Stripe phases — possible ground state of the high-Tc superconductors

Based on the mean-field method applied either to the extended single-band Hubbard model or to the single-band Peierls-Hubbard Hamiltonian we study the stability of both site-centered and bond-centered charge domain walls. The difference in energy between these phases is found to be small. Therefo...

Повний опис

Збережено в:
Бібліографічні деталі
Опубліковано в: :Физика низких температур
Дата:2006
Автори: Raczkowski, M., Oles, A.M., Fresard, R.
Формат: Стаття
Мова:Англійська
Опубліковано: Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України 2006
Теми:
Онлайн доступ:https://nasplib.isofts.kiev.ua/handle/123456789/120185
Теги: Додати тег
Немає тегів, Будьте першим, хто поставить тег для цього запису!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Цитувати:Stripe phases — possible ground state of the high-Tc superconductors / M. Raczkowski, A.M. Oles, R. Fresard // Физика низких температур. — 2006. — Т. 32, № 4-5. — С. 411-429. — Бібліогр.: 96 назв. — англ.

Репозитарії

Digital Library of Periodicals of National Academy of Sciences of Ukraine
_version_ 1859745053932519424
author Raczkowski, M.
Oles, A.M.
Fresard, R.
author_facet Raczkowski, M.
Oles, A.M.
Fresard, R.
citation_txt Stripe phases — possible ground state of the high-Tc superconductors / M. Raczkowski, A.M. Oles, R. Fresard // Физика низких температур. — 2006. — Т. 32, № 4-5. — С. 411-429. — Бібліогр.: 96 назв. — англ.
collection DSpace DC
container_title Физика низких температур
description Based on the mean-field method applied either to the extended single-band Hubbard model or to the single-band Peierls-Hubbard Hamiltonian we study the stability of both site-centered and bond-centered charge domain walls. The difference in energy between these phases is found to be small. Therefore, moderate perturbations to the pure Hubbard model, such as next nearest neighbor hopping, lattice anisotropy, or coupling to the lattice, induce phase transitions, shown in the corresponding phase diagrams. In addition, we determine for stable phases charge and magnetization densities, double occupancy, kinetic and magnetic energies, and investigate the role of a finite electron-lattice coupling. We also review experimental signatures of stripes in the superconducting copper oxides.
first_indexed 2025-12-01T20:34:54Z
format Article
fulltext Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5, p. 411–429 Stripe phases — possible ground state of the high-Tc superconductors Marcin Raczkowski1,2, Andrzej M. Oles� 1, and Raymond Fresard� 2 1 Marian Smoluchowski Institute of Physics, Jagellonian University Reymonta 4, PL-30059 Krak�w, Poland E-mail: A.M.Oles@fkf.mpg.de 2 Laboratoire CRISMAT, UMR CNRS-ENSICAEN(ISMRA) 6508, 6 Bld. du Mar�chal Juin Bld., F-14050 Caen, France Received November 1, 2005 Based on the mean-field method applied either to the extended single-band Hubbard model or to the single-band Peierls-Hubbard Hamiltonian we study the stability of both site-centered and bond-centered charge domain walls. The difference in energy between these phases is found to be small. Therefore, moderate perturbations to the pure Hubbard model, such as next nearest neigh- bor hopping, lattice anisotropy, or coupling to the lattice, induce phase transitions, shown in the corresponding phase diagrams. In addition, we determine for stable phases charge and magnetiza- tion densities, double occupancy, kinetic and magnetic energies, and investigate the role of a finite electron-lattice coupling. We also review experimental signatures of stripes in the superconduct- ing copper oxides. Pacs: 71.10.Fd, 71.27.+a, 74.25.–q, 74.72.–h Keywords: high-Tc superconductivity, Hubbard model, stripe phases. 1. Introduction Since the discovery of high-temperature supercon- ductivity by Bednorz and M�ller [1], the unusual physical properties of the copper oxides have stimu- lated theorists and have led to the appearance of many new ideas [2]. One of the especially appealing new pictures that has emerged is the instability towards a novel type of coexisting incommensurate (IC) charge and magnetic order, i.e., stripe phase. As a rare event in the theory of high temperature superconductivity, the theory preceeded here the experiment and the exis- tence of stripe phases was predicted on the basis of Hartree–Fock (HF) calculations in the two-band model for CuO2 planes of layered La2�xSr xCuO4 (LSCO) [3], before their experimental confirmation. This instability persists as well in the effective sin- gle-band Hubbard model [4–7]. All these calculations yielded solutions with a phase separation manifested in formation of nonmagnetic lines of holes, one-dimen- sional (1D) domain walls or stripes, which separate antiferromagnetic (AF) domains of opposite phases. Such states result from the competition between the superexchange interaction, which stabilize the AF long-range order in the parent Mott insulator, and the kinetic energy of doped holes. Indeed, the magnetic energy is gained when electrons occupy the neighbor- ing sites and their spins order as in the N�el state, whereas the kinetic energy is gained when the holes can move and the AF order is locally suppressed along a domain wall (DW). Thus, a stripe phase provides the best compromise between the superexchange pro- moting the AF order and the kinetic energy of doped holes. However, the debate on the microscopic origin of the stripe instability is far from closed. Two main sce- narios, based on a Ginzburg–Landau free energy, for the driving mechanism of the stripe phase have been discussed [8,9]. In the first one, stripes are charge-density waves with large periodicity arising from the Fermi surface (FS) instability with the tran- sition being spin driven [3]. A general feature of such © Marcin Raczkowski, Andrzej M. Oles� , and Raymond Fr�sard, 2006 an instability is a gap/pseudogap which opens up pre- cisely on the FS. Hence, the spacing between DWs is equal to 1/x, with x denoting doping level so as to maintain a gap/pseudogap on the FS. In this scenario spin and charge order occur at the same temperature or charge stripe order sets in only after spin order has de- veloped. An alternative scenario comes from the Coulomb- frustrated phase separation suggesting that stripe for- mation is charge driven. Indeed, using the Ising model, it has been shown that the competition be- tween long range Coulomb interactions and short range attraction between holes leads to formation of stripes [10]. In this case Ginzburg–Landau consider- ations lead to an onset of charge order prior to spin or- der as the temperature is lowered. However, the above analysis does not take into account spin fluctuations which might be crucial for the nature of the phase transition by precluding the spins from ordering at the charge-order temperature [11]. Moreover, the conjec- ture that long range Coulomb forces are required to stabilize stripe phases has been challenged by the studies of the t–J model, in which the DW structures were obtained without such interactions [12]. In order to investigate the influence of strong elec- tron correlations due to large on-site Coulomb repul- sion U at Cu ions, several methods have been em- ployed to study the stripe phases which go beyond the HF approximation, such as: density matrix re- normalization group (DMRG) [12,13], Slave-Boson approximation (SBA) [14–16], variational local an- satz approximation [17], Exact Diagonalization (ED) of finite clusters [18], analytical approach based on variational trial wave function within the string pic- ture [19], dynamical mean field theory (DMFT) [20,21], Cluster perturbation theory (CPT) [22], and quantum Monte Carlo (QMC) [23,24]. In spite of this huge effort, it remains unclear whether DWs are cen- tered on rows of metal atoms, hereafter named site- centered (SC) stripes, or if they are centered on rows of oxygen atoms bridging the two neighboring metal sites, the so-called bond-centered (BC) stripes, and even calculations performed on larger clusters did not yield a definite answer [25]. Therefore, the purpose of this paper is to study the stability of both structures based on the mean-field method applied either to the extended single-band Hubbard model or the sin- gle-band Peierls-Hubbard Hamiltonian which in- cludes the so-called static phonons [26]. For stable phases we determine charge and magnetization densi- ties, double occupancy, kinetic and magnetic energies, and investigate the role of a finite electron-lattice cou- pling. 2. Experimental signatures of stripes Experimentally, stripe phases are most clearly de- tected in insulating compounds with a static stripe or- der, but there is growing evidence of fluctuating stripe correlations in metallic and superconducting materi- als. The most direct evidence for stripe phases in doped antiferromagnets has come from neutron scat- tering studies in which charge and spin modulations are identified by the appearance of some IC Bragg peaks, in addition to those which correspond to the crystal structure. However, sometimes sufficiently large crystals are not available for such experiments, and one has to resort to other methods capable of probing local order. These methods include nuclear magnetic resonance (NMR), nuclear quadruple reso- nance (NQR), muon spin rotation (�SR), scanning tunneling microscopy (STM), and transmission elec- tron microscopy (TEM). Furthermore, angle-resolved photoemission spectroscopy (ARPES), angle-inte- grated photoemission spectroscopy (AIPES), as well as x-ray photoemission (XPS) and ultraviolet photoemission (UPS) spectroscopies all provide essen- tial information about conspicuous changes in the electronic structure when stripe structure sets in. Finally, a distinct imprint of the 1D spin-charge mod- ulation on transport properties should be detectable as the in-plane anisotropy of the resistivity and the Hall coefficient RH . The abundance of the current evidence on various types of stripe order as well as the recent ARPES re- sults on the spectral weight of the cuprate supercon- ductors is contained in the review articles by Kivelson et al. [27], and by Damascelli et al. [28]. Historically, the first compelling evidence for both magnetic and charge order in the cuprates was accomplished in a neo- dymium codoped compound La2� �x yNd ySr xCuO4 (Nd-LSCO). For y � 0 4. and x � 012. , Tranquada et al. [29,30] found that the magnetic scattering is not char- acterized by the two-dimensional (2D) AF wave vec- tor (1 2 1 2/ /, ), but by IC peaks at the wave vectors ( , )1 2 1 2/ /� � with � � 0118. . Moreover, inspired by the pioneering works demonstrating that the stag- gered magnetization undergoes a phase shift of � at the charge DWs [3–7], the authors found additional charge order peaks ( , )�2 0� , precisely at the expected position 2 0 236� � . . Interestingly, this doping corre- sponds to a local minimum in the doping dependence of the superconducting temperature Tc in Nd-LSCO [31], suggesting that the static stripes are responsible for this anomalous depression of superconductivity. However, it may well be that the apparent correlation is entirely accidental and therefore the role of stripes in superconductivity remains an open question [2]. 412 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 Marcin Raczkowski et al. Unfortunately, in early studies Tranquada et al. [32] detected only magnetic IC peaks at higher doping levels x � 015. and x � 0 2. . Nevertheless, systematic NQR studies of Nd-LSCO revealed the presence of ro- bust charge stripe order throughout the entire super- conducting regime of doping 0 07 0 25. .� �x [33]. Also in a more recent study, both charge and spin superlattice peaks at x � 015. were found recently in the neutron diffraction experiments by Wakimoto et al. [34]. In fact, the reason why static stripes could be de- tected in this compound is a structural transition from the low temperature orthorhombic (LTO) to the low temperature tetragonal (LTT) phase, induced by the substitution for La ions by isovalent Nd ions. This, in turn, provides a pinning potential for dynamic stripes and stabilizes the charge order. Evidence of a similar pinning potential has also been found both in the �SR and NQR studies of La2� �x yEu ySr xCuO4 (Eu-LSCO) with y � 0 2. [35,36]. Moreover, the con- nection between the LTT phase and the appearance of charge and spin stripe order has been clearly demonstrated both in the neutron scattering and x-ray diffraction studies on La2� �x yBa ySr xCuO4 (Ba-LSCO) with y /� 1 8 [37,38]. Finally, static IC charge ( , )2 2 0� � and magnetic ( , )1 2 1 2/ /� � peaks have been detected within the LTT phase of La2�xBa xCuO4 (LBCO) with x /� 1 8 [39]. The po- sition of the peaks and the established incommensurability � � 0118. are exactly the same as those obtained by Tranquada et al. [30] for Nd-LSCO. Notably, the peaks that correspond to charge order ap- pear always at somewhat higher temperature than the magnetic ones, indicating that the stripe order is driven by the charge instability. Let us now discuss the experimental evidence of slowly fluctuating stripes in La2�xSr xCuO4. The main difference between the Ba and Sr codoped sys- tem is the fact that the latter undergoes a structural phase transition from the high-temperature tetragonal (HTT) phase to the LTO phase. As a consequence, in the superconducting regime x � 0 06. , the LSCO sys- tem exhibits purely dynamic magnetic correlations which give rise to IC peaks at the wave vector ( , )1 2 1 2/ /� � specified in tetragonal lattice units 2�/atetra . In their seminal inelastic neutron scattering studies, Yamada et al. [40] established a remarkably simple relation � � x for 0 06 012. .� �x , followed by a lock-in effect at � � 1 8/ for larger x. In contrast, in the insulating spin-glass regime of LSCO x � 0 06. , quasielastic neutron scattering experi- ments with the main weight at zero frequency demon- strate that IC magnetic peaks are located at the wave vectors ( , )1 2 2 1 2 2/ / / /� �� � [41–43]. This phe- nomenon has often been interpreted as the existence of static diagonal stripes, even though no signatures of a charge modulation were observed. Another possible explanation is the formation of a short ranged spiral order as its chirality also breaks the translational sym- metry of the square lattice by a clockwise or anti- clockwise twist [44]. Remarkably, even though the spin modulation changes from a diagonal to verti- cal/horizontal one, i.e., along Cu–O bonds, at x around 0.06, � follows the doping x reasonably well over the entire range 0 03 012. .� �x , as shown in Fig. 1. In fact, just for x � 0 06. , both diagonal ( . )� � 0 053 and vertical/horizontal (� � 0 049. ) IC spin modulations have been found to coexist [43]. In a stripe model this corresponds to a constant density of 0.5 (0.7) holes per Cu atom in the DWs in the verti- cal/horizontal (diagonal) stripe phases, respectively, because of the difference in Cu spacings in the two ge- ometries, i.e., a aortho tetra� 2 . In contrast, in the narrow region 0 02 0 024. .� �x , IC magnetic peaks are located at the wave vector ( , )1 2 2 1 2 2/ / / /� �� � with � � x corresponding to a constant charge of one hole/Cu ion along a diagonal DW [45–47]. However, below x � 0 02. , this does not hold anymore and the incommensurability gets locked with the value � � 0 014. . Stripe phases-possible ground state of the high-Tc superconductors Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 413 0.12 0.08 0.04 0 0.05 0.10 0.15 0.20 0.25 x Nd-LSCO LSCO LCO Zn-LSCO YBCO ktetra ktetra htetra htetra hortho hortho 2 2 1/2 1/2 1/2 1/2 Fig. 1. Summary of experimental data illustrating the doping dependence of incommensurability � in the cuprates. Results have been obtained by different groups: Nd-LSCO (Refs. 29–32); LSCO (Refs. 40–43,45–47); LCO (Ref. 51); Zn-LSCO (Refs. 52,53); YBCO (Refs. 54,55). In LSCO, � has been defined as a distance from the IC peak position to the AF wave vector (1 2 1 2/ /, ) ei- ther in the orthorhombic (x � 006. ) or tetragonal (x � 006. ) notation (see insets), whereas at x � 006. , both definitions are used due to the coexistence of diagonal and parallel to the Cu–O bonds spin modulations. Unfortunately, any concomitant charge ordering has not yet been detected in LSCO. Nevertheless, by comparing the data based on the wipeout effect of 63Cu NQR charge order parameter in LSCO with the ones obtained from charge stripe compounds as (Nd,Eu,Ba)-LSCO, Hunt et al. [48] concluded that a similar stripe instability exists in LSCO over the whole underdoped superconducting region 1 16 1 8/ x /� � . It is also worth mentioning that a very compiling evidence for its existence has been es- tablished in the measurements of the in-plane resistiv- ity and the dynamical infrared conductivity aniso- tropy [49,50]. Experimental detection of IC magnetic peaks in the LTO phase of LSCO suggests that the LTT structure is not essential for the appearance of stripes. This con- jecture has been confirmed in experiments on the oxy- gen doped La2CuO4� � (LCO) with the orthorhombic crystal structure [51]. It is also supported by the evidence for static IC magnetic peaks in another orthorhombic compound La2�xSr xCu1�yZn yO4 (Zn-LSCO) with y up to 0 03. , even though attempts to observe the charge order peaks were unsuccessful [52,53]. In fact, Zn substitution pins the stripe fluctu- ations similarly to the rare-earth elements. However, in contrast to the latter, it does not induce a structural transition to the LTT phase, but provides randomly distributed pinning centers that promote meandering of stripes and correspondingly broadens IC peaks. An important question is whether charge stripes ap- pear solely in monolayered lanthanum compounds or if they are a generic feature of all the cuprates. The latter conjecture seems to be supported by in- elastic neutron scattering experiments on bilayered YBa2Cu3O6� � (YBCO) compounds that have identi- fied the presence of IC spin fluctuations throughout its entire superconducting regime [54]. In fact, as the doped charge is nontrivially distributed between the CuO2 planes and CuO chains, it is very difficult to determine the precise doping level x in the CuO2 sheet of YBCO. Nevertheless, systematic studies by Dai et al. [54] have shown that the incommensur- ability in YBCO increases initially with doping but it saturates faster than in LSCO, i.e., already at x � 01. with the value � � 01. . Unfortunately, there is no any compelling explanation that would account for such a different behavior of � in both systems. Eventually, charge order peaks have been observed in YBCO635. but in spite of several attempts, no static charge order could be detected in YBCO65. and YBCO66. so far [55]. Furthermore, although some neutron scattering ex- periments have been performed on Bi2Sr2CaCu2O8� � (BSCCO) sample, the sample has only produced weak evidence of the IC structure [56]. In contrast, Fourier transform of the recent STM data has revealed some IC peaks corresponding to a four-period modulation of the local density of states along the Cu–O bond direc- tion, which may imply the existence of stripes [57]. Nevertheless, definite answer pertinent to the appear- ance of stripes in all the cuprates remains still unset- tled and further experiments are required to reach an unambiguous conclusion, even though the summary of the experimental data illustrating the doping depen- dence of the incommensurability � in cuprates, de- picted in Fig. 1, includes an array of compounds. Tendency towards phase separation is also a start- ing point to understand the doping evolution of the electronic structure in LSCO and Nd-LSCO. For ex- ample, ARPES spectra measured at the X � ( , )� 0 point in LSCO show that even though the data are solely characterized by a single high binding energy feature in the insulating regime, upon increasing dop- ing one observes a systematic transfer of spectral weight from the high- to the low binding energy part [58]. Consequently, a well-defined quasiparticle (QP) peak develops near the optimal doping. In con- trast, the intensity near the S / /� ( , )� �2 2 point re- mains suppressed for the entire underdoped regime so that a QP peak is observed only for x � 015. . Another peculiar feature of the ARPES band dis- persion is extensively discussed in the literature saddle point at the X point, the so-called flat band [59]. As hole doping increases, the flat band moves mono- tonically upwards and crosses the Fermi level EF at x � 0 2. . This is reflected in the enhancement of the DOS at the chemical potential N( )� observed by AIPES [60]. The experimental distribution of the photoemission spectral weight near the X and S points in doped LSCO has been nicely reproduced using the DMFT approach for vertical SC stripes obtained within the Hubbard model [20]. As a consequence of the stripe order, the obtained spectra along the �–X–M path were not equivalent to those along the �–Y–M one, with � � ( , )0 0 and Y � ( , )0 � . Moreover, as in the ex- periment, the spectral weight along the �–X direction was suppressed close to the � point and simulta- neously enhanced at the X point. Furthermore, in the framework of stripes, the flat QP band near the X point with a large intensity at the maximum below the chemical potential � follows from a superposition of the dispersionless 1D metallic band along the x direc- tion, formed by holes propagating along the vertical domain walls, and an insulating band that stems from the AF domains. In contrast, an AF band at the Y point is characterized by a high binding energy well be- low � and consequently the spectral weight at �� al- 414 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 Marcin Raczkowski et al. most vanishes. Finally, a distinct gap for charge exci- tations should open at � near the S point. This gap follows indeed from the stripe structure — while the system may be metallic along the stripes, i.e., in the antinodal directions �–X or �–Y, the low-energy exci- tations should be noticeably suppressed along the nodal direction �–S crossing all the stripes. This con- jecture is also supported either by the ED studies [18] or by the analytical approach based on variational trial wave function within the string picture [19], both applied to the t–t –t –J model, or by the CPT for the t–J model [22]. In fact, the low-energy spectral weight of Nd-LSCO at x � 012. , a model compound for which the evidence of spin and charge stripe order is the strongest, is also mostly concentrated in flat regions along the �–X and �–Y directions, while there is only little spectral weight along the �–S direction [61]. On the other hand, ARPES spectra of both LSCO and Nd-LSCO at x � 015. have revealed not only the pres- ence of flat bands around the X andY points, but also the existence of appreciable spectral weight at EF in the nodal region [62]. While the observation of flat segments might be directly ascribed to 1D domain walls [63], detection of nodal spectral weight poses a formidable task to develop a theory that would de- scribe the electronic structure resembling the FS of a fully 2D system because, as it was already stressed out, the nodal spectral weight is expected to be sup- pressed in a static SC stripe picture [18–20,22]. In- deed, the experimentally established FS looks rather like the one arising from disorder or from dynamically fluctuating stripes [63]. Alternatively, guided by the CPT results showing that while the SC stripes yield little spectral weight near the nodal region, the BC ones reproduce quite well the nodal segments [22], Zhou et al. [62] have conjectured that the experimental FS may result from the coexistence of the SC and BC stripes. Within this framework, upon increasing doping the BC stripes are formed at the expense of the SC ones. This scenario is particularly interesting because it has been shown that the BC stripe, in contrast to its SC counterpart, en- hances superconducting pairing correlations [64]. The relevance of a bond order at the doping level x � 015. is supported by recent studies of the ARPES spectra in a system with the BC stripes [65]. These studies have yielded pronounced spectral weight both in the nodal and antinodal directions, reproducing quite well the experimental results in Nd-LSCO and LSCO [62]. Furthermore, the stripe scenario would also explain the origin of the already discussed two components seen in the ARPES spectra at the X point near x � 0 05. [58]. Indeed, the response from the AF insulating re- gions would be pushed to the high binding energies due to the Mott gap, whereas the charge stripes would be responsible for the other component near EF . Existence of DWs should also give rise to the ap- pearance of new states inside the charge-transfer gap that would suppress the shift of the chemical potential � in the underdoped regime x /� 1 8 where � increases linearly. Such pinning of � in LSCO was indeed de- duced from XPS experiments [66]. In contrast, in the overdoped region with a lock-in effect of �, the num- ber of stripes per unit cell saturates, doped holes pene- trate into the AF domains, and consequently � would move fast with doping in agreement with the experi- mental data. The picture of broadened stripes and holes spreading out all over the AF domains above x /� 1 8 is also indicated by the doping dependence of the resistivity and the Hall coefficient RH in Nd-LSCO. Namely, a rapid decrease in the magnitude of RH for doping level x /� 1 8 at low temperature provides evidence for the 1D charge transport, whereas for x /� 1 8, relatively large RH suggests a crossover from the 1D to 2D charge transport [67]. Al- together, it appears that the metallic stripe picture does capture the essence of the low-lying physics for Nd-LSCO and LSCO systems. Conversely, it is important to note that so far no evidence of IC peaks has been detected in any electron-doped cuprates superconductors. Instead, the neutron scattering experiments have estab- lished only commensurate spin fluctuations as in Nd2�xCe xCuO4 (NCCO), both in the superconduct- ing and in normal state [68]. Moreover, observation of such peaks is consistent with the XPS measurements in NCCO showing that the chemical potential in- creases monotonously with electron doping [69]. 3. Numerical results In this Section we attempt a systematic investiga- tion of the properties and relative stability of filled vertical and diagonal stripes. We shall see that in spite of the difficulty to stabilize the ground state with half-filled stripes (one hole per every two atoms in a DW), the mean-field framework is useful as pro- viding a generic microscopic description of filled inhomogeneous reference structures with the filling of one doped hole per stripe unit cell. Their special sta- bility rests on a gap that opens in the symmetry bro- ken state between the highest occupied state of the lower Hubbard band and the bottom of the so-called mid-gap bands, i.e., some additional unoccupied bands lying within the Mott-Hubbard gap that are formed due to holes propagating along DWs [26]. Here, we extend early HF studies of the filled DWs [4–7] and determine a phase diagram of the Hubbard Stripe phases-possible ground state of the high-Tc superconductors Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 415 model with an anisotropic nearest-neighbor hopping t by varying the on-site Coulomb repulsionU and inves- tigating locally stable structures for representative hole doping levels x /� 1 8 and x /� 1 6. We also re- port the changes in stability of the stripe structures in the extended Hubbard model due to the next-neighbor hopping t and to the nearest neighbor Coulomb inter- action V. Finally, in order to gain a comprehensive understanding of the competition between different types of stripes in a realistic model, we include lattice degrees of freedom induced by a static Peierls elec- tron-lattice coupling. 3.1. Extended single-band Hubbard model The starting point for the analysis of stripe struc- tures is the extended single-band Hubbard model, which is widely accepted as the generic model for a microscopic description of the cuprate superconduc- tors [70], H t c c U n n V n nij ij i j i i i i j ij � � �� � �� � � � † , (1) where the operator ci † ( )cj creates (annihilates) an electron with spin � on lattice site i (j), and n c c c ci i i i i� � � � � � † † stands for the electron density. The hopping tij is t on the bonds connecting nearest neighbors sites � �i j, and t for second-neighbor sites, while the on-site and nearest neighbor Coulomb inter- actions are, respectively,U and V. The model can be solved self-consistently in real space within the HF, where the interactions are de- coupled into products of one-particle terms becoming effective mean fields that act on each electron with the same strength. This approximation basically involves solving an eigenvalue problem. The obtained wave- functions form a new potential and hence the Hamil- tonian for a new eigenvalue problem. Typically, the new potential is chosen as some linear combination of the current and preceding potential. The iterations are continued until the input and output charge density and energy do not change within some prescribed ac- curacy. The most significant drawback of this method is that it neglects correlations. Electron correlation changes the system properties and manifests itself in the decrease of the ground state energy. The difference between the energy of the exact ground state and the energy obtained within the HF is thus called the correlation energy. It arises from the fact that an electron’s movement is correlated with the electrons around it, and accounting for this effect lowers further the energy, beyond the independent electron approxi- mation. We do not consider noncollinear spin configura- tions, and use the most straightforward version of the HF with a product of two separate Slater determi- nants for up and down spins, whence, n n n n n n n ni i i i i i i i� � � � � � � �� � � � � � �� �� . (2) A similar decoupling is performed for the nearest neighbor Coulomb interaction. Calculations were performed on 12 12� (16 16� ) clusters for x /� 1 6 ( )x /� 1 8 with periodic boundary conditions, and we obtain stable stripe structures with AF domains of width five atoms for x /� 1 6 and seven atoms for x /� 1 8. Typical solutions at x /� 1 8 are shown in Fig. 2 with the local hole density, � � � � � �� �n n nhi i i1 , (3) scaled by the diameter of the black circles and the length of the arrows being proportional to the ampli- tude of local magnetization density, � � � � �� �S n ni z i i 1 2 | | . (4) These structures possess nonmagnetic DWs with enhanced hole density which separate AF domains having hole density almost unchanged with respect to the undoped case. Note that the AF sites on each side of the DWs have a phase shift of �. In order to appreciate better the microscopic rea- sons of such arrangement let us consider a small clus- ter consisting of three atoms filled by two electrons and one hole (with respect to half-filling with the 416 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 Marcin Raczkowski et al. lx ly Fig. 2. Vertical site-centered (VSC) and diagonal site-cen- tered (DSC) stripe phases as found for U/t � 5 at hole doping x /� 1 8. The length of arrows is proportional to the magnetization � �Si z and the hole density � �nhi is scaled by the diameter of black circles. electron density n � 1 per site). For simplicity we as- sume that the electrons are confined to the considered cluster owing to large Coulomb interaction U t�� , and we do not take into account any interactions with the AF background. There are two possible candidates for the ground state. The first one corresponds to a hole added to three atoms of a single AF domain in which, if we suppose that a �-spin electron is replaced by a hole, the two remaining �-spin electrons can be found in one of three allowed configurations: { , , }� �0 , { , , }� � 0 , and { , , }0� � (the other configurations are ex- cluded by the Pauli principle). Hence, this polaronic state gives the total energy, E tP � 2 , (5) and the Coulomb interactionU does not contribute. A different situation is obtained when a hole oc- cupies instead a DW separating two AF domains. Delocalization leads then to similar three configu- rations to those obtained above with opposite spins: { , , }� �0 , {� �, ,0}, and { , , }0� � , but in addition, three con- figurations with one doubly occupied site {��, ,0 0}, { , , }0 0�� , and { , , }0 0�� , can be reached as excited states which cost Coulomb energyU. Moreover, three other configurations with interchanged �- and �-spins are then also accessible via the decay of double occupan- cies: {� �, ,0 }, {� �, ,0}, and {0, ,� �}. In the regime of large U, the total energy in the ground state can be found in a perturbative way, and as a result one obtains, E t t US � 2 4 2 . (6) Therefore, the Hilbert space for the latter solitonic solution is larger and one finds that this solution is al- ways more stable than the polaronic one [26]. The ar- gument applies also to 2D systems, where the DWs are more stable than the lines of polarons in an AF background. We compare the stability of such nonmagnetic SC domain walls with the BC stripe phases in which DWs are formed by pairs of magnetic atoms, as obtained by White and Scalapino [12] (cf. Fig. 3). In the three-band model, SC (BC) stripes correspond to DWs centered at metal (oxygen) sites, respectively [71–74]. 3.2. Effect of hopping anisotropy We begin by setting t � 0 and V � 0 with the goal of elucidating the effects of hopping anisotropy on the stripes. This is motivated by the fact that the first de- tection of static stripes in both charge and spin sectors was accomplished in Nd-LSCO [29] indicating that rare-earth elements doping is in some way helpful for pinning the stripe structure. Indeed, it produces a structural transition in the system from the LTO to LTT phase [75]. Both phases involve a distortion of the CuO2 plane by rotation of the CuO6 octahedra. In the LTO phase the tilt axis runs diagonally within the copper plane, such that all the oxygen atoms are displaced out of the plane. Conversely, in the LTT phase this rotation takes place around an axis oriented along the planar Cu O bonds. Therefore, oxygen at- oms on the tilt axis remain in the plane, while the ones in the perpendicular direction are displaced out of the plane. This provides a microscopic origin for in-plane anisotropies — the Cu–Cu hopping amplitude t de- pends on the Cu O bond and it is isotropic in the LTO phase and anisotropic in the LTT one. For a physical tilt angle of order 5°, the relative anisotropy taking t ty x� , �t x y y t t t � | | , (7) is weak and amounts to �t � 0 015. [76,77]. The direc- tion with a larger hopping amplitude coincides with the direction of a stronger superexchange coupling J. The possible relationship between this anisotropy and the onset of stripe phases has been intensively studied within anisotropic Hubbard (t tx y� ) or t-J (t tx y� , J Jx y� ) models by means of various tech- niques: unrestricted HF approach [76], DMRG [77], and QMC method [23]. The in-plane anisotropies might also be represented theoretically by on-site po- Stripe phases-possible ground state of the high-Tc superconductors Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 417 lx ly Fig. 3. Vertical bond-centered (VBC) and diagonal bond-centered (DBC) stripe phases as found for U/t � 5 at hole doping x /� 1 8. The meaning of the arrows and black circles as in Fig. 2. tentials as in the QMC study by Riera [24]. All these investigations have shown a pronounced tendency to forming stripe phases, which manifests itself by the re- duction of their energy [76,77], accompanied by the appearance of IC peaks in the spin and charge struc- ture factor [23,24]. It appears that a finite anisotropy of the next-nearest neighbor hopping term t might play a role in stabilizing diagonal incommensu- rate peaks observed in the spinglass phase of LSCO ( . . )0 02 0 06� �x [41–43,45–47]. Indeed, although the LTO phase is usually considered as isotropic, which is the case for nearest neighbor hopping and inter- action, a different length of the orthorhombic axes im- plies the need for an anisotropic t parameter. Exact diagonalization studies incorporating such anisotropy have shown that it strongly amplifies hole correlations along one direction and suppresses them along the other, resulting in a 1D pattern of holes [78]. It turns out, however, that the variation of the hop- ping anisotropy �t (7) has only a little visible effect on the local hole density, n l n nh x l lx x ( ) ,( , ), ( , ),� � � �� �1 0 0 (8) shown in Fig. 4 as a function of the x-direction coor- dinate lx for a given y-direction coordinate ly � 0, even at the unrealistically large anisotropy level �t � 0 22. , corresponding to t /tx � 11. and t /ty � 0 9. . Similarly, the anisotropy does not modify the modu- lated magnetization density, S l n nx l l l x x x ( ) ( ) ,( , ), ( , ),� � �� �1 1 2 0 0 with a site dependent factor ( ) 1 lx compensating modulation of the staggered magnetization density within a single AF domain. In contrast, the strong effect of finite anisotropy �t (7) is clearly demonstrated by variation of the expec- tation values of the bond hopping terms along the x- and y-directions, E l t c ct x x x l lx x ( ) , ( , ), † ( , ), � � � �� � 0 10 h. c. (10) E l t c ct y x y l lx x ( ) . ( , ), † ( , ), � � � �� 0 1 h. c. (11) These features are seen in Fig. 4. For the VSC stripes one finds a large anisotropy in the values of the ki- netic energies (10) and (11), which becomes espe- cially pronounced beside the stripes, and is strongly reinforced by the hopping anisotropy. Therefore, tak- ing into account that the hopping between two differ- ent charge densities is favored over motion between equal densities, one should expect that transverse charge fluctuations will always tune the direction of DWs along the weaker hopping direction in the anisotropic model. Analogous conclusion based on Fig. 5 might be drawn concerning the orientation of the VBC stripes. Regarding diagonal stripes, although a finite aniso- tropy in hopping is also reflected in the kinetic energy anisotropy, a system with either the DSC or DBC stripe pattern becomes topologically frustrated and consequently may gain less kinetic energy compared to a system with vertical stripes, taking a full advantage of the hopping anisotropy (cf. Tables 1 and 2). 418 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 Marcin Raczkowski et al. 0 4 8 12 16 0.1 0.2 0.3 0.4 0 4 8 12 16 0.1 0.2 0.3 0.4 0 4 8 12 16 –0.4 –0.2 0 0.2 0.4 0 4 8 12 16 –0.4 –0.2 0 0.2 0.4 0 4 8 12 16 –1.2 –1.0 –0.8 –0.6 –0.4 0 4 8 12 16 –1.2 –1.0 –0.8 –0.6 –0.4 0 4 8 12 16 lx lx –0.8 –0.7 –0.6 –0.5 –0.4 0 4 8 12 16 –0.8 –0.7 –0.6 –0.5 –0.4 n h S E /t tx E /t ty Fig. 4. Local hole n lh x( ) (top) and magnetization S lx ( ) (second row) density; kinetic energy E lt x x( ) (third row) and E lt y x( ) (bottom) projected on the bonds in the x-(y)-directions, respectively, of the VSC (left) and DSC (right) stripe phases shown in Fig. 2 (open circles) as well as of the ones obtained in the anisotropic model with t /tx y � 122. (filled circles). For clarity, the latter are shifted by one lattice constant from the origin of the coor- dinate system. Table 1. Site-normalized ground-state energy Etot, kinetic energy ( , )E Et x t y , and potential energy EU in the isotropic Hubbard model with U/t � 5 and x /� 1 8 as obtained for different stripe phases: vertical site-centered (VSC), diago- nal site-centered (DSC), vertical bond-centered (VBC) and diagonal bond-centered (DBC). In the HF, both types of vertical stripes are degenerate. Et x/t Et y/t EU/t E /ttot VB(S)C –0.6753 –0.6147 0.4900 –0.8000 DBC –0.6375 –0.6375 0.4726 –0.8024 DSC –0.6368 –0.6368 0.4696 –0.8040 Table 2. The same as in Table 1 but with the hopping ani- sotropy �t � 022. . Et x/t Et y/t EU/t E /ttot DBC –0.8143 –0.4807 0.4815 –0.8135 DSC –0.8098 –0.4836 0.4793 –0.8141 VB(S)C –0.8304 –0.4776 0.4938 –0.8142 The effect of an increasing anisotropy illustrates the phase diagram shown in Fig. 6 determined by va- rying U and the ratio t /tx y of the nearest-neighbor hoppings in the x- and y-directions, while maintaining constant t t t /x y� �( ) 2. We observe the generic crossover from vertical to diagonal stripes with in- creasing Coulomb interaction reported in early HF studies [4–7]. The transition from the VSC to DSC stripes appears in the isotropic case at U/t � 41. for x /� 1 8, and at a higher value U/t � 4 6. for x /� 1 6 (cf. Fig. 6,a). The corresponding phase boundary be- tween the VBC and DBC stripes is shifted towards stronger Coulomb interaction and occurs atU/t � 4 4. (5.0) for x /� 1 8 (x /� 1 6), respectively (cf. Fig. 6,b). The results shown in Fig. 6 have a simple physical interpretation. Stripe phases occur as a compromise between, on the one hand, the AF interactions be- tween magnetic ions and the local Coulomb interac- tions responsible for charge localization, and the ki- netic energy of doped holes which on the contrary favors charge delocalization. The kinetic energies in Table 1 show further that the vertical stripes are more favorable for charge dynamics. This result, which is not immediately obvious, has however a straightfor- ward origin. Namely, the HF always leads to a large spin polarization since it is the only way to minimize the on-site Coulomb repulsion. Indeed, removal of a �-spin electron at site i leads to relaxation of the �-spin electron energy level at this site. As a conse- quence, an alternating on-site level shift develops yielding an energetical motivation for the symmetry breaking and forming the AF order. However, the renormalization of the double occu- pancy energy involves a strong reduction of the ki- netic energy in the �-spin channel between site i and its neighboring sites, as an electron incoming into this site encounters a high energy potentialU ni� �� . There- fore, in the HF approximation we shall be able to identify dynamically favorable stripe patterns only by Stripe phases-possible ground state of the high-Tc superconductors Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 419 0 4 8 12 16 0.1 0.2 0.3 0.4 0 4 8 12 16 0.1 0.2 0.3 0.4 0 4 8 12 16 –0.4 –0.2 0 0.2 0.4 0 4 8 12 16 –0.4 –0.2 0 0.2 0.4 0 4 8 12 16 –1.2 –1.0 –0.8 –0.6 –0.4 0 4 8 12 16 –1.2 –1.0 –0.8 –0.6 –0.4 0 4 8 12 16 lx lx –0.8 –0.7 –0.6 –0.5 –0.4 0 4 8 12 16 –0.8 –0.7 –0.6 –0.5 –0.4 n h S E /t tx E /t ty Fig. 5. The same as in Fig. 4 but for the BC stripe phases shown in Fig. 3. 3.5 4.0 4.5 5.0 5.5 6.0 U/t 0.90 0.95 1.00 1.05 1.10 4.0 4.5 5.0 5.5 6.0 U/t 0.90 0.95 1.00 1.05 1.10 a b DSC VSC DBC HBCHSC VBC t /t x Fig. 6. Phase diagrams for stable: (a) site-centered (SC), and (b) bond-centered (BC) stripe structures obtained in the anisotropic Hubbard model on a 16 16� cluster for doping x /� 1 8 (solid lines) and on a 12 12� cluster for x /� 1 6 (dashed lines). Parameters: t� � 0, V � 0. comparing appropriate local magnetization densities. For example, charge fluctuations occur more readily in the VSC stripe geometry presumably due to their greater overall width indicating weaker correlation ef- fects (cf. Fig. 4). This explains their stability at small U where the consequent cost in potential energy EU becomes insignificant. By contrast, the DSC stripes are narrower having larger hole density along non- magnetic DWs. Moreover, magnetization density of their nearest neighbor sites is markedly enhanced as compared to the corresponding VSC stripe magnetiza- tion, as shown in Fig. 4 and in Table 3. The former also illustrates that the bonds connecting DWs with their nearest neighboring sites perpendicularly to the walls, have the main contribution to the kinetic en- ergy gain, in fact suppressed here by larger spin polar- ization. Taken together, the above features are re- flected in a more localized character of the DSC stripes, with a lower net double occupancy and hence a more favorable on-site energy EU (cf. Table 1). This clarifies the mechanism of the transition from the VSC to DSC stripes with increasingU. Table 3. Local hole � �nhi and magnetization � �Si z density of the site-centered stripes shown in Fig. 2, all labeled by decreasing hole density in the x-direction. In parentheses the values for the extended hopping model with t t� � / .015 are given. i 1 2 3 4 5 VSC � �nhi 0.364 0.234 0.067 0.014 0.006 (0.378) (0.234) (0.060) (0.013) (0.006) � �Si z 0.000 0.222 0.348 0.381 0.384 (0.000) (0.234) (0.357) (0.382) (0.384) DSC � �nhi 0.388 0.193 0.070 0.032 0.020 (0.405) (0.195) (0.066) (0.028) (0.017) � �Si z 0.000 0.262 0.352 0.373 0.380 (0.000) (0.272) (0.360) (0.377) (0.382) Turning now to the analogous crossover between the BC stripes, we shall again compare local hole and magnetization densities on and around their DWs. In contrast to the SC case, a VBC stripe phase possesses larger hole density along DWs, as illustrated in Fig. 3 and Table 4, suggesting that it is more localized than the DBC one. Nevertheless, a better renormalization of the double occupancy energy EU by the latter (cf. Table 1) follows from a stronger spin polarization not only of the DW atoms but also their nearest neighbors (cf. Fig. 3 and Table 4). This enhancement is directly responsible for a substantial reduction of the kinetic energy along bonds joining these atoms. Correspond- ingly, it accounts for a crossover from the DBC to VBC stripes in the smallU regime when the larger ki- netic energy gain becomes crucial. Table 4. The same as in Table 3 but for the bond-centered stripes. VBC stripe is unstable in the extended hopping model with t /t� � 015. — data in parentheses. i 1 2 3 4 VBC � �nhi 0.326 0.136 0.030 0.007 � �Si z 0.118 0.301 0.371 0.384 DBC � �nhi 0.314 0.115 0.047 0.023 (0.323) (0.110) (0.046) (0.021) � �Si z 0.145 0.322 0.365 0.378 (0.155) (0.333) (0.368) (0.380) We would like to emphasize that the above transi- tion between different types of stripe phases is not an artefact of the HF and occurs also between filled stripes obtained within more realistic approaches in- cluding local electron correlations. Indeed, slave-boson studies of the Hubbard model at the dop- ing x /� 1 9 have established that the transition from the filled VSC to DSC stripe phase appears at the value U/t � 5 7. , being much higher than that pre- dicted by the HF, which yieldsU/t � 3 8. [14]. In this method, enhanced stability of the VSC stripes follows from an additional variational parameter per each site di , reducing the on-site energy without a strong sup- pression of the kinetic energy. Remarkably, the total energy difference between the vertical SC and BC stripes at both doping levels is comparable to the accu- racy of the present calculation. Such degeneracy was also reported in the HF studies of the charge-transfer model [71]. However, when electron correlations are explicitly included the BC stripes are more stable at and above x /� 1 8 doping [21,79]. 3.3. Effect of the next-neighbor hopping t We now turn to the effect of a next-neighbor hop- ping t on the relative stability of the stripes. There are numerous experimental and theoretical results which support the presence of finite t in the cuprates. For ex- ample, recent slave-boson studies have revealed that the phenomena of the half-filled vertical stripes in LSCO requires a finite next-neighbor hopping t t / .� 0 2 [16]. Let us pause now for a moment to clarify the influ- ence of t on the DOS as well as on the FS using the electronic band which follows from a simple tight- binding model [80], E t k k t k kx y x y( ) (cos cos ) cos cos .k � � 2 4 (12) 420 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 Marcin Raczkowski et al. By the reduction from the CuO2 multiband model to an effective single-band model it has been found that t � 0 and t � 0 for hole doped system, and t � 0 and t � 0 in electron doped system [70]. Although an accidental cancellation of the various contributions re- sults in almost perfect electron-hole symmetry of the nearest neighbor hopping t, the next-neighbor hop- ping t asymmetry appears owing to the fact that the dominant contribution to the latter comes from a di- rect O–O hopping tpp in the case of a hole hopping. On the contrary, an electron hopping follows from a third order Cu � O � O � Cu process, being there- fore dominated by the Cu–O hopping element tpd . In the noninteracting limit the role of t is to shift the van Hove singularity away from the middle of the band, either to higher or to lower energy depending on its sign [80]. Figure 7 shows the tight-binding DOS, centered at � 0 with the condition N d( )� � 0, and the occupied states at the doping x /� 1 4. In the hole-doped case, with the vacuum as the zero electron state, the van Hove singularity lies in the lower part of the band. Conversely, in the case of electron dop- ing, with the vacuum as the zero hole state, the van Hove singularity is shifted towards higher energy part of the band, unoccupied by holes. Apart from breaking the electron-hole symmetry, the extra parameter t modifies the shape of the FS of the free electrons and indeed it becomes more consis- tent with the FS topology seen by ARPES [59,81,82]. In the electron-doped system NCCO, the low-energy spectral weight at the doping x � 0 04. is concentrated in small electron pockets around the (��,0) and (0,� �) points. Upon increasing doping, one observes both the modification of the hole pockets and the emergence of new low-lying spectral weights around ( , )� �� �/ /2 2 . Finally, at x � 015. the FS pieces evolve into a large holelike curve centered at M � ( , )� � . In contrast, it has been observed that in the lightly doped regime( . )x � 0 03 doped holes in LSCO enter into the hole pockets around (� �� �/ /2 2, ) points [83], implying that the FS is holelike and centered at the M point. However, in the heavy overdoped regime x � 0 3. it converts into the electronlike FS around the � � ( , )0 0 point. Figure 8,a shows that the model (12) with t � 0 has a nested square FS at half-filling which becomes electronlike and shrinks around the � point upon hole doping. However, negative t � 0 3. removes the FS nesting at half filling, and the FS expands in the ( , )�k 0 and ( , )0 � k directions, while contracts along the nodal ( , )k k� and ( , )�k k directions due to a large gradient dE/dk along the latter. Indeed, the eigenenergy map, illustrated in Fig. 9,a, has in this case a valleylike character with a minimum at the � point. Therefore the FS turns into a holelike one with experimentally observed arcs (cf. Fig. 8,a). In con- trast, the nearest neighbor hopping t with the same sign as t interchanges the expansion- and contraction directions which results in the electronlike FS. Stripe phases-possible ground state of the high-Tc superconductors Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 421 0.80.8 0.60.6 0.40.4 0.20.2 00 N ( )( 1 /t ) � –4 –4–2 –20 02 24 46 6 �/t �/t a b Fig. 7. Effect of the next-neighbor hopping t t� � / .03 on the noninteracting 2D DOS at the doping x /� 1 4: (a) hole doping (t � 1); (b) electron doping (t � 1). Dotted line shows the Fermi energy in the undoped case, whereas the gray area shows the states occupied by either electrons (a) or holes (b). k / y � 1.01.0 00 –1.0–1.0 k /x � –1.0 0 1.0 k /x � –1.0 0 1.0 a b Fig. 8. FS obtained in the tight binding model at the dop- ing x /� 1 4: (a) hole doping with t � 1 and: t� � 03. (black solid line), t� � 03. (gray solid line), and t� � 0 (dashed line); (b) electron doping with t � 1 and: t� � 03. (black solid line), t� � 03. (gray solid line), and t� � 0 (dashed line). The long-dashed line in both panels corre- sponds to the undoped case with t� � 0. The excessively large value of | | .t� � 03 as compared to LSCO was chosen only for more clarity of the figure. a b 3 0 –3 –6 6 3 0 –3 –� –� –� –� 0 0 0 0 � � � � kx kx ky ky E(k)/tE(k)/t Fig. 9. Eigenenergy maps of the tight-binding model (12) with t t� � / .03 as obtained for: (a) hole doping (t � 1); (b) electron doping (t � 1). Regarding the electron doped case with t � 1, shown in Fig. 8,b, positive t � 0 3. (dark solid line) also leads to the appearance of arc segments of the FS and makes it closer to experimental observations. In this case, however, the minimum energy is found at the M point, as illustrated in Fig. 9,b. It should be noted in passing that this FS describes the same situa- tion as the one obtained with t � 1 and t � 0 3. , indi- cated by the gray solid line in Fig. 8,a. In fact, the sign of t is less important and turns out to be equiva- lent to the ( , )� � shift of the momentum without changing the corresponding eigenvalues. Conse- quently, in order to imitate the effect of hole and elec- tron doping it is sufficient to study the Hamiltonian (1) only below half-filling and the alternation bet- ween two regimes is possible by the particle-hole transform, c ci i i † ( ) ,� 1 (13) mapping the model (12) with t � 0 onto the one with t � 0. Therefore, in order to avoid any further confu- sion concerning the signs of t and t in Eq. (12), we set hereafter t to be positive; then a negative t ( / )t t � 0 corresponds to hole doping, whereas a posi- tive one (t t �/ 0) indicates electron doping. The remarkable differences of the electronic struc- ture due to the broken hole-electron symmetry by t , result in different phase diagrams of LSCO and NCCO. In the former the long-range AF order is al- ready suppressed in the lightly doped regime x � 0 03. , while in the latter the antiferromagnetism is known to be quite robust at increasing electron doping, hence only commensurate spin fluctuations are observed at x � 015. [68]. The robustness of the commensurate spin fluctuations in the electron doped regime is consistent with the ED studies of the t–t –J [84,85] and t–t –t –J [86,87] models. It is also supported by the conclusion that a negative t promotes incommensuration at a lower doping level than a positive one, reached using the QMC technique applied to the extended Hubbard model [88]. Finally, the XPS measurements in NCCO show that the chemical potential monotonously in- creases with electron doping [69], whereas its shift is suppressed in the underdoped region of LSCO [66]. These data have been nicely reproduced in Ref. 86 for both compounds, except for the low doping regime of LSCO where stripes are expected. All these numerical and experimental results indicate that doped electrons might selforganize in a different way than holes do — in the latter case DWs are formed. Nevertheless, sta- ble diagonal stripes with one doped electron per site in a DW have been obtained in the slave-boson studies of a more realistic extended three-band model [73], so the problem is still open. Turning back to the competition between stripes in a doped system, Fig. 10,a shows that negative t stabi- lizes the DSC stripes, whereas positive t favors the VSC ones, within the parameter range where t does not drive a stripe melting. Analogous crossover from vertical stripes at small | |t to more complex in shape diagonal ones at t t � / .01 and t t � / .0 2 has been found in other HF studies [89]. The explanation is contained in Table 5: negative t gives a positive ki- netic energy contribution, which is much more readily minimized by the diagonal charge configuration. In- deed, despite the solitonic mechanism yielding a no- ticeable kinetic energy loss due to the transverse hop- ping t t � / .015, the overall kinetic energy loss in the case of DSC stripes along the diagonal ( )11 and antidiagonal ( )11 directions is smaller than the corre- sponding one for the VSC stripe. A more careful anal- ysis shows that hole propagation along the DSC stripe results in a contribution having the same sign as t . However, it is entirely canceled by the ones coming from diagonal bonds of the AF domains so that Et x y � � � 0. One observes further that positive t reduces the anisotropy between the kinetic energy gains in the x- and y-directions for the VSC stripes, and makes their sum more favorable, while negative t has the opposite effect. For the DSC stripes the total kinetic energy also follows the same trend. The explanation of these results follows from the reinforcement of stripe order by a negative t (cf. values in parenthesis in Table 3), which suppresses the hopping contributions, and its smearing out by positive t where hopping is enhanced. These trends agree with the earlier finding within the DMFT that the VSC stripe phase is destabilized by kink fluctuations [21]. However, this stripe (dis)or- dering tendency also leads to a considerably greater change in the Coulomb energy EU , listed in Table 5, for the DSC than for VSC stripes, which contributes significantly to the predominance of the former struc- ture for negative t . In fact, it follows from the in- 422 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 Marcin Raczkowski et al. 3.5 4.0 4.5 5.0 5.5 6.0 6.0 U/t –0.2 –0.1 0 0.1 0.2 3.5 4.0 4.5 5.0 5.5 U/t –0.2 –0.1 0 0.1 0.2 a b DSC V(H)SC DBC V(H)BC t /t’ Fig. 10. Phase boundaries for: (a) site-centered, and (b) bond-centered stripes as obtained in the extended Hubbard model with the next-neighbor hopping t� for doping x /� 1 8 (solid line) and x /� 1 6 (dashed line). crease of hole density within the nonmagnetic stripes and the magnetization density enhancement within the AF domains (cf. Table 3). Like their SC counterparts, DBC stripes are also stabilized by negative t resulting in a phase diagram shown in Fig. 10,b. In this case, expelling holes from the AF domains enhances not only magnetization of their atoms but also increases magnetic moment of the hole rich DWs, as illustrated in Table 4. This enhance- ment must, however, strongly suppress the dominant transverse kinetic energy gain of the VBC stripes. Therefore, the latter are already unstable at t t � / .015. It is worth noting that a finite diagonal hopping t should directly affect the competition between the d-wave pairing correlations and stripes. Indeed, a sys- tematic comparison of stripe and pairing instabilities within the DMRG framework has shown that when the stripes are weakened by positive t , the latter are strongly enhanced due to increasing pair mobility [13]. This effect is accompanied by a simultaneous en- hancement of the AF correlations [87]. Conversely, negative t reinforcing a static stripe order results in the suppression of pair formation in the underdoped region, as found both in the DMRG technique and Variational Monte Carlo (VMC) [90]. However, the enhanced pairing correlation, attributed to the change of the FS topology in LSCO, has been obtained in the optimally doped and overdoped regimes [91]. 3.4. Effect of the nearest-neighbor Coulomb interaction V We now investigate the changes in the stripe stabil- ity due to either repulsive (V � 0) or attractive (V � 0) nearest-neighbor Coulomb interaction, which give the phase boundaries between the VSC and DSC stripe phases shown in Fig. 1,a. We have found that realistic repulsive V favors the latter. The tendency towards the DSC stripe formation at V � 0 is primar- ily due to a large difference between charge densities at the atoms of the DW itself and at all their near- est-neighbor sites, a situation which is avoided in the case of VSC stripe phases (cf. Fig. 2). Consequently, the former optimize better the repulsive potential en- ergy component EV , as shown by the data reported in Table 4. Similarly, the fact that the nearest-neighbor interaction V is well minimized only by inhomo- geneous charge densities makes the DBC stripe phase more favorable than the VBC one, as shown in Fig. 11,b. While this is also the leading mechanism for both diagonal stripe suppression atV � 0, the asymme- try of the curve in Fig. 11 arises from the fact that the lower U values at the transition favor the higher ki- netic energy contributions available for the vertical stripes. Stripe phases-possible ground state of the high-Tc superconductors Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 423 3.5 4.0 4.5 5.0 5.5 6.0 6.0 U/t –0.4 –0.2 0.0 0.2 0.4 V /t 3.5 4.0 4.5 5.0 5.5 U/t –0.4 –0.2 0.0 0.2 0.4 a b DSC V(H)SC DBC V(H)BC Fig. 11. Phase diagrams for the site-centered (a) and bond-centered (b) stripes obtained in the extended Hub- bard model with the nearest-neighbor Coulomb interaction V for doping x /� 1 8 (solid line) and x /� 1 6 (dashed line). Table 5. Energies per site: ground-state energy Etot, kinetic energy contributions for the bonds along (10) Et x, (01) Et y, (11) Et x y � � and (11) Et x y � � directions, as well as the potential energy EU, all normalized per one site, in the extended hop- ping Hubbard model with U/t � 5 and x /� 1 8. VBC stripe is unstable at t t� � / .015. t t� � �/ .015 Et x/t Et y/t Et x y/t� � Et x y/t� � EU/t E /ttot VSC –0.15 –0.6876 –0.5886 0.0140 0.0140 0.4778 –0.7704 DBC –0.15 –0.6279 –0.6279 0.0000 0.0183 0.4562 –0.7813 DSC –0.15 –0.6275 –0.6275 0.0000 0.0188 0.4533 –0.7829 DBC 0.15 –0.6442 –0.6442 0.0000 –0.0282 0.4883 –0.8283 DSC 0.15 –0.6437 –0.6437 0.0000 –0.0279 0.4855 –0.8298 VB(S)C 0.15 –0.6612 –0.6372 –0.0169 –0.0169 0.4997 –0.8325 Table 6. Energies per site: ground-state energy Etot, ki- netic energy ( , )E Et x t y and potential energy ( , )E EU V com- ponents in the extended Hubbard model with the nea- rest-neighbor Coulomb interaction V for U/t � 5 and x /� 1 8. V/t Et x/t Et y/t EU/t EV/t E /ttot DBC –0.4 –0.6322 –0.6322 0.4626 –0.6194 –1.4212 DSC –0.4 –0.6319 –0.6319 0.4602 –0.6193 –1.4229 VB(S)C –0.4 –0.6655 –0.6083 0.4749 –0.6251 –1.4240 VB(S)C 0.4 –0.6838 –0.6214 0.5063 0.6207 –0.1782 DBC 0.4 –0.6424 –0.6424 0.4829 0.6176 –0.1843 DSC 0.4 –0.6412 –0.6412 0.4789 0.6171 –0.1864 However, it has been argued based on the results obtained using the SBA that an increasing repulsive interaction V favors half-filled vertical stripes, hence the latter take over at V/t � 01. in the parameter re- gime of x /� 1 8 andU/t � 10 [15]. This finding could naturally explain the appearance of filled diagonal stripes in the nickelates, provided that they were char- acterized by a small V term, and the stability of the half-filled vertical ones in the Nd-codoped cuprates due to possibly larger value ofV. It is also worth men- tioning other HF [92] and variational [93] studies in which a variety of intriguing stripe phases, coexisting at V/t � 15. with charge order, has been found in a broad doping region. 3.5. Effect of the lattice deformations So far, we have demonstrated that a finite aniso- tropy of the transfer integral t can tip the balance be- tween vertical and diagonal stripes. Here we will show that such anisotropy naturally emerges in a doped system with DWs, described by a single-band Peierls-Hubbard Hamiltonian, H t u c c U n n K uij ij ij i j i i i ij ij � � �� � �� � � � ( ) .† 1 2 2 (14) In this model we keep only the leading term and as- sume a linear dependence of the nearest neighbor hop- ping element tij on the lattice displacements uij , t u t uij ij ij( ) ( )� �0 1 � . (15) Furthermore, we include the elastic energy � K which allows to investigate the stability of the system with respect to a given lattice deformation and to de- termine the equilibrium configuration. For conve- nience, we parametrize the electron-lattice coupling with a single quantity, � �� 2 0t /K, with the parame- ter values K/t0 18� and � � 3 assumed following the earlier HF studies [26]. As previously, we focus on the doping x /� 1 8 (x /� 1 6) and present the results of calculations performed on 16 16� ( )12 12� clusters, respectively, with periodic boundary conditions. These calculations have shown that such clusters give the most stable filled stripe solutions for the selected doping levels. The model (14) was solved self-consis- tently in real space within the HF (2). Thereby, we used an approximate saddle-point formula for the equilibrium relation between the actual deformation 424 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 Marcin Raczkowski et al. 0 4 8 12 16 0.1 0.2 0.3 0.4 0 4 8 12 16 0.1 0.2 0.3 0.4 0 4 8 12 16 –0.4 –0.2 0 0.2 0.4 0 4 8 12 16 –0.4 –0.2 0 0.2 0.4 0 4 8 12 16 0.08 0.10 0.12 0.14 0.16 0 4 8 12 16 0.08 0.10 0.12 0.14 0.16 0 4 8 12 16 lx –1.4 –1.2 –1.0 –0.8 –0.6 –0.4 0 4 8 12 16 lx –1.4 –1.2 –1.0 –0.8 –0.6 –0.4 n h E /t t 0 x S u ;D x(0 ) , Å Fig. 12. Local hole n lh x( ) (top) and magnetization S lx ( ) (second row) density; fractional change of the length for the bonds to the right nearest-neighbor along the x-direc- tion ux ( )0 (circles) and double occupancy D lx( ) (squares) (third row), as well as the kinetic energy E lt x x( ) projected on the bonds in the x-direction (bottom) of the VSC (left) and DSC (right) stripe phases, as obtained in the Peierls-Hubbard model (14) with U/t � 5, � 05. and x /� 1 8 (filled symbols). For comparison the results ob- tained with � 0 are shown by open symbols. uij of a given bond and the bond-charge density � �c ci j † , u t K c cij i j ( ) † ,0 0� � � � �� h.c. (16) being a consequence of the linearity assumption in Eq. (15). Quite generally, it is a widely spread out belief that inhomogeneous states at finite doping are very sensitive to small changes of �, supported both by the HF [94,95] and ED studies [96]. Further, it has been shown that the electron-lattice interaction favors DW solutions over other possible phases, such as isolated polarons or bipolarons [26]. Therefore, a complete dis- cussion of the stripe phase stability in correlated ox- ides has to include the coupling to the lattice. We turn now to the most important aspect of this Section. Figures 12 and 13 illustrate the effect of the finite electron-lattice coupling � � 0 5. on the SC and BC stripes, respectively. Both figures give a clear demonstration that, in contrast to the hopping aniso- tropy �t (7) discussed above, finite � markedly modi- fies both the local hole density (8) and modulated magnetization (9) [cf. also Table 3 with 7 (SC stripes) and Table 4 with 8 (BC stripes)]. Basically, the influ- ence of � resembles the effect of positive t , smearing out the stripe order by ejecting holes from the DWs, being however much stronger. In fact, hole deloca- lization not only suppresses the magnetization within the AF domains, but also noticeably quenches mag- netic moments of the BC domain walls. These trends can be understood by considering energy increments: the kinetic Et , on-site EU , and elastic energy EK , as explained below. Table 7. Local hole � �n ih and magnetization � �Si z density at nonequivalent atoms of the SC stripe phases, all labeled by decreasing hole density in the x-direction, in the Peierls-Hubbard model on a 16 16� cluster with U/t � 5, � 05. and x /� 1 8. i 1 2 3 4 5 VSC � �nhi 0.270 0.212 0.103 0.038 0.022 � �Si z 0.000 0.146 0.259 0.310 0.321 DSC � �nhi 0.292 0.179 0.094 0.058 0.046 � �Si z 0.000 0.193 0.277 0.305 0.314 Table 8. The same as in Table 7 but for the BC stripe phases. i 1 2 3 4 VBC � �nhi 0.255 0.156 0.063 0.026 � �Si z 0.074 0.209 0.291 0.319 DBC � �nhi 0.248 0.130 0.073 0.049 � �Si z 0.103 0.243 0.294 0.312 One should realize that a system described by the Hamiltonian (14) might be unstable towards lattice deformations only if the covalency increase is large enough to compensate both the EU and EK energy cost. Without the electron-lattice coupling, a compro- mise solution is mainly reached by developing a strong magnetic order in the AF domains, where a possible kinetic energy gain is irrelevant, and by forming non- magnetic or weakly magnetic DWs with large hole density. As we have already shown, transverse charge fluctuations around the DWs yield the leading kinetic energy contribution. However, enhanced covalency and mixing of the lower � �d and higher � ��d U energy states between a DW and the surrounding sites partly delocalize these states and increase double oc- cupancy, Stripe phases-possible ground state of the high-Tc superconductors Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 425 0 4 8 12 16 0.1 0.2 0.3 0.4 0 4 8 12 16 0.1 0.2 0.3 0.4 0 4 8 12 16 –0.4 –0.2 0 0.2 0.4 0 4 8 12 16 –0.4 –0.2 0 0.2 0.4 0 4 8 12 16 0.08 0.10 0.12 0.14 0.16 0 4 8 12 16 0.08 0.10 0.12 0.14 0.16 0 4 8 12 16 lx –1.4 –1.2 –1.0 –0.8 –0.6 –0.4 0 4 8 12 16 lx –1.4 –1.2 –1.0 –0.8 –0.6 –0.4 n h E /t t 0 x S u ;D x(0 ) , Å Fig. 13. The same as in Fig. 12 but for the bond-centered stripes. D l n nx l lx x ( ) .( , ), ( , ),� � �� �0 0 (17) Indeed, in the � � 0 case, double occupancy D lx( ) reaches its maximum at the DWs, as illustrated in Figs. 12 and 13. The only exception is the DSC stripe phase (right panels of Fig. 12) with the largest D lx( ) in the AF domains. As a consequence, the latter is the most localized one with the smallest kinetic energy gain (cf. Table 1). The situation changes when turning on the elec- tron-lattice coupling. When the electrons couple to the lattice (� � 0), the bonds contract, and the saddle point values of the distortions (16): u uij ij ( )0 � � � along (10) and (01) direction, respectively, are finite. How- ever, a nonuniform charge distribution results in a dif- ferent bondlength in the cluster. This is illustrated in Figs. 12 and 13 showing a fractional change of the length for the bonds to the right nearest-neighbor along the x-direction ux ( )0 (third row). Although the values of uij ( )0 in the AF domains are also substantial, the largest lattice deformations � � �c ci j † appear ei- ther on the bonds connecting atoms of the DWs with their nearest neighbors (cf. Fig. 12), or on the bonds which join two atoms of the bond-centered DWs (cf. Fig. 13). Accordingly, a strengthening nearest neigh- bor hopping (15) enables a larger kinetic energy gain on these bonds (cf. bottom of Figs. 12 and 13). Table 9. Ground-state energy Etot per site, kinetic energy ( , )E Et x t y and potential energy ( , )E EU K components, as ob- tained in the Peierls-Hubbard model. Parameters: U/t � 5, � 05. , and x /� 1 8. Et x/t Et y/t EU/t EK/t E /ttot DBC –0.9679 –0.9679 0.6478 0.2548 –1.0332 DSC –0.9670 –0.9670 0.6450 0.2544 –1.0346 VB(S)C –1.0496 –0.9248 0.6719 0.2638 –1.0387 As expected, the increasing covalency is accompa- nied by partial quenching of magnetic moments. In or- der to appreciate this tendency, let us consider a site in the AF domain with larger density of �-spin elec- trons (at A sublattice). Once the magnetization is re- duced, the corresponding �-spin energy level which belongs to the lower Hubbard band is pushed up- wards, and the �-spin of the upper Hubbard band goes down. As a result, the locally raised �-spin state be- comes stronger mixed with �-spin states at the sur- rounding sites of B sublattice, and simultaneously bond-charge density increases. At the same time, elec- trons, jumping forth and back between the central site with the �-spin polarization and its nearest neighbors with the �-spin one, enhance considerably double oc- cupancy D lx( ), as shown in Figs. 12 and 13. This weakens the stripe order and results in a more uniform distribution of D lx( ). Of course, the increase of the elastic energy and concomitant enhancement of the on-site energy, both owing to finite bond contractions (16), is compensated by the kinetic energy gain and the total energy is lowered (cf. Tables 1 and 9). We close this Section with the phase diagrams shown in Fig. 14. They were obtained by varying U and the coefficient �, while maintaining constant K/t0 18� . The increased stability of vertical stripes follows from the relative stronger enhancement of the local hopping elements (15) (and consequently larger gain of the kinetic energy), especially on the bonds in the direction perpendicular to the DWs itself. 4. Summary In summary, we have shown that a competition be- tween magnetic energy of interacting almost localized electrons and the kinetic energy of holes created by doping leads to the formation of new type of coexist- ing charge and spin order — the stripe phases. We have shown that vertical (horizontal) and diagonal stripes dominate the behavior of the charge structures formed by doping the antiferromagnet away from half filling, using the solutions obtained for the Hubbard model within the HF approximation in the physically interesting regime of the Coulomb interaction. The de- tailed charge distribution and the type of stripe order depend on the ratio U/t, on the value of the next-neighbor hopping t , and on the nearest-neighbor Coulomb interaction V. We have also shown that a strong electron-lattice coupling might be responsible for the appearance of the vertical stripes observed in the superconducting cuprates at x /� 1 8. Altogether, although some experimentally ob- served trends could be reproduced already in the HF approach, the presented results indicate that strong electron correlations play a crucial role in the stripe 426 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 Marcin Raczkowski et al. 4.0 5.0 6.0 7.0 7.0 U/t 0 0.1 0.2 0.3 0.4 4.0 5.0 6.0 U/t 0 0.1 0.2 0.3 0.4 a b DSC VSC DBC VBC � � Fig. 14. Phase diagrams for site-centered (a) and bond- centered (b) stripe structures as calculated from the Peierls-Hubbard model for doping x /� 1 8 (solid line) and x /� 1 6 (dashed line). phases and have to be included for a more quantitative analysis. Further progress both in the experiment and in the theory is necessary to establish the possible role of stripes in the phenomenon of high temperature su- perconductivity. M. Raczkowski was supported by a Marie Curie fellowship of the European Community program un- der number HPMT2000-141. This work was supported by the the Polish Ministry of Scientific Research and Information Technology, Project No. 1 P03B 068 26, and by the Minist�re Fran�ais des Affaires Etrang�res under POLONIUM 09294VH. 1. J.G. Bednorz and K.A. M�ller, Z. Phys. B64, 189 (1986). 2. E.W. Carlson, V.J. Emery, S.A. Kivelson, D. Orgad, Concepts in High Temperature Superconductivity in: The Physics of Conventional and Unconventional Su- perconductors, K.H. Bennemann and J.B. Ketterson (eds.), Springer-Verlag (2002), cond-mat/0206217. 3. J. Zaanen and O. Gunnarsson, Phys. Rev. B40, 7391 (1989). 4. D. Poilblanc and T.M. Rice, Phys. Rev. B39, 9749 (1989). 5. H.J. Schulz, J. Phys. (Paris) 50, 2833 (1989); Phys. Rev. Lett. 64, 1445 (1990). 6. M. Kato, K. Machida, H. Nakanishi, and M. Fujita, J. Phys. Soc. Jpn. 59, 1047 (1990). 7. M. Inui and P.B. Littlewood, Phys. Rev. B44, 4415 (1991). 8. O. Zachar, S.A. Kivelson, and V.J. Emery, Phys. Rev. B57 1422 (1998); J. Zaanen, Physica C317, 217 (1999). 9. S.A. Kivelson and V.J. Emery, Synth. Metals 80, 151 (1996). 10. U. L�w, V.J. Emery, K. Fabricius, and S.A. Kivelson, Phys. Rev. Lett. 72, 1918 (1994). 11. C.N.A. van Duin and J. Zaanen, Phys. Rev. Lett. 80, 1513 (1998). 12. S.R. White and D.J. Scalapino, Phys. Rev. Lett. 80, 1272 (1998). 13. S.R. White and D.J. Scalapino, Phys. Rev. B60, R753 (1999). 14. G. Seibold, E. Sigmund, and V. Hizhnyakov, Phys. Rev. B57, 6937 (1998). 15. G. Seibold, C. Castellani, C. Di Castro, and M. Grilli, Phys. Rev. B58, 13506 (1998). 16. G. Seibold and J. Lorenzana, Phys. Rev. B69, 134513 (2004). 17. D. G�ra, K. Ro�ciszewski, and A.M. Ole�, Phys. Rev. B60, 7429 (1999). 18. T. Tohyama, S. Nagai, Y. Shibata, and S. Maekawa, Phys. Rev. Lett. 82, 4910 (1999). 19. P. Wr�bel and R. Eder, Phys. Rev. B62, 4048 (2000). 20. M. Fleck, A.I. Lichtenstein, E. Pavarini, and A.M. Ole�, Phys. Rev. Lett. 84, 4962 (2000). 21. M. Fleck, A.I. Lichtenstein, and A.M. Ole�, Phys. Rev. B64, 134528 (2001). 22. M.G. Zacher, R. Eder, E. Arrigoni, and W. Hanke, Phys. Rev. Lett. 85, 2585 (2000); Phys. Rev. B65, 045109 (2002). 23. F. Becca, L. Capriotti, and S. Sorella, Phys. Rev. Lett. 87, 167005 (2001). 24. J. Riera, Phys. Rev. B64, 104520 (2001). 25. M. Raczkowski, R. Fr�sard, and A.M. Ole�, Physica B359–361, 780 (2005). 26. J. Zaanen and A.M. Ole�, Ann. Phys. (Leipzig) 5, 224 (1996). 27. S.A. Kivelson, I.P. Bindloss, E. Fradkin, V. Ogane- syan, J.M. Tranquada, A. Kapitulnik, and C. Howald, Rev. Mod. Phys. 75, 1201 (2003). 28. A. Damascelli, Z. Hussain, and Z.-X. Shen, Rev. Mod. Phys. 75, 473 (2003). 29. J.M. Tranquada, B.J. Sternlieb, J.D. Axe, Y. Na- kamura, and S. Uchida, Nature 375, 561 (1995). 30. J.M. Tranquada, J.D. Axe, N. Ichikawa, Y. Naka- mura, S. Uchida, and B. Nachumi, Phys. Rev. B54, 7489 (1996). 31. N. Ichikawa, S. Uchida, J.M. Tranquada, T. Nie- m�ller, P.M. Gehring, S.-H. Lee, and J.R. Schneider, Phys. Rev. Lett. 85, 1738 (2000). 32. J.M. Tranquada, J.D. Axe, N. Ichikawa, A.R. Moo- denbaugh, Y. Nakamura, and S. Uchida, Phys. Rev. Lett. 78, 338 (1997). 33. P.M. Singer, A.W. Hunt, A.F. Cederstr�m, and T. Imai, Phys. Rev. B60, 15345 (1999). 34. S. Wakimoto, R.J. Birgeneau, Y. Fujimaki, N. Ichi- kawa, T. Kasuga, Y.J. Kim, K.M. Kojima, S.-H. Lee, H. Niko, J.M. Tranquada, S. Uchida, and M. von Zimmermann, Phys. Rev. B67, 184419 (2003). 35. G.B. Teitel’baum, B. B�chner, and H. de Gronckel, Phys. Rev. Lett. 84, 2949 (2000). 36. H.-H. Klauss, W. Wagener, M. Hillberg, W. Kop- mann, H. Walf, F.J. Litterst, M. H�cker, and B. B�chner, Phys. Rev. Lett. 85, 4590 (2000). 37. M. Fujita, H. Goka, K. Yamada, and M. Matsuda, Phys. Rev. Lett. 88, 167008 (2002). 38. H. Kimura, Y. Noda, H. Goka, M. Fujita, K. Ya- mada, M. Mizumaki, N. Ikeda, and H. Ohsumi, Phys. Rev. B70, 134512 (2004). 39. M. Fujita, H. Goka, K. Yamada, J.M. Tranquada, and L.P. Regnault, Phys. Rev. B70, 104517 (2004). 40. K. Yamada, C.H. Lee, K. Kurahashi, J. Wada, S. Wa- kimoto, S. Ueki, H. Kimura, Y. Endoh, S. Hosoya, G. Shirane, R.J. Birgeneau, M. Greven, M.A. Kastner, and Y.J. Kim, Phys. Rev. B57, 6165 (1998). 41. S. Wakimoto, G. Shirane, Y. Endoh, K. Hirota, S. Ueki, K. Yamada, R.J. Birgeneau, M.A. Kastner, Y.S. Lee, P.M. Gehring, and H.S. Lee, Phys. Rev. B60, R769 (1999). 42. S. Wakimoto, R.J. Birgeneau, M.A. Kastner, Y.S. Lee, R. Erwin, P.M. Gehring, S.H. Lee, M. Fujita, K. Yamada, Y. Endoh, K. Hirota, and G. Shirane, Phys. Rev. B61, 3699 (2000). 43. M. Fujita, K. Yamada, H. Hiraka, P.M. Gehring, S.H. Lee, S. Wakimoto, and G. Shirane, Phys. Rev. B65, 064505 (2002). Stripe phases-possible ground state of the high-Tc superconductors Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 427 44. N. Hasselmann, A.H. Castro Neto, and C. Morais Smith, Phys. Rev. B69, 014424 (2004). 45. M. Matsuda, Y.S. Lee, M. Greven, M.A. Kastner, R.J. Birgeneau, K. Yamada, Y. Endoh, P. B�ni, S.-H. Lee, S. Wakimoto, and G. Shirane, Phys. Rev. B61, 4326 (2000). 46. M. Matsuda, M. Fujita, K. Yamada, R.J. Birgeneau, M.A. Kastner, H. Hiraka, Y. Endoh, S. Wakimoto, and G. Shirane, Phys. Rev. B62, 9148 (2000). 47. M. Matsuda, M. Fujita, K. Yamada, R.J. Birgeneau, Y. Endoh, and G. Shirane, Phys. Rev. B65, 134515 (2002). 48. A.W. Hunt, P.M. Singer, K.R. Thurber, and T. Imai, Phys. Rev. Lett. 82, 4300 (1999). 49. Y. Ando, K. Segawa, S. Komiya, and A.N. Lavrov, Phys. Rev. Lett. 88, 137005 (2002). 50. M. Dumm, S. Komiya, Y. Ando, and D.N. Basov, Phys. Rev. Lett. 91, 077004 (2003). 51. B.O. Wells, Y.S. Lee, M.A. Kastner, R.J. Chris- tanson, R.J. Birgeneau, K. Yamada, Y. Endoh, and G. Shirane, Science 277, 1067 (1997); Y.S. Lee, R.J. Birgeneau, M.A. Kastner, Y. Endoh, S. Wakimoto, K. Yamada, R.W. Erwin, S.-H. Lee, and G. Shirane, Phys. Rev. B60, 3643 (1999). 52. K. Hirota, K. Yamada, I. Tanaka, and H. Kojima, Physica B241, 817 (1998). 53. H. Kimura, K. Hirota, H. Matsushita, K. Yamada, Y. Endoh, S.-H. Lee, C.F. Majkrzak, R. Erwin, G. Shi- rane, M. Greven, Y.S. Lee, M.A. Kastner, and R.J. Birgeneau, Phys. Rev. B59, 6517 (1999); J.M. Tran- quada, N. Ichikawa, K. Kakurai, and S. Uchida, J. Phys. Chem. Solids 60, 1019 (1999). 54. P. Dai, H.A. Mook, R.D. Hunt, and F. Dov�an, Phys. Rev. B63, 054525 (2001). 55. H.A. Mook, P. Dai, and F. Dov�an, Phys. Rev. Lett. 88, 097004 (2002). 56. H.A. Mook and B.C. Chakoumakos, J. Supercond. 10, 389 (1997). 57. C. Howald, H. Eisaki, N. Kaneko, M. Greven, and A. Kapitulnik, Phys. Rev. B67, 014533 (2003); A. Fang, C. Howald, N. Kaneko, M. Greven, A. Kapitulnik, Phys. Rev. B70, 214514 (2004). 58. A. Ino, C. Kim, M. Nakamura, T. Yoshida, T. Mizo- kawa, Z.-X. Shen, A. Fujimori, T. Kakeshita, H. Ei- saki, and S. Uchida, Phys. Rev. B62, 4137 (2000). 59. A. Ino, C. Kim, M. Nakamura, T. Yoshida, T. Mi- zokawa, A. Fujimori, Z.-X. Shen, T. Kakeshita, H. Eisaki, and S. Uchida, Phys. Rev. B65, 094504 (2002). 60. A. Ino, T. Mizokawa, K. Kobayashi, A. Fujimori, T. Sasagawa, T. Kimura, K. Kishio, K. Tamasaku, H. Eisaki, and S. Uchida, Phys. Rev. Lett. 81, 2124 (1998). 61. X.J. Zhou, P. Bogdanov, S.A. Kellar, T. Noda, H. Eisaki, S. Uchida, Z. Hussain, and Z.-X. Shen, Sci- ence 286, 268 (1999). 62. X.J. Zhou, T. Yoshida, S.A. Kellar, P.V. Bogdanov, E.D. Lu, A. Lanzara, M. Nakamura, T. Noda, T. Kake- shita, H. Eisaki, S. Uchida, A. Fujimori, Z. Hussain, and Z.-X. Shen, Phys. Rev. Lett. 86, 5578 (2001). 63. M.I. Salkola, V.J. Emery, and S.A. Kivelson, Phys. Rev. Lett. 77, 155 (1996). 64. V.J. Emery, S.A. Kivelson, and O. Zachar, Phys. Rev. B56, 6120 (1997). 65. P. Wr�bel, A. Maci�g, and R. Eder, cond-mat/0408703. 66. A.Ino, T. Mizokawa, A. Fujimori, K. Tamasaku, H. Eisaki, S. Uchida, T. Kimura, T. Sasagawa, and K. Kishio, Phys. Rev. Lett. 79, 2101 (1997). 67. T. Noda, H. Eisaki, and S. Uchida, Science 286, 265 (1999). 68. K. Yamada, K. Kurahashi, T. Uefuji, M. Fujita, S. Park, S.-H. Lee, and Y. Endoh, Phys. Rev. Lett. 90, 137004 (2003). 69. N. Harima, J. Matsuno, A. Fujimori, Y. Onose, Y. Taguchi, and Y. Tokura, Phys. Rev. B64, 220507 (2001). 70. L.F. Feiner, J.H. Jefferson and R. Raimondi, Phys. Rev. B53, 8751 (1996). 71. T. Mizokawa and A. Fujimori, Phys. Rev. B56, 11920 (1997). 72. Z.G. Yu, J. Zang, J.T. Gammel, and A.R. Bishop, Phys. Rev. B57, R3241 (1998). 73. A. Sadori and M. Grilli, Phys. Rev. Lett. 84, 5375 (2000). 74. J. Lorenzana and G. Seibold, Phys. Rev. Lett. 89, 136401 (2002). 75. A. B�chner, M. Breuer, A. Freimuth, and A.P. Kampf, Phys. Rev. Lett. 73, 1841 (1994). 76. A. Normand and A.P. Kampf, Phys. Rev. B64, 024521 (2001). 77. A.P. Kampf, D.J. Scalapino, and S.R. White, Phys. Rev. B64, 052509 (2001). 78. J.M. Tipper and K.J.E. Vos, Phys. Rev. B67, 144511 (2003). 79. P. Wr�bel, A. Maci�g, and R. Eder, J. Phys. Condens. Matter 18, 1249 (2006). 80. M. Fleck, A.M. Ole�, and L. Hedin, Phys. Rev. B56, 3159 (1997). 81. A.M. King, Z.-X. Shen, D.S. Dessau, B.O. Wells, W.E. Spicer, A.J. Arko, D.S. Marshall, J. DiCarlo, A.G. Loeser, C.H. Park, E.R. Ratner, J.L. Peng, Z.Y. Li, and R.L. Greene, Phys. Rev. Lett. 70, 3159 (1993). 82. N.P. Armitage, F. Ronning, D.H. Lu, C. Kim, A. Damascelli, K.M. Shen, D.L. Feng, H. Eisaki, Z.-X. Shen, P.K. Mang, N. Kaneko, M. Greven, Y. Onose, Y. Taguchi, and Y. Tokura, Phys. Rev. Lett. 88, 257001 (2002). 83. T. Yoshida, X.J. Zhou, T. Sasagawa, W.L. Yang, P.V. Bogdanov, A. Lanzara, Z. Hussain, T. Mizokawa, A. Fujimori, H. Eisaki, Z.-X. Shen, T. Kakeshita, and S. Uchida, Phys. Rev. Lett. 91, 027001 (2003). 84. T. Tohyama and S. Maekawa, Phys. Rev. B49, 3596 (1994). 85. R.J. Gooding, K.J. E. Vos, and P.W. Leung, Phys. Rev. B50, 12866 (1994). 428 Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 Marcin Raczkowski et al. 86. T. Tohyama and S. Maekawa, Phys. Rev. B67, 092509 (2003). 87. T. Tohyama, Phys. Rev. B70, 174517 (2004). 88. A. Duffy and A. Moreo, Phys. Rev. B52, 15607 (1995). 89. A. Normand and A.P. Kampf, Phys. Rev. B65, 020509 (2002). 90. A. Himeda, T. Kato, and M. Ogata, Phys. Rev. Lett. 88, 117001 (2002). 91. A.T. Shih, T.K. Lee, R. Eder, C.-Y. Mou, and Y.C. Chen, Phys. Rev. Lett. 92, 227002 (2004). 92. T. Kato and M. Kato, J. Phys. Soc. Jpn. 69, 3972 (2000). 93. K. Ro�ciszewski and A.M. Ole�, J. Phys.: Condens. Matter 15, 8363 (2003). 94. K. Yonemitsu, A.R. Bishop, and J. Lorenzana, Phys. Rev. Lett. 69, 965 (1992); Phys. Rev. B47, 8065 (1993); ibid. B47, 12059 (1993). 95. M. Raczkowski and A.M. Ole�, Physica C387, 82 (2003). 96. A. Dobry, A. Greco, J. Lorenzana, and J. Riera, Phys. Rev. B49, 505 (1994); J. Lorenzana and A. Dobry Phys. Rev. B50, 16094 (1994). Stripe phases-possible ground state of the high-Tc superconductors Fizika Nizkikh Temperatur, 2006, v. 32, Nos. 4/5 429
id nasplib_isofts_kiev_ua-123456789-120185
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
issn 0132-6414
language English
last_indexed 2025-12-01T20:34:54Z
publishDate 2006
publisher Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
record_format dspace
spelling Raczkowski, M.
Oles, A.M.
Fresard, R.
2017-06-11T12:05:42Z
2017-06-11T12:05:42Z
2006
Stripe phases — possible ground state of the high-Tc superconductors / M. Raczkowski, A.M. Oles, R. Fresard // Физика низких температур. — 2006. — Т. 32, № 4-5. — С. 411-429. — Бібліогр.: 96 назв. — англ.
0132-6414
Pacs: 71.10.Fd, 71.27.+a, 74.25.–q, 74.72.–h
https://nasplib.isofts.kiev.ua/handle/123456789/120185
Based on the mean-field method applied either to the extended single-band Hubbard model or to the single-band Peierls-Hubbard Hamiltonian we study the stability of both site-centered and bond-centered charge domain walls. The difference in energy between these phases is found to be small. Therefore, moderate perturbations to the pure Hubbard model, such as next nearest neighbor hopping, lattice anisotropy, or coupling to the lattice, induce phase transitions, shown in the corresponding phase diagrams. In addition, we determine for stable phases charge and magnetization densities, double occupancy, kinetic and magnetic energies, and investigate the role of a finite electron-lattice coupling. We also review experimental signatures of stripes in the superconducting copper oxides.
M. Raczkowski was supported by a Marie Curie fellowship of the European Community program under number HPMT2000-141. This work was supported by the the Polish Ministry of Scientific Research and Information Technology, Project No. 1 P03B 068 26, and by the Ministre Franais des Affaires Etrangres under POLONIUM 09294VH.
en
Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
Физика низких температур
General Aspects
Stripe phases — possible ground state of the high-Tc superconductors
Article
published earlier
spellingShingle Stripe phases — possible ground state of the high-Tc superconductors
Raczkowski, M.
Oles, A.M.
Fresard, R.
General Aspects
title Stripe phases — possible ground state of the high-Tc superconductors
title_full Stripe phases — possible ground state of the high-Tc superconductors
title_fullStr Stripe phases — possible ground state of the high-Tc superconductors
title_full_unstemmed Stripe phases — possible ground state of the high-Tc superconductors
title_short Stripe phases — possible ground state of the high-Tc superconductors
title_sort stripe phases — possible ground state of the high-tc superconductors
topic General Aspects
topic_facet General Aspects
url https://nasplib.isofts.kiev.ua/handle/123456789/120185
work_keys_str_mv AT raczkowskim stripephasespossiblegroundstateofthehightcsuperconductors
AT olesam stripephasespossiblegroundstateofthehightcsuperconductors
AT fresardr stripephasespossiblegroundstateofthehightcsuperconductors