to the HVSC case,
368: enhancement of the $E_{t'}$ energy cost upon increasing $|t'|$.
369:
370: In conclusion, we have found that the next-nearest-neighbor hopping $t'$ plays
371: an important role in affecting the relative stability between filled
372: and half-filled stripe phases. However, the established value
373: $|t'|/t\simeq 0.3$ opening a window for the stability of the HVSC stripe
374: phase, is excessively larger than $t'/t\simeq -0.1$ of LSCO compounds
375: \cite{Pav01}. Interestingly, $t'/t\simeq -0.3$ corresponds to the value
376: appropriate for YBa$_2$Cu$_3$O$_{6+\delta}$ (YBCO) and indeed a
377: half-filled vertical stripe phase has been reported in this compound for
378: $\delta=0.35$ \cite{Moo02}. Therefore, our result calls for a further
379: experimental characterization of stripes in YBCO. Finally,
380: it may be expected that a better treatment of electron
381: correlations would stabilize further the half-filled stripe phases
382: \cite{Sei04} for realistic values of $t'$.
383:
384:
385: \begin{acknowledgement}
386: We thank P. Dai for valuable discussions. M. Raczkowski was supported
387: by a Marie Curie Fellowship of the European Community program under
388: number HPMT2000-141. This work was also supported by the Polish
389: Ministry of Scientific Research and Information Technology,
390: Project No. 1~P03B~068~26.
391: \end{acknowledgement}
392:
393:
394: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
395:
396: \begin{thebibliography}{10}
397:
398: \bibitem{Kiv03} S.~A. Kivelson {\it et al.},
399: Rev. Mod. Phys. {\bf 75}, 1201 (2003).
400:
401: \bibitem{Zaa89} J.~Zaanen and O.~Gunnarsson,
402: Phys. Rev. B {\bf 40}, 7391 (1989);
403: D. Poilblanc and T. M. Rice,
404: {\it ibid.} {\bf 39}, 9749 (1989).
405:
406: \bibitem{Tra95Nd} J.~M.~Tranquada, B.~J.~Sternlieb, J.~D.~Axe,
407: Y.~Nakamura, and S.~Uchida,
408: Nature {\bf 375}, 561 (1995).
409:
410: \bibitem{Zaa96} J. Zaanen and A.~M. Ole\'s,
411: Ann. Phys. (Leipzig) {\bf 5}, 224 (1996).
412:
413: \bibitem{Rac03} M. Raczkowski, B. Normand, and A.~M. Ole\'s,
414: Phys. Stat. Sol. (b) \textbf{236}, 376 (2003).
415:
416: \bibitem{Fle01} M.~Fleck, A.~I.~Lichtenstein, and A.~M.~Ole\'s,
417: Phys. Rev. B {\bf 64}, 134528 (2001).
418:
419: \bibitem{Pav01} E. Pavarini, I. Dasgupta, T. Saha-Dasgupta, O. Jepsen,
420: and O. K. Andersen,
421: Phys. Rev. Lett. {\bf 87}, 047003 (2001).
422:
423: \bibitem{Moo02} H.~A. Mook, P. Dai, and F. Do\v{g}an,
424: Phys. Rev. Lett. {\bf 88}, 097004 (2002).
425:
426: \bibitem{Sei04} G. Seibold and J. Lorenzana,
427: Phys. Rev. B {\bf 69}, 134513 (2004).
428:
429: \end{thebibliography}
430:
431: \end{document}
432: \endinput
433: