cond-mat0501511/dp1.tex
1: \documentclass[letter,twocolumn]{jpsj2}
2: 
3: \usepackage{graphicx}
4: 
5: \def\ve#1{\mbox{\boldmath $#1$}}
6: 
7: \def\Tc{T_\mathrm{c}}
8: \def\Re{\mathrm{Re}}
9: 
10: \title{%
11: Analysis of Superconductivity in 
12: d-p Model on Basis of Perturbation Theory\\
13: }
14: 
15: \author{%
16: Sotaro \textsc{Sasaki},
17: Hiroaki \textsc{Ikeda}
18: and Kosaku \textsc{Yamada}
19: }
20: 
21: \inst{
22: Department of Physics, Kyoto University, Sakyo-ku, Kyoto 606-8502 \\
23: }
24: 
25: \recdate{\today}
26: 
27: %%%%%%%%%%%%%%%%%%%%%%% Abstract %%%%%%%%%%%%%%%%%%%%%%%%%%%
28: \abst{%
29: We investigate the mass enhancement factor and the superconducting transition temperature in the d-p model 
30: for the high-$\Tc$ cuprates. 
31: We solve the \'Eliashberg equation using the third-order perturbation theory 
32: with respect to the on-site Coulomb repulsion $U$. 
33: We find that when the energy difference between the d-level and p-level is large, 
34: the mass enhancement factor becomes large and $\Tc$ tends to be suppressed 
35: owing to the difference of the density of state for d-electron at the Fermi 
36: level. 
37: From another viewpoint, 
38: when the energy difference is large, the d-hole number approaches 
39: unity and the electron correlation becomes strong 
40: and enhances the effective mass. This behavior for the electron number 
41: is the same as that for the 
42: f-electron number in the heavy fermion systems. 
43: The mass enhancement factor plays an essential role in understanding the 
44: difference in $\Tc$ between the LSCO and YBCO systems. 
45: }
46: 
47: %%%%%%%%%%%%%%%%%%%%%%% Keyword %%%%%%%%%%%%%%%%%%%%%%%%%%%%
48: \kword{% 
49: d-p model, high-$\Tc$ cuprate, mass enhancement factor, transition temperature, third-order perturbation theory
50: }
51: 
52: %%%%%%%%%%%%%%%%%%% Main text begins here. %%%%%%%%%%%%%%%%%
53: \begin{document}
54: \sloppy
55: \maketitle
56: 
57: %\section{Introduction}
58: Currently, superconductivity in high-$\Tc$ cuprates is being intensively investigated. 
59: For example, from the point of view of strong coupling, 
60: the $t$-$J$ model has been investigated. ~\cite{rf:Lee}
61: Also, Hubbard model has been investigated using the variational Monte Carlo 
62: method~\cite{rf:Nakanishi} and the quantum Monte Carlo method.~\cite{rf:Kuroki} On the other hand, from the point of view of weak coupling, 
63: the Hubbard model has been investigated using the fluctuation-exchange approximation and the perturbation theory. In the latter case, 
64: the nature of superconductivity in the high-$\Tc$ cuprates has almost been clarified on the basis of the nearly 
65: antiferromagnetic Fermi-liquid theory.~\cite{rf:Moriya,rf:Yanase} 
66: However, we have not yet explained the observed differences between the high-$\Tc$ systems, 
67: YBa$_2$Cu$_3$O$_{7-\delta}$(YBCO) and La$_{2-x}$Sr$_x$CuO$_4$(LSCO) and so on, 
68: in particular, the principal reason why the transition temperature $\Tc$ observed 
69: in LSCO is relatively low. 
70: The nuclear quadrupole resonance (NQR) experiment by Zheng {\it et al}. shows that when the ratio of the 
71: d-hole number ($n_d$) to the p-hole number ($n_p$), $n_d/2n_p$, is large, 
72: $\Tc$ 
73: is suppressed. ~\cite{rf:Zheng}
74: Actually, LSCO has a large ratio, as compared with YBCO. 
75: Also, in the specific heat experiment, 
76: the $\gamma$-value per mole in LSCO 
77: is approximately as large as that in 
78: YBCO. ~\cite{rf:Loram,rf:Momono}
79: Considering that YBCO has two CuO$_2$ layers and three Cu atoms 
80: in the unit, we find that the 
81: $\gamma$-value per layer in LSCO is large. 
82: The effect of the strong mass enhancement can be also seen in the nuclear magnetic resonance (NMR) relaxation rate. 
83: From the  systematic study of $(T_1T)^{-1}$, we can see that the effective Fermi energy in LSCO is rather smaller than that in YBCO.~\cite{rf:Kitaoka}
84: Thus, the strong electron correlation leads to a large mass enhancement, 
85: and then the $\Tc$ in LSCO is reduced. 
86: This proposal is very important to clarify a relation to the other materials, 
87: such as the heavy fermion systems, and to promote further progress on the 
88: unified picture in the strongly correlated systems. 
89: In this study, using the third-order perturbation theory, 
90: we investigate the mass enhancement factor and $\Tc$ in the d-p model 
91: for high-$\Tc$ cuprates, and clarify the difference in $\Tc$ between LSCO and 
92: YBCO. 
93: Husimi discussed it in the same framework and concluded that it 
94: originates from the difference in the effective on-site Coulomb repulsion 
95: $U$.~\cite{rf:Husimi} 
96: We would like to stress here that the mass enhancement factor plays an 
97: essential role. 
98: 
99: %\section{d-p model}
100: Since the CuO$_2$ plane is essential for superconductivity in 
101: high-$\Tc$ cuprates, we can consider the lattice structure with only the 
102: CuO$_2$ network shown in Fig. \ref{fig:dpmodel}. 
103: This structure has the $d_{x^2-y^2}$ orbitals on 
104: Cu sites and the $p_x$ and $p_y$ orbitals on O sites in the primitive cell. 
105: For simplicity, we consider only the hopping integrals $t_{dp}$ and $t_{pp}$ 
106: shown in Fig. 
107: \ref{fig:dpmodel}. 
108: This is the so-called d-p model. 
109: \begin{figure}[t]
110: \begin{center}
111: \includegraphics[width=8cm]{dpmodel.eps}
112: \end{center}
113: \caption{
114: Lattice structure of d-p model. 
115: $t_{\rm dp}$ and $t_{\rm pp}$ are the hopping integrals. 
116: }
117: \label{fig:dpmodel}
118: \end{figure}
119: In this study, we investigate in detail the nature of superconductivity 
120: in the d-p model. 
121: The model Hamiltonian is written as 
122: \begin{equation}
123: \begin{split}
124: H=H_0+H_{\rm int},
125: \end{split}
126: \end{equation}
127: where, 
128: \[
129: H_0=
130: \left(
131: \begin{array}{ccc}
132: d_{k\sigma}^{\dagger} & p_{k\sigma}^{x\dagger} & p_{k\sigma}^{y\dagger}
133: \end{array}
134: \right)
135: \left(
136: \begin{array}{ccc}
137: \varepsilon_d & \xi_k^x & \xi_k^y \\
138: \xi_k^x & \varepsilon_p & \xi^p_k \\
139: \xi_k^y & \xi^p_k & \varepsilon_p
140: \end{array}
141: \right)
142: \left(
143: \begin{array}{c}
144: d_{k\sigma}\\
145: p_{k\sigma}^{x}\\
146: p_{k\sigma}^{y}
147: \end{array}
148: \right).
149: \]
150: Here, $\varepsilon_d$ and $\varepsilon_p$ include the chemical potential 
151: $\mu$. The essential parameter is the level splitting 
152: $\varepsilon_d-\varepsilon_p$. 
153: The off-diagonal parts are given by 
154: \begin{equation}
155: \begin{split}
156: &\xi^{i}_k=-2t_{dp}\sin\frac{k_i}{2}, (i=x,y),\\
157: &\xi^{p}_k=4t_{pp}\sin\frac{k_x}{2}\sin\frac{k_y}{2}.
158: \end{split}
159: \end{equation}
160: The second term denotes the on-site Coulomb repulsion between the 
161: d-electrons, 
162: \begin{equation}
163: \begin{split}
164: H_{\rm int}=\frac{U}{N}\sum_k\sum_{q\neq 0}d^{\dagger}_{k+q\uparrow}d^{\dagger}_{k'-q\downarrow}d_{k'\downarrow}d_{k\uparrow}.
165: \end{split}
166: \end{equation}
167: We set $t_{dp}=1.0$ as an energy unit, and we also fix $t_{pp}=0.30$ and 
168: $n=4.90$ to reproduce the Fermi surfaces. 
169: In Fig. \ref{fig:fs}, we show the Fermi surfaces for
170: $\varepsilon_d-\varepsilon_p=1.0$ 
171: and $\varepsilon_d-\varepsilon_p=3.0$ in the 
172: case of 
173: $T=0.01$. 
174: We see that the Fermi surfaces coincide with those of the high-$\Tc$ cuprates, 
175: YBCO and LSCO, respectively. 
176: In this model, the electron number $n< 5.0$ represents the hole-doped case, 
177: and $n> 5.0$ represents the electron-doped case. 
178: We fix $T=0.01$ except when we discuss $\Tc$. 
179: \begin{figure}[t]
180: \begin{center}
181: \includegraphics[width=4cm]{fs2.eps}
182: \end{center}
183: \caption{
184: The unbroken and broken lines are the Fermi surfaces 
185: in the case of
186: $T=0.01$, $t_{\rm pp}=0.30$ and $n=4.90$ for 
187: $\varepsilon_d-\varepsilon_p=1.0$ and $\varepsilon_d-\varepsilon_p=3.0$, 
188: respectively. 
189: }
190: \label{fig:fs}
191: \end{figure}
192: \begin{figure}[t]
193: \begin{center}
194: \includegraphics[width=8cm]{dosband1.eps}
195: \end{center}
196: \caption{
197: Band structure and density of states for d-electron $\rho_d(\varepsilon)$ 
198: in case of 
199: $\varepsilon_d-\varepsilon_p=1.0$. 
200: }
201: \label{fig:band1}
202: \end{figure}
203: \begin{figure}[t]
204: \begin{center}
205: \includegraphics[width=8cm]{dosband3.eps}
206: \end{center}
207: \caption{
208: Band structure and density of states for d-electron $\rho_d(\varepsilon)$ 
209: in case of 
210: $\varepsilon_d-\varepsilon_p=3.0$. 
211: }
212: \label{fig:band3}
213: \end{figure}
214: Figs. \ref{fig:band1} and \ref{fig:band3} show the band structures and 
215: the density of states for d-electrons $\rho_d(\varepsilon)$ for 
216: $\varepsilon_d-\varepsilon_p=1.0$ and $\varepsilon_d-\varepsilon_p=3.0$, 
217: respectively. 
218: We see that when $\varepsilon_d-\varepsilon_p$ is large, 
219: $\rho_d(\varepsilon=0)$ at the Fermi level is large. \\
220: The bare Green's function $\hat{G}^{(0)}(k)$ is written as
221: \begin{equation}
222: \begin{split}
223: \hat{G}^{(0)}(k)=(i\omega_n\hat{1}-\hat{H}_0)^{-1}.
224: \end{split}
225: \end{equation}
226: We define the matrix elements of $\hat{G}^{(0)}(k)$ as 
227: \[
228: \hat{G}^{(0)}(k)=
229: \left(
230: \begin{array}{ccc}
231: G^{(0)}_{\rm dd}(k) & G^{(0)}_{\rm dp_x}(k) & G^{(0)}_{\rm dp_y}(k) \\
232: G^{(0)}_{\rm p_xd}(k) & G^{(0)}_{\rm p_xp_x}(k) & G^{(0)}_{\rm p_xp_y}(k) \\
233: G^{(0)}_{\rm p_yd}(k) & G^{(0)}_{\rm p_yp_x}(k) & G^{(0)}_{\rm p_yp_y}(k) 
234: \end{array}
235: \right).
236: \]
237: We need $G^{(0)}_{\rm dd}(k)$ to only describe the normal self-energy 
238: $\hat{\Sigma}(k)$ and 
239: the anomalous self-energy $\Delta(k)$, 
240: since the interaction exists only between 
241: the d-electrons.  
242: \[
243: \hat{\Sigma}(k)=
244: \left(
245: \begin{array}{ccc}
246: \Sigma_{\rm dd}(k) & 0 & 0 \\
247: 0 & 0 & 0 \\
248: 0 & 0 & 0 
249: \end{array}
250: \right).
251: \] 
252: The Green's function is given by 
253: \begin{equation}
254: \begin{split}
255: \hat{G}(k)=(i\omega_n\hat{1}-\hat{H}_0-\hat{\Sigma}(k))^{-1}.
256: \end{split}
257: \end{equation}
258: In this study, the chemical potential $\mu$ is determined 
259: so as to fix the total electron number, $n$, 
260: \begin{equation}
261: \begin{split}
262: n=2\frac{T}{N}\sum_k {\rm Tr}\,\hat{G}_0(k)=2\frac{T}{N}\sum_k {\rm Tr}\, \hat{G}(k).
263: \end{split}
264: \end{equation}
265: 
266: %\section{Third-order perturbation theory}
267: We apply the third-order perturbation theory
268: with respect to $U$. 
269: The normal self-energy is given by
270: \begin{equation}
271: \begin{split}
272: \Sigma_{\rm dd}(k)&= \frac{T}{N}\sum_{k'} [U^2 \chi_0(k-k') G_{\rm dd}^{(0)}(k') \\
273: & +U^3 \left( \chi_0^2(k-k')+\phi_0^2(k+k') \right) G_{\rm dd}^{(0)}(k')]. 
274: \end{split}
275: \end{equation}
276: Since the first-order term is constant, 
277: we can include the first-order term in $\varepsilon_d-\varepsilon_p$. 
278: Here, 
279: \begin{equation}
280: \begin{split}
281: &\chi_0(q)=-\frac{T}{N}\sum_{k} G_{\rm dd}^{(0)}(k)G_{\rm dd}^{(0)}(q+k), \\
282: &\phi_0(q)=-\frac{T}{N}\sum_{k}G_{\rm dd}^{(0)}(k)G_{\rm dd}^{(0)}(q-k).
283: \end{split}
284: \end{equation}
285: We also expand the effective pairing interaction up to the third-order terms 
286: with respect to $U$. \\
287: For the spin-singlet state, the effective pairing interaction is given by
288: \begin{equation}
289: \begin{split}
290: V(k;k')=V_{\rm RPA}(k;k')
291: +V_{\rm Vertex}(k;k'),
292: \end{split}
293: \end{equation}
294: where
295: \begin{equation}
296: \begin{split}
297: V_{\rm RPA}(k;k')=U+U^2\chi_0(k-k')
298: +2U^3\chi_0^2(k-k'), 
299: \end{split}
300: \end{equation}
301: and
302: \begin{equation}
303: \begin{split}
304: &V_{\rm Vertex}(k;k')=2(T/N)\Re\Big [\sum_{k_1}G_{\rm dd}^{(0)}(k_1) \\
305: &\times(\chi_0(k+k_1)-\phi_0(k+k_1))G_{\rm dd}^{(0)}(k+k_1-k')U^3\Big ].
306: \end{split}
307: \end{equation}
308: Here, $V_{\rm RPA}(k,k')$ is called the RPA terms and 
309: $V_{\rm Vertex}(k,k')$ is called the vertex 
310: corrections. 
311: Near the transition point, the anomalous self-energy $\Delta(k)$ satisfies 
312: the linearized \'Eliashberg equation,
313: \begin{equation}
314: \begin{split}
315: \lambda_{\rm max}\Delta(k)=-\frac{T}{N}\sum_{k'}V(k;k')|G_{\rm dd}(k')|^2\Delta(k'),
316: \end{split}
317: \end{equation}
318: where $\lambda_{\rm max}$ is the largest positive eigenvalue. 
319: In this equation, the temperature with $\lambda_{\rm max}=1$ 
320: corresponds to $T_{\rm c}$.
321: The symmetry of the superconductivity obtained here is the spin-singlet 
322: $d_{x^2-y^2}$ wave. 
323: 
324: %\section{Numerical Results}
325: %\subsection{$\varepsilon_d-\varepsilon_p$ dependence of $n_d/2n_p$}
326: We take 64 $\times$ 64 $\ve{k}$-meshes for the 
327: first Brillouin zone and 2048 Matsubara frequencies in the numerical 
328: calculation. 
329: \begin{figure}[t]
330: \begin{center}
331: \includegraphics[width=8cm]{dpratio1.eps}
332: \end{center}
333: \caption{
334: $\varepsilon_d-\varepsilon_p$ dependence of $n_d/2n_p$. 
335: }
336: \label{fig:dpratio1}
337: \end{figure}
338: First, we show the physical quantities in the unperturbed state. 
339: In Fig. \ref{fig:dpratio1}, we show the $\varepsilon_d-\varepsilon_p$ 
340: dependence of $n_d/2n_p$. 
341: When $\varepsilon_d-\varepsilon_p$ is large, $n_d/2n_p$ is large. 
342: Therefore, in the d-p model, the difference between LSCO 
343: and YBCO is represented as the difference of 
344: $\varepsilon_d-\varepsilon_p$. That is to say, 
345: $\varepsilon_d-\varepsilon_p$ is large for LSCO 
346: and small for YBCO, in this model. 
347: %\subsection{$\varepsilon_d-\varepsilon_p$ dependence of DOS}
348: \begin{figure}[t]
349: \begin{center}
350: \includegraphics[width=8cm]{dpdos1.eps}
351: \end{center}
352: \caption{
353: $\varepsilon_d-\varepsilon_p$ dependence of $\rho_d(0)$. 
354: }
355: \label{fig:dpdos1}
356: \end{figure}
357: In Figs. \ref{fig:dpdos1} and \ref{fig:dpnd1}, we show the 
358: $\varepsilon_d-\varepsilon_p$ dependences of $\rho_d(0)$ and $n_d$. 
359: When $\varepsilon_d-\varepsilon_p$ is large, these quantites also become 
360: large, and $n_d$ approaches unity. 
361: This indicates that LSCO is located in the relatively strongly correlated 
362: regime as compared with YBCO. 
363: %\subsection{$\varepsilon_d-\varepsilon_p$ dependence of $n_d$}
364: \begin{figure}[t]
365: \begin{center}
366: \includegraphics[width=8cm]{dpnd1.eps}
367: \end{center}
368: \caption{
369: $\varepsilon_d-\varepsilon_p$ dependence of $n_d$. 
370: }
371: \label{fig:dpnd1}
372: \end{figure}
373: %\subsection{$\varepsilon_d-\varepsilon_p$ dependence of mass enhancement factor. }
374: \begin{figure}[t]
375: \begin{center}
376: \includegraphics[width=8cm]{dpmass1.eps}
377: \end{center}
378: \caption{
379: $\varepsilon_d-\varepsilon_p$ dependence of 
380: $z^{-1}=\langle z^{-1}(\ve{k})\rangle$. 
381: }
382: \label{fig:dpmass1}
383: \end{figure}
384: 
385: Now, let us evaluate the mass enhancement factor using the 
386: third-order perturbation theory. 
387: \begin{equation}
388: \begin{split}
389: z^{-1}(\ve{k})=\left (1-\frac{\partial \Sigma(\ve{k},\omega)}{\partial \omega}\right )_{\omega\rightarrow 0}. 
390: \end{split}
391: \end{equation}
392: Fig. \ref{fig:dpmass1} shows the $\varepsilon_d-\varepsilon_p$ 
393: dependence of the average $z^{-1}=\langle z^{-1}(\ve{k})\rangle$. 
394: Here, $\langle\cdots\rangle$ represents the average over the momentum space. 
395: With increasing $\varepsilon_d-\varepsilon_p$, the mass enhancement factor 
396: increases. 
397: This is consistent with the finding that the mass enhancement of LSCO seems to be 
398: larger than that of YBCO, as mentioned above. ~\cite{rf:Loram,rf:Momono} 
399: This is due to the difference in $\rho_d(0)$. 
400: From another viewpoint, 
401: when the value of $\varepsilon_d-\varepsilon_p$ is large, $n_d$ 
402: approaches unity, and the electron mass is strongly enhanced by the strong 
403: electron correlation, similarly to the Mott transition. 
404: Such behavior has been markedly observed in the Ce-based heavy 
405: fermion systems. 
406: 
407: %\subsection{$\varepsilon_d-\varepsilon_p$ dependence of $\Tc$}
408: \begin{figure}[t]
409: \begin{center}
410: \includegraphics[width=8cm]{dptc.eps}
411: \end{center}
412: \caption{
413: $\varepsilon_d-\varepsilon_p$ dependence of $\Tc$. 
414: }
415: \label{fig:dptc1}
416: \end{figure}
417: 
418: Next, we discuss the superconducting transition temperature $\Tc$. 
419: From the experimental results, for large $\varepsilon_d-\varepsilon_p$, 
420: $\Tc$ is relatively low. 
421: In Fig. \ref{fig:dptc1}, we show the $\varepsilon_d-\varepsilon_p$ 
422: dependence of $\Tc$. 
423: $\Tc$ for $U=4.0$ increases as a function of $\varepsilon_d-\varepsilon_p$, 
424: and for $U=6.0$ is almost unchanged. Although the normal 
425: self-energy correction 
426: effect markedly suppresses $\Tc$, 
427: we cannot derive the results observed in the experiments within the present 
428: calculation. 
429: On the basis of the following reason, 
430: however, we can expect that by calculating the higher-order terms, 
431: we will obtain a better dependence of $\Tc$ on $\varepsilon_d-\varepsilon_p$. 
432: In Fig. \ref{fig:dptc1}, the filled (open) 
433: circles and squares denote $\Tc$ for $U=4.0$ ($U=6.0$) with and 
434: without the normal self-energy correction, respectively. 
435: The difference is mainly due to the effect of the mass renormalization. 
436: The larger $\varepsilon_d-\varepsilon_p$ is, the more this effect increases. 
437: This tendency is considered to be more marked in the higher-order terms of 
438: the normal self-energy. 
439: On the other hand, higher-order terms of the pairing interaction were evaluated by 
440: Nomura {\it et al.} in the Hubbard model. 
441: The results show that the convergency for 
442: the pairing interaction is very good for the spin-singlet pairing 
443: near the half-filling.~\cite{rf:Nomura} 
444: Namely, the contribution of the fourth-order terms to the pairing interaction 
445: is small compared with those of second-order and third-order terms. 
446: Here, we can expect the same trend. 
447: Thus, with the inclusion of the fourth-order terms 
448: in the \'Eliashberg equation, 
449: $\Tc$ will become relatively low for large $\varepsilon_d-\varepsilon_p$. 
450: This prediction is an important problem to be confirmed in the future. 
451: 
452: %\section{summary}
453: In conclusion, 
454: we have investigated the mass enhancement factor and the superconducting 
455: transition temperature in the d-p model for the high-$\Tc$ cuprates. 
456: We have solved the \'Eliashberg equation using the third-order perturbation theory 
457: with respect to the on-site repulsion $U$. 
458: We find that when $\varepsilon_d-\varepsilon_p$ is large, 
459: the mass enhancement factor becomes large and $\Tc$ tends to be suppressed 
460: owing to the difference in $\rho_d(0)$. 
461: From another viewpoint, when the d-hole number approaches 
462: unity, the electron correlation 
463: between d-holes (electrons) becomes strong and the effective mass increases. 
464: In fact, LSCO with d-hole number near unity shows 
465: strong mass enhancement . 
466: Here, we consider that $\Tc$ is given by the renormalization factor $z$ and
467: $\Tc^{\ast}$ as  
468: \begin{equation}
469: \begin{split}
470: \Tc\simeq z\Tc^{\ast},
471: \end{split}
472: \end{equation}
473: where $\Tc^{\ast}$ is the critical temperature determined by the calculation 
474: without any renormalization due to the normal self-energy correction. 
475: LSCO possesses small $z$ and exhibits low $\Tc$. Also, in the 
476: heavy fermion systems, when the number of 
477: f-electrons approaches unity, the effective mass is large and $\Tc$ is 
478: suppressed. 
479: This is the important unified
480: theory which holds for all the strongly correlated electron systems from 
481: cuprates to heavy fermions.  
482: Thus, our calculation shows that, in order to systematically discuss 
483: the physical quantities such as 
484: $\Tc$ and the 
485: mass enhancement factor, we need to use the d-p model. The Hubbard 
486: Hamiltonian is insufficient to represent the difference among 
487: material systems. \\
488: 
489: {\bf Acknowledgments}\\
490: 
491: Numerical calculation in this work was carried out at 
492: the Yukawa Institute Computer Facility.
493: 
494: %%%%%%%%%%%%%%%%%%References%%%%%%%%%%%%%%%%%%%
495: \begin{thebibliography}{99}
496: \bibitem{rf:Lee} P. A. Lee, N. Nagaosa and X-G. Wen:
497: cond-mat/0410445. 
498: \bibitem{rf:Nakanishi} T. Nakanishi, K. Yamaji and T. Yanagisawa:
499: J. Phys. Soc. Jpn. {\bf 66} (1997) 294.
500: \bibitem{rf:Kuroki} K. Kuroki and H. Aoki: 
501: Phys. Rev. B {\bf 56} (1997) 14287. 
502: \bibitem{rf:Moriya} T. Moriya and K. Ueda:
503: Adv. in Phys. {\bf 49} (2000) 555. 
504: \bibitem{rf:Yanase} Y. Yanase, T. Jujo, T. Nomura, H. Ikeda, T. Hotta and K. Yamada:
505: Phys. Rep. {\bf 387} (2003) 1. 
506: \bibitem{rf:Zheng} G.-q. Zheng, Y. Kitaoka, K. Ishida and K. Asayama:
507: J. Phys. Soc. Jpn. {\bf 64} (1995) 2524. 
508: \bibitem{rf:Loram} J. W. Loram, K. A. Mirza, J. R. Cooper and W. Y. Liang: 
509: Phys. Rev. Lett. {\bf 71} (1993) 1740. 
510: \bibitem{rf:Momono} N. Momono, T. Matsuzaki, M. Oda and M. Ido:
511: J. Phys. Soc. Jpn. {\bf 71} (2002) 2832. 
512: \bibitem{rf:Kitaoka} Y. Kitaoka, K. Ishida, S. Osugi, K. Fujiwara, G.-q. Zheng and K. Asayama: Appl. Magn. Reson. {\bf 3} (1992) 549. 
513: \bibitem{rf:Husimi} Y. Husimi:
514: Master thesis in Kyoto University. 
515: \bibitem{rf:Nomura} T. Nomura and K. Yamada:
516: J. Phys. Soc. Jpn. {\bf 72} (2003) 2053. 
517: \end{thebibliography}
518: 
519: \end{document}
520: 
521: