1: \documentclass{ws-ijmpb}
2:
3: \usepackage{graphicx}
4: \usepackage{amssymb}
5:
6: \begin{document}
7:
8: \markboth{A.~Sherman \& M.~Schreiber} {Incommensurate spin dynamics in
9: underdoped cuprate perovskites}
10:
11: \title{INCOMMENSURATE SPIN DYNAMICS\\ IN UNDERDOPED CUPRATE PEROVSKITES}
12:
13: \author{A.~Sherman}
14:
15: \address{Institute of Physics, University of Tartu, Riia 142, 51014
16: Tartu, Estonia\\alexei@fi.tartu.ee}
17:
18: \author{M.~Schreiber}
19:
20: \address{Institut f\"ur Physik, Technische Universit\"at, D-09107
21: Chemnitz, Federal Republic of Germany}
22:
23: \maketitle
24:
25: \begin{history}
26: \received{Day Month Year} \revised{Day Month Year}
27: %\accepted{(Day Month Year)}
28: %\comby{(xxxxxxxxxx)}
29: \end{history}
30:
31: \begin{abstract}
32: The incommensurate magnetic response observed in normal-state cuprate
33: pe\-rov\-ski\-tes is interpreted based on the projection operator
34: formalism and the $t$-$J$ model of Cu-O planes. In agreement with
35: experiment the calculated dispersion of maxima in the susceptibility
36: has the shape of two parabolas with upward and downward branches which
37: converge at the antiferromagnetic wave vector. The maxima are located
38: at the momenta $(\frac{1}{2},\frac{1}{2}\pm\delta)$,
39: $(\frac{1}{2}\pm\delta,\frac{1}{2})$ and at
40: $(\frac{1}{2}\pm\delta,\frac{1}{2}\pm\delta)$,
41: $(\frac{1}{2}\pm\delta,\frac{1}{2}\mp\delta)$ in the lower and upper
42: parabolas, respectively. The upper parabola reflects the dispersion of
43: magnetic excitations of the localized Cu spins, while the lower
44: parabola arises due to a dip in the spin-excitation damping at the
45: antiferromagnetic wave vector. For moderate doping this dip stems from
46: the weakness of the interaction between the spin excitations and holes
47: near the hot spots. The frequency dependence of the susceptibility is
48: shown to depend strongly on the hole bandwidth and damping and varies
49: from the shape observed in YBa$_2$Cu$_3$O$_{7-y}$ to that inherent in
50: La$_{2-x}$Sr$_x$CuO$_4$.
51: \end{abstract}
52:
53: \keywords{Cuprate superconductors; magnetic properties; $t$-$J$ model.}
54:
55: \section{Introduction}
56: One of the most interesting features of the inelastic neutron
57: scattering in lanthanum cuprates is that for hole concentrations $x
58: \gtrsim 0.04$, low temperatures and small energy transfers the
59: scattering is peaked at incommensurate momenta
60: $(\frac{1}{2},\frac{1}{2}\pm\delta)$,
61: $(\frac{1}{2}\pm\delta,\frac{1}{2})$ in the reciprocal lattice units
62: $2\pi/a$ with the lattice period $a$.\cite{Yoshizawa} For $x\lesssim
63: 0.12$ the incommensurability parameter $\delta$ is approximately equal
64: to $x$.\cite{Yamada} For larger $x$ the parameter saturates near the
65: value $\delta\approx 0.12$. The incommensurate response was observed
66: both below and above $T_c$.\cite{Mason93} Recently the analogous
67: low-frequency incommensurability was observed also in
68: YBa$_2$Cu$_3$O$_{7-y}$.\cite{Dai} This gives ground to suppose that the
69: incommensurability is a common feature of cuprate perovskites which
70: does not depend on subtle details of the energy structure. However, for
71: larger frequencies the susceptibility differs essentially in these two
72: types of cuprates. In the underdoped YBa$_2$Cu$_3$O$_{7-y}$ and some
73: other cu\-p\-ra\-tes both below and above $T_c$ a pronounced maximum is
74: observed at frequencies $\omega_r=25-40$~meV.\cite{Bourges} In the
75: momentum space the magnetic response is sharply peaked at the
76: antiferromagnetic wave vector ${\bf Q}=(\frac{1}{2},\frac{1}{2})$ for
77: this frequency. Contrastingly, no maximum at $\omega_r$ was observed in
78: lanthanum cuprates. Instead for low temperatures and frequencies of
79: several millielectronvolts a broad feature was detected.\cite{Aeppli}
80: For even larger frequencies the magnetic response becomes again
81: incommensurate in both types of cuprates with peaks located at
82: $(\frac{1}{2}\pm\delta,\frac{1}{2}\pm\delta)$,
83: $(\frac{1}{2}\pm\delta,\frac{1}{2}\mp
84: \delta)$.\cite{Bourges,Hayden96,Hayden04,Tranquada} In contrast to the
85: low-frequency incommensurability in which the incommensurability
86: parameter decreases with increasing frequency, the parameter of the
87: high-frequency incommensurability grows or remains practically
88: unchanged with frequency. Thus, the dispersion of maxima in the
89: susceptibility resembles two parabolas with upward- and
90: downward-directed branches which converge at {\bf Q} and near the
91: frequency $\omega_r$.\cite{Dai,Tranquada}
92:
93: The nature of the magnetic incommensurability is the subject of active
94: discussion now. The most frequently used approaches for its explanation
95: are based on the picture of itinerant electrons with the susceptibility
96: calculated in the random phase approximation\cite{Liu,Brinckmann} and
97: on the stripe domain picture.\cite{Tranquada,Hizhnyakov} In the former
98: approach the low-frequency incommensurability is connected with the
99: Fermi surface nesting in the normal state or with the nesting in
100: constant-energy contours in the superconducting case. This imposes
101: rather stringent requirements on the electron energy spectrum, since
102: the nesting has to persist in the range of hole concentrations $0.04
103: \lesssim x\lesssim 0.18$ where the incommensurability is observed and
104: the nesting momentum has to change in a specific manner with doping to
105: ensure the known dependence of the incommensurability parameter
106: $\delta$ on $x$. It is unlikely that these conditions are fulfilled in
107: La$_{2-x}$Sr$_x$CuO$_4$.\cite{Ino} Besides, the applicability of the
108: picture of itinerant electrons for underdoped cuprates casts doubts. As
109: for the second notion, it should be noted that in the elastic neutron
110: scattering the charge-density wave connected with stripes is observed
111: only in crystals with the low-temperature tetragonal or the
112: low-temperature less-orthorhombic phases (La$_{2-x}$Ba$_x$CuO$_4$
113: \linebreak[4] and La$_{2-y-x}$Nd$_y$Sr$_x$CuO$_4$) and is not observed
114: in the crystal La$_{2-x}$Sr$_x$CuO$_4$ in the low-temperature
115: orthorhombic phase.\cite{Kimura} At the same time the magnetic
116: incommensurability is similar in these phases. It can be supposed that
117: the magnetic incommensurability is the cause rather than the effect of
118: stripes which are formed with an assistance of phonons.
119:
120: In the present work the general formula for the magnetic susceptibility
121: derived in the projection operator formalism\cite{Mori} is used. For
122: the description of spin excitations in the doped antiferromagnet the
123: $t$-$J$ model of a Cu-O plane is employed. In this approach the
124: mentioned peculiarities of the magnetic properties of cuprates are
125: reproduced including the proper frequency and momentum location of the
126: susceptibility maxima. The incommensurability for $\omega>\omega_r$ is
127: connected with the dispersion of spin
128: excitations.\cite{Barzykin,Sherman02} The incommensurability for lower
129: frequencies is related to the dip in the spin-excitation damping at
130: ${\bf Q}$. For small $x$ the dip appears due to the nesting of the hole
131: pockets around $(\pm\frac{1}{4},\pm\frac{1}{4})$ forming the Fermi
132: surface.\cite{Sherman04} For moderate doping this dip stems from the
133: weakness of the interaction between the spin excitations and holes near
134: the hot spots -- the intersection points of the Fermi surface and the
135: boundary of the magnetic Brillouin zone. Such a weak interaction
136: follows from the fact that due to a short-range interaction between
137: holes and spins a decaying site spin excitation creates a fermion pair
138: with components residing on the same and neighbor sites. The
139: spin-excitation damping was found to depend strongly on details of the
140: hole dispersion, bandwidth and damping, so that the change in these
141: characteristics leads to the conversion of well-defined spin
142: excitations to overdamped ones. As this takes place, the frequency
143: dependence of the susceptibility at {\bf Q} is transformed from a
144: pronounced maximum\cite{Bourges} at $\omega_r$ which is inherent in
145: underdoped YBa$_2$Cu$_3$O$_{7-y}$ to a broad low-frequency feature
146: characteristic for lanthanum cuprates.\cite{Aeppli} The increased
147: spin-excitation damping has no marked effect on the low-frequency
148: incommensurability, however for $\omega>\omega_r$ the incommensurate
149: peaks are shifted to {\bf Q} and form a broad maximum. Such form of the
150: momentum dependence of the susceptibility is also observed
151: experimentally.\cite{Endoh}
152:
153: \section{Main formulas}
154: The imaginary part of the magnetic susceptibility which determines the
155: cross-section of the magnetic scattering\cite{Kastner} is calculated
156: from the relations $\chi''({\bf k}\omega)=-4\mu_B^2\Im\langle\langle
157: s^z_{\bf k}|s^z_{\bf -k}\rangle\rangle_\omega$, $\langle\langle
158: s^z_{\bf k}|s^z_{\bf -k}\rangle\rangle_\omega=\omega((s^z_{\bf
159: k}|s^z_{\bf -k}))_\omega-(s^z_{\bf k},s^z_{\bf -k})$. Here $\mu_B$ is
160: the Bohr magneton, $\langle\langle s^z_{\bf k}|s^z_{\bf
161: -k}\rangle\rangle_\omega$ and $((s^z_{\bf k}|s^z_{\bf -k}))_\omega$ are
162: the Fourier transforms of the retarded Green's and Kubo's relaxation
163: functions, $$\langle\langle s^z_{\bf k}|s^z_{\bf
164: -k}\rangle\rangle_t=-i\theta(t)\langle[s^z_{\bf k}(t),s^z_{\bf
165: -k}]\rangle,\quad ((s^z_{\bf k}|s^z_{\bf -k}))_t=\theta(t)\int_t^\infty
166: dt'\langle[s^z_{\bf k}(t'),s^z_{\bf -k}]\rangle,$$ $s^z_{\bf
167: k}=N^{-1/2}\sum_{\bf n}e^{-i{\bf kn}}s^z_{\bf n}$ with the number of
168: sites $N$ and the $z$ component of the spin $s^z_{\bf n}$ on the
169: lattice site {\bf n}, for arbitrary operators $A$ and $B$
170: $(A,B)=i\int_0^\infty dt\langle[A(t),B]\rangle$ where the angular
171: brackets denote the statistical averaging and $A(t)=e^{iHt}Ae^{-iHt}$
172: with the Hamiltonian $H$.
173:
174: Using the projection operator technique\cite{Mori} the relaxation
175: function $((s^z_{\bf k}|s^z_{\bf -k}))_\omega$ can be calculated from
176: the recursive relations
177: \begin{equation}\label{cf}
178: R_n(\omega)=[\omega-E_n-F_nR_{n+1}(\omega)]^{-1}, \quad n=0,1,2,\ldots
179: \end{equation}
180: where $R_n(\omega)$ is the Laplace transform of
181: $R_n(t)=(A_{nt},A_n^\dagger)(A_{n},A_n^\dagger)^{-1}$, the time
182: dependence in $A_{nt}$ is determined by the relation
183: $$i\frac{d}{dt}A_{nt}=\prod_{k=0}^{n-1}(1-P_k)[A_{nt},H], \quad
184: A_{n,t=0}=A_n$$ with the projection operators $P_k$ defined as
185: $P_kB=(B,A_k^\dagger) (A_k,A_k^\dagger)^{-1}A_k$. The parameters $E_n$
186: and $F_n$ in relations (\ref{cf}) and operators $A_n$ in the functions
187: $R_n(t)$ are calculated recursively using the procedure\cite{Sherman02}
188: \begin{eqnarray}
189: &&[A_n,H]=E_nA_n+A_{n+1}+F_{n-1}A_{n-1},\quad
190: E_n=([A_n,H],A_n^\dagger)(A_n,A_n^\dagger)^{-1}, \nonumber\\
191: &&\label{Lanczos}\\
192: &&F_{n-1}=(A_n,A_n^\dagger)(A_{n-1},A_{n-1}^\dagger)^{-1},\quad
193: F_{-1}=0. \nonumber
194: \end{eqnarray}
195: As the starting operator for this procedure we set $A_0=s^z_{\bf k}$.
196: In this case $((s^z_{\bf k}|s^z_{\bf -k}))_\omega=(s^z_{\bf k},s^z_{\bf
197: -k})R_0(\omega)$ where $R_0(\omega)$ is calculated from Eq.~(\ref{cf}).
198:
199: To describe the spin excitations of Cu-O planes which determine the
200: magnetic properties of cuprates\cite{Kastner} the $t$-$J$
201: model\cite{Izyumov} is used. The model was shown to describe correctly
202: the low-energy part of the spectrum of the realistic extended Hubbard
203: model.\cite{Zhang,Jefferson} The Hamiltonian of the two-dimensional
204: $t$-$J$ model reads
205: \begin{equation}\label{hamiltonian}
206: H=\sum_{\bf nm\sigma}t_{\bf nm}a^\dagger_{\bf n\sigma}a_{\bf
207: m\sigma}+\frac{1}{2}\sum_{\bf nm}J_{\bf nm}{\bf s_n s_m},
208: \end{equation}
209: where $a_{\bf n\sigma}=|{\bf n}\sigma\rangle\langle{\bf n}0|$ is the
210: hole annihilation operator, {\bf n} and {\bf m} label sites of the
211: square lattice, $\sigma=\pm 1$ is the spin projection, $J_{\bf nm}$ and
212: $t_{\bf nm}$ are the exchange and hopping constants, respectively,
213: $|{\bf n}\sigma\rangle$ and $|{\bf n}0\rangle$ are site states
214: corresponding to the absence and presence of a hole on the site. These
215: states may be considered as linear combinations of the products of the
216: $3d_{x^2-y^2}$ copper and $2p_\sigma$ oxygen orbitals of the extended
217: Hubbard model.\cite{Jefferson} The spin-$\frac{1}{2}$ operators can be
218: written as $s^z_{\bf n}=\frac{1}{2}\sum_\sigma\sigma|{\bf
219: n}\sigma\rangle\langle{\bf n}\sigma|$ and $s^\sigma_{\bf n}=|{\bf
220: n}\sigma\rangle\langle{\bf n},-\sigma|$.
221:
222: With Hamiltonian (\ref{hamiltonian}) and $A_0=s^z_{\bf k}$ we find from
223: Eq.~(\ref{Lanczos})
224: \begin{eqnarray}
225: &&E_0(s^z_{\bf k},s^z_{\bf -k})=(i\dot{s}^z_{\bf k},s^z_{\bf
226: -k})=\langle[s^z_{\bf k},s^z_{\bf -k}]\rangle=0, \nonumber\\
227: &&A_1=A_1^s+A_1^h=\frac{1}{2\sqrt{N}}\sum_{\bf l}e^{-i{\bf
228: kl}}\bigg[\sum_{\bf nm}J_{\bf mn}(\delta_{\bf ln}-\delta_{\bf
229: lm})s_{\bf n}^{+1}s_{\bf m}^{-1}\label{fstep}\\
230: &&\hspace{1.5em}+\sum_{\bf nm\sigma}t_{\bf mn}(\delta_{\bf
231: lm}-\delta_{\bf ln})\sigma a_{\bf n\sigma}^\dagger a_{\bf
232: m\sigma}\bigg],\nonumber
233: \end{eqnarray}
234: where $i\dot{s}^z_{\bf k}=[s^z_{\bf k},H]$. To obtain a tractable form
235: for the spin-excitation damping it is convenient to approximate the
236: quantity $(A_{1t},A_1^\dagger)$ in the $R_1(\omega)$ by the sum
237: $$(A_1^h(t),A_1^{h\dagger})+(A_{1t}^s,A_1^{s\dagger})$$ where the first
238: term describes the influence of holes on the spin excitations.
239: Continuing calculations (\ref{Lanczos}) with the second term of the sum
240: we get
241: \begin{equation}\label{sstep}
242: F_0=4JC_1(\gamma_{\bf k}-1)(s^z_{\bf k},s^z_{\bf -k})^{-1}, \quad
243: E_1=0,
244: \end{equation}
245: where only the nearest neighbor interaction between spins was taken
246: into account, $J_{\bf nm}=J\sum_{\bf a}\delta_{\bf n,m+a}$, the four
247: vectors {\bf a} connect the nearest neighbor sites, $C_1=\langle
248: s^{+1}_{\bf n}s^{-1}_{\bf n+a}\rangle$ is the spin correlation on
249: neighbor sites and $\gamma_{\bf k}=\frac{1}{2}[\cos(k_x)+\cos(k_y)]$.
250:
251: To calculate the quantity $(s^z_{\bf k},s^z_{\bf -k})$ let us notice
252: that in accord with procedure (\ref{Lanczos}) the interruption of
253: calculations at this stage actually means that $(A_2,A_2^\dagger)$ in
254: the parameter $F_1$ is set to zero. Here $A_2=i^2\ddot{s}^z_{\bf
255: k}-F_0s^z_{\bf k}$. The substitution of this expression into
256: $(A_2,A_2^\dagger)=0$ gives an equation for $(s^z_{\bf k},s^z_{\bf
257: -k})$. Using the decoupling in calculating $i^2\ddot{s}^z_{\bf k}$ we
258: get\cite{Sherman02}
259: \begin{equation}\label{inpr}
260: (s^z_{\bf k},s^z_{\bf -k})^{-1}=4\alpha J(\Delta+1+\gamma_{\bf k}),
261: \end{equation}
262: where $\alpha\sim 1$ is the decoupling parameter.\cite{Kondo} The
263: meaning of the parameter $\Delta$, which can be expressed in terms of
264: spin correlations, will be discussed later.
265:
266: Using the decoupling in $(A_1^h(t),A_1^{h\dagger})$ we find from the
267: above formulas
268: \begin{equation}\label{chi}
269: \chi''({\bf k}\omega)=-\frac{4\mu^2_B\omega\Im R({\bf k}\omega)
270: }{[\omega^2-\omega f_{\bf k}\Re R({\bf k}\omega)-\omega^2_{\bf
271: k}]^2+[\omega f_{\bf k}\Im R({\bf k}\omega)]^2},
272: \end{equation}
273: where
274: \begin{eqnarray}
275: &&f_{\bf k}^{-1}=4J|C_1|(1-\gamma_{\bf k}), \quad \omega_{\bf
276: k}^2=16J^2\alpha|C_1|(1-\gamma_{\bf k})(\Delta+1+
277: \gamma_{\bf k}), \nonumber\\
278: &&\Im R({\bf k}\omega)=\frac{8\pi\omega^2_{\bf k}}{N}
279: \sum_{\bf k'}g_{\bf kk'}^2\int_{-\infty}^\infty d\omega' A({\bf
280: k'}\omega') \label{chitJ}\\
281: &&\hspace{4em}\times A({\bf k+k'},\omega+\omega')
282: \frac{n_F(\omega+\omega')-n_F(\omega')}{\omega},\nonumber
283: \end{eqnarray}
284: the interaction constant $g_{\bf kk'}=t_{\bf k'}-t_{\bf k+k'}$ with
285: $t_{\bf k}=\sum_{\bf n}e^{i{\bf k(n-m)}}t_{\bf nm}$,
286: $n_F(\omega)=[\exp(\omega/T)+1]^{-1}$, $T$ is the temperature and
287: $A({\bf k}\omega)$ is the hole spectral function. Since the incoherent
288: part of the spectral function is unlikely to lead to sharp structure in
289: $\chi''$, only the coherent part of $A({\bf k}\omega)$ is taken into
290: account in this work,
291: \begin{equation}\label{hsf}
292: A({\bf k}\omega)=\frac{\eta/\pi}{(\omega-\varepsilon_{\bf
293: k}+\mu)^2+\eta^2}.
294: \end{equation}
295: Here $\mu$ is the chemical potential, $\eta$ is the artificial
296: broadening, and $\varepsilon_{\bf k}$ is the hole dispersion. The real
297: part of $R({\bf k}\omega)$ can be calculated from the imaginary part
298: $\Im R({\bf k}\omega)$ and the Kramers-Kronig relation.
299:
300: Notice that the interaction constant $g_{\bf kk'}$ is determined by the
301: Fourier transform of the hole hopping constant $t_{\bf nm}$. If the
302: hopping to the nearest and next nearest sites is taken into account the
303: constant acquires the form
304: \begin{equation}\label{intcon}
305: g_{\bf kk'}=t(\gamma_{\bf k'}-\gamma_{\bf k+k'})+t'(\gamma'_{\bf
306: k'}-\gamma'_{\bf k+k'}),
307: \end{equation}
308: where $\gamma'_{\bf k}=\cos(k_x)\cos(k_y)$. This constant vanishes for
309: ${\bf k=Q}$ when the vector ${\bf k'}$ is located at the boundary of
310: the magnetic Brillouin zone. In other words, fermions near hot spots
311: interact weakly with spin excitations. This is connected with the
312: short-range character of the interaction described by constant
313: (\ref{intcon}) -- the decaying spin excitation on the site {\bf n}
314: creates the fermion pair on the same and neighbor sites which is
315: reflected in the above form of the interaction constant.
316:
317: As the quantity $\omega f_{\bf k}\Re R({\bf k}\omega)$ influences the
318: frequency of spin excitations only near {\bf Q}, it is convenient to
319: incorporate it in $\omega_{\bf k}$. This modifies the parameter
320: $\Delta>0$ which, as seen from Eqs.~(\ref{chi}) and (\ref{chitJ}),
321: describes a gap in the spin-excitation spectrum at the
322: antiferromagnetic wave vector {\bf Q}. The most exact way to determine
323: this parameter is to use the constraint of zero site magnetization
324: \begin{equation}\label{zsm}
325: \langle s^z_{\bf n}\rangle=\frac{1}{2}(1-x)-\langle s^{-1}_{\bf
326: n}s^{+1}_{\bf n}\rangle=0,
327: \end{equation}
328: which has to be fulfilled in the short-range antiferromagnetic
329: ordering. It can be shown that $\Delta\propto\xi^{-2}$ where $\xi$ is
330: the correlation length of the short-range order.\cite{Sherman03} Thus,
331: in this case the frequency of spin excitations at {\bf Q} is nonzero,
332: in contrast to the classical antiferromagnetic magnons. As follows from
333: Eq.~(\ref{chitJ}), the dispersion of spin excitations has a local
334: minimum at {\bf Q} and can be approximated as
335: \begin{equation}\label{mdisp}
336: \omega_{\bf k}=[\omega^2_{\bf Q}+c^2({\bf k-Q})^2]^{1/2}
337: \end{equation}
338: near this momentum. In Fig.~\ref{Fig_i} the calculated dispersion of
339: spin excitations\cite{Sherman02} near {\bf Q} is compared with the
340: dispersion of the maximum in the susceptibility in
341: YBa$_2$Cu$_3$O$_{6.5}$.\cite{Bourges} This is a bilayer crystal and the
342: symmetry allows one to divide the susceptibility into odd and even
343: parts. For the antiferromagnetic intrabilayer coupling the dispersion
344: of the maximum in the odd part can be compared with our calculations
345: carried out for a single layer. This comparison demonstrates that the
346: observed dispersion of the susceptibility maxima above $\omega_{\bf
347: Q}$, which we identify with the resonance frequency $\omega_r$, is
348: closely related to the dispersion of spin excitations.
349: \begin{figure}[t]
350: \centerline{\includegraphics[width=6.5cm]{Fig1.eps}} \caption{The
351: dispersion of spin excitations calculated in a 20$\times$20 lattice for
352: $x=0.06$ and $T=17$~K (filled squares).\protect\cite{Sherman02} The
353: solid line is the fit of Eq.~(\protect\ref{mdisp}) to these data. Open
354: squares are the dispersion of the peak in the odd susceptibility in
355: YBa$_2$Cu$_3$O$_{6.5}$ ($x\approx 0.075$,
356: Ref.~\protect\refcite{Tallon}) at
357: $T=5$~K.\protect\cite{Bourges}}\label{Fig_i}
358: \end{figure}
359:
360: Previous calculations \cite{Sherman03} show that the variation of the
361: temperature in the range from 0 to approximately 100~K leads only to
362: some broadening of maxima in the susceptibility. Therefore to simplify
363: calculations and use larger lattices, which is necessary to resolve the
364: low-frequency incommensurability, let us set $T=0$. In calculating $\Im
365: R({\bf k}\omega)$ the integration over frequencies in Eq.~(\ref{chitJ})
366: is the most time-consuming operation. For $T=0$ and $\omega\geq 0$ this
367: integral reduces to
368: $$\int_{-\omega}^0d\omega'\,A({\bf k'}\omega')
369: A({\bf k+k'},\omega+\omega')$$ and is easily integrated for the
370: spectral function (\ref{hsf}). The same result is obtained for
371: $\omega<0$, since $\Im R({\bf k}\omega)$ is an even function of
372: frequency. Notice that for $\eta\ll\omega$ the states with energies
373: \begin{equation}\label{energies}
374: -\omega<\varepsilon_{\bf k'}-\mu<0\;\;{\rm and}\;\; 0<\varepsilon_{\bf
375: k+k'}-\mu<\omega
376: \end{equation}
377: make the main contribution to this integral.
378:
379: In the following, we use the values of $C_1$, $\Delta$ and $\alpha$
380: calculated self-con\-sis\-ten\-t\-ly in the $t$-$J$ model on a
381: 20$\times$20 lattice for the range of hole concentrations $0\leq
382: x\lesssim 0.16$.\cite{Sherman03} The calculations were carried out for
383: the parameters $t=0.5$~eV and $J=0.1$~eV corresponding to hole-doped
384: cuprates.\cite{McMahan} In Eq.~(\ref{hsf}), for $\varepsilon_{\bf k}$
385: we apply the hole dispersion
386: \begin{eqnarray}
387: \varepsilon_{\bf k}&=&-0.0879+0.5547\gamma_{\bf k}-0.1327\gamma'_{\bf
388: k}-0.0132\gamma_{2\bf k}\nonumber\\
389: &+&0.09245[\cos(2k_x)\cos(k_y)+\cos(k_x)\cos(2k_y)]
390: -0.0265\gamma'_{2\bf k} \label{disp}
391: \end{eqnarray}
392: proposed from the analysis of photoemission data in
393: Bi$_2$Sr$_2$CaCu$_2$O$_8$.\cite{Norman} Here the coefficients are in
394: electronvolts. Results which are analogous to those discussed in the
395: next section can also be obtained with other model dispersions
396: suggested for cuprates.\cite{Liu,Brinckmann,Norman} Results do not
397: change qualitatively either with the variation of the parameter $t'$ in
398: Eq.~(\ref{intcon}) in the range from 0 to $-0.4t$ (notice that
399: parameters $t$ and $t'$ of the hole hopping part of Hamiltonian
400: (\ref{hamiltonian}) are only indirectly connected with the coefficients
401: in Eq.~(\ref{disp}), since to a great extent the hole dispersion is
402: shaped by the interaction between holes and spin
403: excitations\cite{Sherman04b}).
404:
405: \section{Magnetic susceptibility}
406: \begin{figure}[t]
407: \centerline{\includegraphics[width=6.5cm]{Fig2.eps}} \caption{The
408: momentum dependence of $\chi''({\bf k}\omega)$ for $T=0$, $x \approx
409: 0.12$, $\mu=-40$~meV, $t'=-0.2t$ and $\omega=70$~meV, $\eta=30$~meV
410: (a), $\omega=35$~meV, $\eta=15$~meV (b), $\omega=2$~meV, $\eta=1.5$~meV
411: (c). Calculations were carried out in a 1200$\times$1200 lattice. The
412: solid lines correspond to the scans along the edge of the Brillouin
413: zone, ${\bf k}=(\kappa,\frac{1}{2})$; the dashed lines are for the zone
414: diagonal, ${\bf k}=(\kappa,\kappa)$.} \label{Fig_ii}
415: \end{figure}
416: The momentum dependence of $\chi''({\bf k}\omega)$ calculated with the
417: above equations for three energy transfers are shown in
418: Fig.~\ref{Fig_ii}. The contour plots of the susceptibility for the same
419: parameters are demonstrated in Fig.~\ref{Fig_iii}.
420: \begin{figure}[t]
421: \centerline{\includegraphics[width=5cm]{Fig3a.eps}
422: \hspace{0.5cm}\includegraphics[width=5cm]{Fig3b.eps}}
423: \vspace*{0.5cm}\centerline{\includegraphics[width=5cm]{Fig3c.eps}}
424: \caption{The contour plots of $\chi''({\bf k}\omega)$. Parameters in
425: parts (a), (b) and (c) are the same as in the respective parts of
426: Fig.~\protect\ref{Fig_ii}.} \label{Fig_iii}
427: \end{figure}
428: As seen from these figures, there are three frequency regions with
429: different shapes of the momentum dependence of $\chi''({\bf k}\omega)$.
430: The first region is the vicinity of the frequency $\omega_{\bf Q}$ of
431: the gap in the dispersion of spin excitations at the antiferromagnetic
432: wave vector {\bf Q}. For the parameters of Fig.~\ref{Fig_ii}
433: $\omega_{\bf Q}\approx 37$~meV. In this region the susceptibility is
434: peaked at the wave vector {\bf Q}. For smaller and larger frequencies
435: the magnetic response is incommensurate. The dispersion of maxima in
436: $\chi''({\bf k}\omega)$ for scans along the edge and the diagonal of
437: the Brillouin zone and their full widths at half maximum (FWHM) are
438: shown in Fig.~\ref{Fig_iv}. Analogous dispersion was obtained in
439: Ref.~\refcite{Norman} in the itinerant-carrier approach for the
440: superconducting state.
441: \begin{figure}[t]
442: \centerline{\includegraphics[width=6.5cm]{Fig4.eps}} \caption{The
443: dispersion of maxima in $\chi''({\bf k}\omega)$ for scans along the
444: edge [${\bf k}=(\kappa,\frac{1}{2})$, solid lines] and the diagonal
445: [${\bf k}=(\kappa,\kappa)$, dashed lines] of the Brillouin zone. The
446: dispersion along the diagonal is shown only in the frequency range in
447: which these maxima are more intensive than those along the edge.
448: Parameters are the same as in Fig.~\protect\ref{Fig_ii}. Horizontal
449: bars are FWHM for maxima along the edge of the Brillouin zone.}
450: \label{Fig_iv}
451: \end{figure}
452:
453: The momentum dependencies of the susceptibility which are similar to
454: those shown in Fig.~\ref{Fig_ii} and \ref{Fig_iii} were observed in
455: yttrium and lanthanum cuprates.\cite{Mason93,Dai,Tranquada} The
456: dispersion of the peaks in $\chi''({\bf k}\omega)$ which is similar to
457: that shown in Fig.~\ref{Fig_iv} was derived from experimental data in
458: YBa$_2$Cu$_3$O$_{7-y}$ and La$_{2-x}$Ba$_x$CuO$_4$ in
459: Refs.~\refcite{Dai,Tranquada}. As seen from Fig.~\ref{Fig_ii}, for
460: frequencies $\omega<\omega_{\bf Q}$ the susceptibility is peaked at the
461: wave vectors ${\bf k}=(\frac{1}{2},\frac{1}{2}\pm\delta)$ and
462: $(\frac{1}{2}\pm\delta,\frac{1}{2})$, while for $\omega>\omega_{\bf Q}$
463: the maxima are located at
464: $(\frac{1}{2}\pm\delta,\frac{1}{2}\pm\delta)$,
465: $(\frac{1}{2}\pm\delta,\frac{1}{2}\mp\delta)$ for the parameters used.
466: This result is also in agreement with experimental
467: observations.\cite{Dai,Hayden04,Tranquada} Notice, however, that for
468: $\omega>\omega_{\bf Q}$ the positions of maxima in the momentum space
469: may vary with parameters.
470:
471: To understand the above results one should notice that Eq.~(\ref{chi})
472: contains the resonance denominator which will dominate in the momentum
473: dependence for $\omega \geq\omega_{\bf Q}$ if the spin excitations are
474: not overdamped. Parameters of Fig.~\ref{Fig_ii} correspond to this
475: case. For $\omega \geq\omega_{\bf Q}$ the equation $\omega=\omega_{\bf
476: k}$ determines the positions of the maxima in $\chi''({\bf k}\omega)$
477: which are somewhat shifted by the momentum dependence of the
478: spin-excitation damping $f_{\bf k}\Im R({\bf k}\omega)$. Using
479: Eq.~(\ref{mdisp}) we find that the maxima in $\chi''({\bf k}\omega)$
480: are positioned near a circle centered at {\bf Q} with the radius
481: $c^{-1}(\omega^2-\omega^2_{\bf Q})^{1/2}$.\cite{Barzykin,Sherman02}
482:
483: In the region $\omega<\omega_{\bf Q}$ the nature of the
484: incommensurability is completely different. It is most easily seen in
485: the limit of small frequencies when Eq.~(\ref{chi}) reduces to
486: \begin{equation}\label{lowfreq}
487: \chi''({\bf k}\omega)\approx -4\mu_B^2\omega\frac{\Im R({\bf
488: k}\omega)}{\omega_{\bf k}^4}.
489: \end{equation}
490: As seen in Fig.~\ref{Fig_i}, $\omega^{-4}_{\bf k}$ is a decreasing
491: function of the difference ${\bf k-Q}$ which acts in favor of a
492: commensurate peak. However, if $\Im R({\bf k}\omega)$ in the numerator
493: of Eq.~(\ref{lowfreq}) has a pronounced dip at {\bf Q} the commensurate
494: peak splits into several incommensurate maxima. For hole concentrations
495: $x\lesssim 0.06$, when the Fermi surface consists of four ellipses
496: centered at
497: $(\pm\frac{1}{4},\pm\frac{1}{4})$,\cite{Izyumov,Sherman04b,Damascelli}
498: $\Im R({\bf k}\omega)$ has a dip at {\bf Q} due to the nesting of the
499: ellipses with this wave vector.\cite{Sherman04} For larger $x$ the
500: mechanism of the dip formation is the following.
501: \begin{figure}[t]
502: \centerline{\includegraphics[width=5cm]{Fig5.eps}} \caption{The Fermi
503: surface for dispersion (\protect\ref{disp}) and $\mu=-40$~meV (solid
504: lines). Dashed lines show the boundary of the magnetic Brillouin zone,
505: gray circles are the hot spots, the dotted arrow is the
506: antiferromagnetic wave vector.} \label{Fig_v}
507: \end{figure}
508: As follows from Eq.~(\ref{energies}), for ${\bf k=Q}$ and small
509: frequencies $\omega$ hole states which make the main contribution to
510: the spin-excitation damping (\ref{chitJ}) are located near the hot
511: spots (see Fig.~\ref{Fig_v}). For these wave vectors the interaction
512: constant $g_{\bf Qk'}$, Eq.~(\ref{intcon}), is small which leads to the
513: smallness of $\Im R({\bf Q}\omega)$. With the wave vector moving away
514: from {\bf Q} momenta of states contributing to the spin-excitation
515: damping recede from the hot spots, the interaction constant grows, and
516: with it the spin-excitation damping. Thus, the damping has a dip at
517: {\bf Q} which leads to the low-frequency incommensurability shown in
518: Fig.~\ref{Fig_ii}c.
519:
520: Let us compare the discussed mechanisms of the low- and high-frequency
521: incommensurability with those based on the picture of itinerant
522: electrons and the random phase approximation. In this latter approach
523: incommensurability arises due to maxima in the noninteracting
524: susceptibility $\chi_0$ described by the fermion
525: bubbles.\cite{Liu,Brinckmann,Norman} For low frequencies such a maximum
526: appears if the Fermi surface has nesting. As mentioned, this mechanism
527: imposes rather stringent requirements on the electron energy spectrum,
528: because to reproduce known experimental results the nesting has to
529: persist in the wide range of hole concentrations and the nesting
530: momentum has to change in a specific manner with doping. In
531: Ref.~\refcite{Brinckmann} the notion was proposed that the nesting for
532: constant-energy contours can appear in the superconducting state. The
533: application of this idea also requires fine tuning of
534: parameters.\cite{Norman} Besides, this mechanism cannot explain the
535: incommensurability above $T_c$ which is observed both in lanthanum and
536: yttrium cuprates.\cite{Mason93,Dai,Tranquada} In the approach discussed
537: in this paper requirements on the Fermi surface are substantially
538: relaxed: the Fermi surface has to intersect with the boundary of the
539: magnetic Brillouin zone, i.e. the Fermi surface has to contain hot
540: spots where the interaction constant $g_{\bf kk'}$ is small which leads
541: to the dip in ${\Im R}$ at {\bf Q}. According to the available
542: photoemission data\cite{Ino,Damascelli} Fermi surfaces of this type,
543: which resemble that shown in Fig.~\ref{Fig_v}, are indeed observed in
544: underdoped cuprates. Apart from Eq.~(\ref{disp}) we used some other
545: model dispersions present in the literature\cite{Liu,Brinckmann,Norman}
546: and obtained results which are qualitatively similar to those shown in
547: Fig.~\ref{Fig_ii} -- \ref{Fig_iv}. Hence the discussed mechanism is
548: robust with respect to changes of the hole energy spectrum, provided
549: that the Fermi surface contains hot spots. The mechanism is equally
550: applicable for the superconducting state, since the same interaction
551: constant $g_{\bf kk'}$ enters into the expression for the
552: susceptibility in this state.\cite{Sherman02} We suppose that in
553: certain conditions the magnetic incommensurability may trigger the
554: corrugation of Cu-O planes and formation of stripes.
555:
556: The dependence of the incommensurability parameter $\delta$ on $x$ for
557: the low frequencies is shown in Fig.~\ref{Fig_vi}. In agreement with
558: experiment (see the inset in Fig.~\ref{Fig_vi}) $\delta$ grows nearly
559: linearly with $x$ up to $x\lesssim 0.12$ and then saturates. In this
560: calculation dispersion~(\ref{disp}) was used for the entire range of
561: hole concentrations $0.06\leq x\leq 0.16$. This is not quite correct,
562: since the photoemission data\cite{Damascelli} and
563: calculations\cite{Sherman04b} demonstrate that the dispersion changes
564: substantially with doping. However, we suppose that the growth of the
565: spin-excitation frequency with doping in accord with the relations
566: $\omega_{\bf Q}\propto\xi^{-1}\propto x^{1/2}$,\cite{Sherman03} which
567: leads to a weaker momentum dependence in Eq.~(\ref{mdisp}) and in the
568: denominator of Eq.~(\ref{lowfreq}), is more essential for the
569: dependence $\delta(x)$ than the variation of the hole dispersion.
570: \begin{figure}[t]
571: \centerline{\includegraphics[width=6.5cm]{Fig6.eps}} \caption{The
572: incommensurability parameter $\delta$ vs.\ $x$ for $\omega=2$~meV. The
573: value of $\delta$ for $x=0.043$ was taken from
574: Ref.~\protect\refcite{Sherman04}. Inset: experimental
575: data\protect\cite{Yamada} for La$_{2-x}$Sr$_x$CuO$_4$. Connecting lines
576: are a guide to the eye.} \label{Fig_vi}
577: \end{figure}
578:
579: We found also that the low-frequency incommensurability disappears when
580: the hole damping $\eta$ is greater than $\omega$. Besides, this
581: incommensurability disappears if the chemical potential $\mu$
582: approaches the extended van Hove singularities at $(0,\frac{1}{2})$,
583: $(\frac{1}{2},0)$. In this case for ${\bf k=Q}$ the entire region of
584: these singularities in which the interaction constant $g_{\bf Qk'}$ is
585: not small contributes to the spin-excitation damping. As a result the
586: dip in the damping becomes shallower or disappears completely. In the
587: $t$-$J$ model, $\mu$ approaches the van Hove singularities for
588: $x\approx 0.18$ for the parameters of hole-doped
589: cuprates.\cite{Sherman04b} This may be the reason of the disappearance
590: of the incommensurability in overdoped cuprates.\cite{Yamada} The
591: low-frequency incommensurability disappears also in lattices with size
592: less than 30$\times$30 sites. The use of a smaller lattice and an
593: increased artificial broadening needed for stabilizing the iteration
594: procedure accounts for the lack of low-frequency incommensurability in
595: the self-consistent calculations of Ref.~\refcite{Sherman03}.
596:
597: \begin{figure}[t]
598: \centerline{\includegraphics[width=6.5cm]{Fig7.eps}} \caption{The
599: frequency dependence of $\chi''$. The solid lines are our results for
600: $T=0$, $x \approx 0.12$, $\mu=-40$~meV, $\eta=3.5$~meV, ${\bf k=Q}$ (a)
601: and for ${\bf k}=(0.42,0.5)$ with the hole dispersion scaled by the
602: factor 0.4 (b, see text). Squares are the odd susceptibility
603: measured\protect\cite{Bourges} in the normal-state
604: YBa$_2$Cu$_3$O$_{6.83}$ ($x\approx 0.14$, \protect\refcite{Tallon}) at
605: $T=100$~K and ${\bf k=Q}$ (a) and the susceptibility in
606: La$_{1.86}$Sr$_{0.14}$CuO$_4$ for $T=35$~K at the incommensurate peak
607: \protect\cite{Aeppli} (b).} \label{Fig_vii}
608: \end{figure}
609: As mentioned above, for the parameters chosen spin excitations are not
610: overdamped near the antiferromagnetic wave vector. As a consequence,
611: the frequency dependence of $\chi''({\bf Q}\omega)$ has a pronounced
612: maximum at $\omega\approx\omega_{\bf Q}$ which resembles the
613: susceptibility observed in the underdoped YBa$_2$Cu$_3$O$_{7-y}$ in the
614: superconducting and normal states. As seen in Fig.~\ref{Fig_vii}a, the
615: experimental width of the maximum is approximately twice as large as
616: the calculated one. Partly this is connected with the difference in
617: temperatures for the two sets of the data. Besides, the width and shape
618: of the frequency dependence of $\chi''$ vary essentially with the
619: change of the hole dispersion and damping. Figure~\ref{Fig_vii}b
620: demonstrates that, in particular, the decrease of the hole bandwidth
621: leads to a substantial growth of the spin-excitation damping which in
622: its turn results in the overdamping of spin excitations. For this
623: figure the calculated results were obtained with dispersion
624: (\ref{disp}) scaled by the factor 0.4. The calculations were carried
625: out for ${\bf k}=(0.42,0.5)$ which corresponds to the wave vector of
626: the low-frequency peak in Fig.~\ref{Fig_ii}c. The overdamping of spin
627: excitations leads to the red shift of the maximum in $\chi''(\omega)$.
628: Its position is no longer connected with the frequency of spin
629: excitations. The similar frequency dependence of $\chi''$ without a
630: well-defined peak of spin excitations is observed in
631: La$_{2-x}$Sr$_x$CuO$_4$.\cite{Aeppli} Thus, we suppose that the
632: observed dissimilarity of the frequency dependencies of the
633: susceptibility in lanthanum and yttrium cuprates is connected with the
634: different values of the spin-excitation damping.
635:
636: The increased spin-excitation damping obtained above with the scaled
637: hole dispersion does not affect markedly the low-frequency
638: incommensurability, however, for the frequencies $\omega_{\bf
639: Q}\geq\omega\geq 150$~meV we found only broad commensurate maxima
640: instead of the incommensurate peaks shown in Fig.~\ref{Fig_ii}a. Such
641: spectra are also observed experimentally.\cite{Endoh}
642:
643: Our consideration was restricted to the normal state. In the considered
644: approach the opening of the superconducting gap suppresses the
645: spin-excitation damping for frequencies below the gap and increases the
646: damping above it. The respective redistribution of the intensity takes
647: place also in the
648: susceptibility.\cite{Mason93,Aeppli,Barzykin,Sherman02} As mentioned,
649: for the momentum dependence the same mechanisms which lead to the
650: incommensurate magnetic response in the normal state operate also in
651: the superconducting state. In this state the suppressed spin-excitation
652: damping produces sharper peaks in the susceptibility, however, their
653: location in the momentum space is approximately the same as in the
654: normal state.\cite{Dai,Kimura}
655:
656: \section{Concluding remarks}
657: Mori's projection operator formalism and the $t$-$J$ model of Cu-O
658: planes were used for the interpretation of the magnetic susceptibility
659: in normal-state cu\-p\-ra\-te perovskites. It was shown that the
660: calculated momentum and frequency dependencies of the imaginary part of
661: the susceptibility $\chi''$, the dispersion and location of maxima in
662: it and the concentration dependence of the incommensurability parameter
663: are similar to those observed in lanthanum and yttrium cuprates. The
664: dispersion of the maxima in $\chi''$ resembles two parabolas with
665: upward- and downward-directed branches which converge at the
666: antiferromagnetic wave vector {\bf Q} and at the respective frequency
667: of spin excitations $\omega_{\bf Q}$. This frequency corresponds to a
668: local minimum in the dispersion of spin excitations and its value is
669: connected with the correlation length of the short-range
670: antiferromagnetic order. We relate the upper parabola to the
671: spin-excitation dispersion. The incommensurability connected with the
672: lower parabola is related to the dip in the spin-excitation damping at
673: {\bf Q}. For moderate doping the dip arises due to the smallness of the
674: interaction between spin excitations and holes near the hot spots,
675: which is a consequence of the short-range character of this
676: interaction. In agreement with experiment the incommensurate peaks
677: which form the lower parabola are located at momenta
678: $(\frac{1}{2},\frac{1}{2}\pm\delta)$ and
679: $(\frac{1}{2}\pm\delta,\frac{1}{2})$, while peaks in the upper parabola
680: are at $(\frac{1}{2}\pm\delta,\frac{1}{2}\pm\delta)$ and
681: $(\frac{1}{2}\pm\delta,\frac{1}{2}\mp\delta)$. Also in agreement with
682: experiment the low-frequency incommensurability parameter $\delta$
683: grows linearly with the hole concentration $x$ for $x\lesssim 0.12$ and
684: then saturates. This behavior of $\delta$ is mainly connected with the
685: concentration dependence of the frequency $\omega_{\bf Q}$ of the spin
686: gap at the antiferromagnetic wave vector. We found that the
687: incommensurability for the transfer frequencies $\omega<\omega_{\bf Q}$
688: disappears if the damping of holes with energies $\pm\omega$ is greater
689: than $\omega$. This incommensurability vanishes also when the chemical
690: potential approaches the extended van Hove singularities at
691: $(0,\frac{1}{2})$ and $(\frac{1}{2},0)$. The incommensurability for
692: $\omega>\omega_{\bf Q}$ disappears for large spin-excitation damping.
693: The value of this damping depends heavily on the hole damping and on
694: the shape and width of the hole band. We suppose that the marked
695: difference in the frequency dependencies of the susceptibility in
696: YBa$_2$Cu$_3$O$_{7-y}$ and La$_{2-x}$Sr$_x$CuO$_4$ -- a pronounced peak
697: at $\omega\approx 25-40$~meV for ${\bf k=Q}$ in the former crystal and
698: a broad feature at $\omega\approx 10$~meV in the latter -- is a
699: consequence of the difference in the electron spectra. The larger
700: spin-excitation damping in La$_{2-x}$Sr$_x$CuO$_4$ leads to overdamping
701: of spin excitations, while in the underdoped YBa$_2$Cu$_3$O$_{7-y}$ the
702: excitations are well-defined even in the normal state.
703:
704: \section*{Acknowledgements}
705: This work was partially supported by the ESF grant No.~5548 and by the
706: DFG.
707:
708: \begin{thebibliography}{99}
709: \bibitem{Yoshizawa}H.~Yoshizawa, S.~Mitsuda, H.~Kitazawa, and
710: K.~Katsumata, {\it J.\ Phys.\ Soc.\ Jpn.} {\bf 57}, 3686 (1988);
711: R.~J.~Birgeneau, Y.~Endoh, Y.~Hidaka, K.~Kakurai, M.~A.~Kastner,
712: T.~Murakami, G.~Shirane, T.~R.~Thurston, and K.~Yamada, {\it Phys.\
713: Rev.} {\bf B39}, 2868 (1989).
714:
715: \bibitem{Yamada}K.~Yamada, C.~H.~Lee, K.~Kurahashi, J.~Wada,
716: S.~Wakimoto, S.~Ueki, H.~Kimura, Y.~Endoh, S.~Hosoya, G.~Shirane,
717: R.~J.~Birgeneau, M.~Greven, M.~A.~Kastner, and Y.~J.~Kim, {\it Phys.\
718: Rev.} {\bf B57}, 6165 (1998).
719:
720: \bibitem{Mason93}T.~E.~Mason, G.~Aeppli, S.~M.~Hayden, A.~P.~Ramirez,
721: and H.~A.~Mook, {\it Phys.\ Rev.\ Lett.} {\bf 71}, 919 (1993);
722: M.~Matsuda, K.~Yamada, Y.~Endoh, T.~R.~Thurston, G.~Shirane,
723: R.~J.~Birgeneau, M.~A.~Kastner, I.~Tanaka, and H.~Kojima, {\it Phys.\
724: Rev.} {\bf B49}, 6958 (1994).
725:
726: \bibitem{Dai}P.~Dai, H.~A.~Mook, and F.~Dogan, {\it Phys.\ Rev.\ Lett.}
727: {\bf 80}, 1738 (1998); M.~Arai, T.~Nishijima, Y.~Endoh, T.~Egami,
728: S.~Tajima, K.~Tamimoto, Y.~Shiohara, M.~Takahashi, A.~Garrett, and
729: S.~M.~Bennington, {\it Phys.\ Rev.\ Lett.} {\bf 83}, 608 (1999);
730: P.~Bourges, Y.~Sidis, H.~F.~Fong, L.~P.~Regnault, J.~Bossy, A.~Ivanov,
731: and B.~Keimer, {\it Science} {\bf 288}, 1234 (2000).
732:
733: \bibitem{Bourges}P.~Bourges, in {\it The Gap Symmetry and Fluctuations
734: in High Temperature Superconductors}, ed.\ J.~Bok, G.~Deutscher,
735: D.~Pavuna, and S.~A.~Wolf (Plenum Press, New York, 1998), p.~349;
736: H.~He, Y.~Sidis, P.~Bourges, G.~D.~Gu, A.~Ivanov, N.~Koshizuka,
737: B.~Liang, C.~T.~Lin, L.~P.~Regnault, E.~Schoenherr, and B.~Keimer, {\it
738: Phys.\ Rev.\ Lett.} {\bf 86}, 1610 (2001).
739:
740: \bibitem{Aeppli}G.~Aeppli, T.~E.~Mason, S.~M.~Hayden, H.~A.~Mook,
741: and J.~Kulda, {\it Science} {\bf 279}, 1432 (1997).
742:
743: \bibitem{Hayden96}S.~M.~Haiden, G.~Aeppli, H.~A.~Mook, T.~G.~Perring,
744: T.~E.~Mason, S.-W.~Cheong, and Z.~Fisk, {\it Phys.\ Rev.\ Lett.} {\bf
745: 76}, 1344 (1996).
746:
747: \bibitem{Hayden04}S.~M.~Haiden, H.~A.~Mook, P.~Dai, T.~G.~Perring, and
748: F.~Do\v{g}an, {\it Nature} {\bf 429}, 531 (2004).
749:
750: \bibitem{Tranquada}J.~M.~Tranquada, H.~Woo, T.~G.~Perring, H.~Goka,
751: G.~D.~Gu, G.~Xu, M.~Fujita, and K.~Yamada, {\it Nature} {\bf 429}, 534
752: (2004).
753:
754: \bibitem{Liu}D.~Z.~Liu, Y.~Zha, and K.~Levin, {\it Phys.\ Rev.\ Lett.}
755: {\bf 75}, 4130 (1995); N.~Bulut and D.~J.~Scalapino, {\it Phys.\ Rev.}
756: {\bf B53}, 5149 (1996).
757:
758: \bibitem{Brinckmann}J.~Brinckmann and P.~A.~Lee, {\it Phys.\ Rev.\
759: Lett.} {\bf 82}, 2915 (1999).
760:
761: \bibitem{Hizhnyakov}V.~Hizhnyakov and E.~Sigmund, {\it Physica} {\bf
762: C156}, 655 (1988); J.~Zaanen and O.~Gunnarsson, {\it Phys.\ Rev.} {\bf
763: B40}, R7391 (1989); K.~Machida, {\it Physica} {\bf C158}, 192 (1989).
764:
765: \bibitem{Ino}A.~Ino, C.~Kim, M.~Nakamura, T.~Yoshida, T.~Mizokawa,
766: A.~Fujimori, Z.-X.~Shen, T.~Kakeshita, H.~Eisaki and S.~Uchida, {\it
767: Phys.\ Rev.} {\bf B65}, 094504 (2002).
768:
769: \bibitem{Kimura}H.~Kimura, K.~Hirota, H.~Matsushita, K.~Yamada,
770: Y.~Endoh, S.-H.~Lee, C.~F.~Majkrzak, R.~Erwin, G.~Shirane, M.~Greven,
771: Y.~S.~Lee, M.~A.~Kastner, and R.~J.~Birgeneau, {\it Phys.\ Rev.} {\bf
772: B59}, 6517 (1999); M.~Fujita, H.~Goka, K.~Yamada, and M.~Matsuda, {\it
773: Phys.\ Rev.\ Lett.} {\bf 88}, 167008 (2002).
774:
775: \bibitem{Mori}H.~Mori, {\it Progr.\ Theor.\ Phys.} {\bf 34}, 399
776: (1965); S.~Onoda and M.~Imada, {\it J.\ Phys. Chem.\ Solids} {\bf 63},
777: 2225 (2002).
778:
779: \bibitem{Barzykin}V.~Barzykin and D.~Pines, {\it Phys.\ Rev.} {\bf
780: B52}, 13585 (1995).
781:
782: \bibitem{Sherman02}A.~Sherman and M.~Schreiber, {\it Phys.\ Rev.} {\bf
783: B65}, 134520 (2002); {\bf 68}, 094519 (2003).
784:
785: \bibitem{Sherman04}A.~Sherman and M.~Schreiber, {\it Phys.\ Rev.} {\bf
786: B69}, 100505(R) (2004).
787:
788: \bibitem{Endoh}Y.~Endoh, T.~Fukuda, S.~Wakimoto, M.~Arai, K.~Yamada,
789: and S.~M.~Bennington, {\it J.\ Phys.\ Soc.\ Japan} {\bf 69}, Suppl.~B,
790: 16 (2000).
791:
792: \bibitem{Kastner}M.~A.~Kastner, R.~J.~Birgeneau, G.~Shirane, and
793: Y.~Endoh, {\it Rev.\ Mod.\ Phys.} {\bf 70}, 897 (1998).
794:
795: \bibitem{Izyumov}Yu.~A.~Izyumov, {\it Usp.\ Fiz.\ Nauk} {\bf 167}, 465 (1997)
796: [{\it Phys.-Usp.\ (Russia)} {\bf 40}, 445 (1997)]; E.~Dagotto, {\it
797: Rev.\ Mod.\ Phys.} {\bf 66}, 763 (1994).
798:
799: \bibitem{Zhang}F.~C.~Zhang and T.~M.~Rice, {\it Phys.\ Rev.} {\bf B37},
800: 3759 (1988).
801:
802: \bibitem{Jefferson}J.~H.~Jefferson, H.~Eskes and L.~F.~Feiner, {\it Phys.\
803: Rev.} {\bf B45}, 7959 (1992); ~A.~V.~Sherman, {\it Phys.\ Rev.} {\bf
804: B47}, 11521 (1993).
805:
806: \bibitem{Kondo}J.~Kondo and K.~Jamaji, {\it Progr.\ Theor.\ Phys.} {\bf 47},
807: 807 (1972); H.~Shimahara and S.~Takada, {\it J.\ Phys.\ Soc.\ Japan}
808: {\bf 60}, 2394 (1991); S.~Winterfeldt and D.~Ihle, {\it Phys.\ Rev.}
809: {\bf B56}, 5535 (1997).
810:
811: \bibitem{Sherman03}A.~Sherman and M.~Schreiber, {\it Eur.\ Phys.\ J.} {\bf B32},
812: 203 (2003).
813:
814: \bibitem{Tallon}J.~L.~Tallon, C.~Bernhard, H.~Shaked, R.~L.~Hitterman,
815: and J.~D.~Jorgensen, {\it Phys.\ Rev.} {\bf B51}, R12911 (1995).
816:
817: \bibitem{McMahan}A.~K.~McMahan, J.~F.~Annett, and R.~M.~Martin, {\it Phys.\
818: Rev.} {\bf B42}, 6268 (1990); V.~A.~Gavrichkov, S.~G.~Ovchinnikov,
819: A.~A.~Borisov, and E.~G.~Goryachev, {\it Zh.\ Eksp.\ Teor.\ Fiz.} {\bf
820: 118}, 422 (2000) [{\it JETP (Russia)} {\bf 91}, 369 (2000)].
821:
822: \bibitem{Norman}M.~R.~Norman, {\it Phys.\ Rev.} {\bf B61}, 14751 (2000).
823:
824: \bibitem{Sherman04b}A.~Sherman, {\it Phys.\ Rev.} {\bf B70}, 184512 (2004).
825:
826: \bibitem{Damascelli}A.~Damascelli, Z.~Hussain, and Z.-X.~Shen, {\it Rev.\ Mod.\
827: Phys.} {\bf 75}, 473 (2003).
828: \end{thebibliography}
829: \end{document}
830: