1: \documentclass[twocolumn]{jpsj2}
2: %\documentclass{jpsj2}
3:
4: \def\runtitle{Effective Heisenberg-Model Description to the
5: Coupled Spin-Pseudospin ...}
6: \def\runauthor{Tohru {\sc Nakaegawa} and Yukinori {\sc Ohta}}
7:
8: \title{Effective Heisenberg-Model Description of the Coupled
9: Spin-Pseudospin Model\\ for Quarter-Filled Ladders}
10:
11: \author{Tohru {\sc Nakaegawa}$^{1}$ and Yukinori {\sc Ohta}$^{1,2}$
12: \thanks{E-mail: ohta@science.s.chiba-u.ac.jp}}
13:
14: \inst{$^1$Graduate School of Science and Technology, Chiba
15: University, Chiba 263-8522\\
16: $^2$Department of Physics, Chiba University, Chiba 263-8522}
17:
18: \recdate{April 5, 2002}
19:
20: \abst{Quantum Monte Carlo method is used to study the coupled
21: spin-pseudospin Hamiltonian in one-dimension (1D) that models
22: the charge-ordering instability of the anisotropic Hubbard
23: ladder at quarter filling. We calculate the temperature
24: dependence of the uniform spin susceptibility and the
25: spin and charge excitation spectra of the system to show
26: that there is a parameter and temperature region where the
27: spin degrees of freedom behave like a 1D antiferromagnetic
28: Heisenberg model. Anomalous spin dynamics in the
29: disorder phase of a typical charge-ordered material
30: $\alpha'$-NaV$_2$O$_5$ is thereby considered.}
31:
32: \kword{charge ordering, charge fluctuation, Hubbard ladder,
33: quarter filling, pseudospin}
34:
35: \begin{document}
36: \sloppy
37: \maketitle
38:
39: %%%%%%%%%%%%%%%%%%%%%%
40: \section{Introduction}
41: %%%%%%%%%%%%%%%%%%%%%%
42:
43: Charge-ordering (CO) instability has recently been one
44: of the major topics in the field of strongly correlated
45: electron systems. Here, elucidation of the observed
46: anomalous behaviors of electrons associated with the CO
47: phase transition has been the central issue. This includes
48: questions on the charge dynamics above the transition
49: temperature $T_{\rm CO}$ as well as on the CO spatial
50: patterns realized below $T_{\rm CO}$.
51: A well-known example is the vanadate bronze
52: $\alpha'$-NaV$_2$O$_5$ where the system may be modeled
53: as a lattice of coupled ladders (or a trellis lattice)
54: at quarter filling.\cite{smolinski,seo,nishimoto1,
55: mostovoy1,thalmeier}
56: Strong intersite Coulomb interaction between electrons
57: is believed to be the origin of the CO
58: instability.\cite{seo,nishimoto1} In this material,
59: the CO with a zigzag ordering pattern is observed below
60: $T_{\rm CO}=34$ K,\cite{isobe,ohama1,sawa,johnston,ohama2}
61: and associated with this, a number of anomalous behaviors,
62: which can be related to the slow dynamics of charge
63: carriers (or charge fluctuation), have been observed
64: above $T_{\rm CO}$.\cite{ravy,nakao,damascelli,presura,
65: ohama2,nishimoto2,nishimoto3,mostovoy2}
66: Anomalous response of the spin degrees of freedom
67: has also been noticed.\cite{hemberger,johnston,ohama3}
68: It seems therefore quite natural to wonder how in such
69: systems the spin degrees of freedom behave near the CO
70: phase transition when they are on the slowly fluctuating
71: charge carriers.
72: In this paper, we thus consider the issue: what are
73: the consequences of charge fluctuation at $T>T_{\rm CO}$
74: to the spin degrees of freedom?
75:
76: One of the simplest models that allow for such situation
77: is the anisotropic Hubbard ladders at quarter filling
78: with the strong intersite Coulomb repulsion.
79: We here use an effective Hamiltonian written in terms of
80: the spin and pseudospin (representing charge degrees of
81: freedom) operators.\cite{mostovoy1,cuoco,sa,mostovoy2}
82: This Hamiltonian is derived from the Hubbard ladder model
83: by the perturbation theory\cite{mostovoy1,cuoco,sa} where
84: the hopping parameter between the rungs of the ladder is
85: assumed to be small compared with the onsite and intersite
86: Coulomb repulsions as well as the hopping parameter in the
87: rung (i.e., the {\em anisotropic} ladder).\cite{nishimoto1}
88: Although the CO is not realized in this model (since it
89: is the 1D quantum-spin model), we can simulate anomalous
90: behaviors of the spin degrees of freedom under the strong
91: charge fluctuation.
92: We will apply the quantum Monte Carlo (QMC) method to
93: this model to calculate the temperature dependence of the
94: uniform spin susceptibility and the spin and charge excitation
95: spectra, thereby clarifying consequences of the interplay
96: between its spin and charge degrees of freedom.
97:
98: We note that, in this coupled spin-pseudospin model,
99: the spin exchange interaction is necessarily associated with
100: the charge excitation; i.e., the spin excitations cannot
101: occur without making the exchange of the pseudospins.
102: We will then show that nevertheless there is a parameter
103: and temperature region where the spin degrees of freedom
104: behave like a 1D antiferromagnetic Heisenberg model;
105: i.e., the spin degrees of freedom are `separated' from the
106: charge degrees of freedom in this region.
107: We will moreover show that the spin system behaves in different
108: manner depending on whether the temperature $T$ is below or
109: above a crossover temperature $T^*$ which is related to the
110: pseudospin excitations; at $T\lesssim T^*$, it behaves like
111: a 1D antiferromagnetic Heisenberg model with a $T$-independent
112: effective exchange coupling constant $J_{\rm eff}$ with large
113: renormalization, whereas at $T\gtrsim T^*$, $J_{\rm eff}$
114: decreases rapidly with increasing $T$, where the effective
115: Heisenberg-model description ceases to be valid.
116:
117: This paper is organized as follows. In \S2, we define
118: the coupled spin-pseudospin model that describes the spin
119: and charge degrees of freedom of the anisotropic Hubbard
120: ladder at quarter filling. Some details of the method
121: of calculation are also given. In \S3, we present the
122: results of calculation which include the staggered
123: susceptibility for pseudospins, the spin and pseudospin
124: excitation spectra, and the temperature dependence of
125: the uniform spin susceptibility. Discussion on the
126: experimental relevance to $\alpha'$-NaV$_2$O$_5$ and
127: summary of the paper will be given in \S4.
128:
129: %%%%%%%%%%%%%%%%%%%%%%
130: \section{Model and Method}
131: %%%%%%%%%%%%%%%%%%%%%%
132:
133: Our effective spin-pseudospin Hamiltonian may be written
134: as a sum
135: \begin{equation}
136: {\cal H}={\cal H}_0+{\cal H}_{\rm ST}
137: \end{equation}
138: of the quantum Ising Hamiltonian for pseudospins
139: \begin{equation}
140: {\cal H}_0=J_1\big(-\frac{g}{2}\sum_iT_i^x
141: +\sum_iT_i^zT_{i+1}^z\big)
142: \end{equation}
143: and the spin-pseudospin coupling term
144: \begin{equation}
145: {\cal H}_{\rm ST}=J_2\sum_i
146: \big({\bf S}_i\cdot{\bf S}_{i+1}-\frac{1}{4}\big)
147: \big(T_i^+T_{i+1}^-+{\rm H.c.}\big).
148: \end{equation}
149: The standard notation is used here.
150: ${\bf S}_i$ and ${\bf T}_i$ are, respectively, the spin
151: and pseudospin operators of spin-1/2 at site $i$, where
152: $T_i^z=-1/2$ ($+1/2)$ means the electron is on the left
153: (right) site on the rung of the ladder.
154: $J_1$ is the energy scale of the pseudospin system and
155: $J_2$ is the coupling strength between the spin and
156: pseudospin systems.
157:
158: From the second-order perturbation
159: theory,\cite{mostovoy1,cuoco,sa} we have the relations
160: $J_1=2V_\parallel$ and $J_2=4t_\parallel^2/V_\perp$,
161: where $t_\parallel$ and $V_\parallel$ ($t_\perp$ and
162: $V_\perp$) are the nearest-neighbor hopping parameter
163: and Coulomb repulsion of the leg (rung) of the ladder,
164: respectively. We should then have $J_1>J_2$, which
165: we assume throughout the present work.
166: We also assume the onsite Coulomb repulsion to be
167: $U\rightarrow\infty$.
168: Relative strength of the transverse field to the
169: pseudospins is measured by
170: $g=4t_\perp/J_1=2t_\perp/V_\parallel$.
171: Note that $g$ in the quantum Ising model represents
172: the relative strength of the fluctuation of a charge
173: in the rung: if we assume one electron in a rung, we
174: have the prefactor $gJ_1/2$ in the first term of eq.~(2),
175: which is the difference between the energies of the
176: bonding and antibonding levels of the rung, $2t_\perp$.
177: Thus, if $g$ (or $t_\perp$) is large the electron is
178: stable in the bonding level of the rung, but if $g$
179: (or $t_\perp$) is small the effect of $V_\parallel$
180: easily leads the system to CO.
181:
182: We use the conventional world-line QMC method for the
183: analysis of the model. We use a 32-site cluster (where
184: a site contains a spin and a pseudospin) with
185: periodic boundary condition; the cluster-size dependence
186: of the calculated results are examined by using clusters
187: up to 96 sites but we find no significant size
188: dependence in the results.
189: Because the model does not conserve the total pseudospin,
190: we have examined a number of ways of the spin flips
191: and confirmed that available analytical results are
192: reproduced correctly.\cite{nakaegawa}
193: The maximum-entropy method is used to calculate
194: the dynamical quantities like the spin and pseudospin
195: excitation spectra.
196:
197: %%%%%%%%%%%%%%%%%%%%%%
198: \section{Calculated Results}
199: %%%%%%%%%%%%%%%%%%%%%%
200:
201: %%%%%%%%%%%%%%%%%%%%%%
202: \subsection{Staggered susceptibility for pseudospins}
203: %%%%%%%%%%%%%%%%%%%%%%
204:
205: %%%% FIG.1 %%%%
206: \begin{figure}[h]
207: \vspace{5pt}
208: \begin{center}
209: %\epsfxsize=6.5cm \epsfbox{fig1.eps}
210: \includegraphics[width=6.5cm,clip]{fig1.eps}
211: \caption{Temperature dependence of the staggered susceptibility
212: for pseudospins $\chi_{\rm T}(\pi)$ calculated for the coupled
213: spin-pseudospin Hamiltonian.}
214: \end{center}
215: \label{fig:1}
216: \end{figure}
217: %%%%%%%%%%%%%%%
218: The response function is defined as
219: \begin{equation}
220: \chi_{ij}=\int_0^\beta\!{\rm d}\lambda\,
221: \big(\langle S_j^z(-i\lambda)S_i^z\rangle
222: -\langle S_j^z\rangle\langle S_i^z\rangle\big)
223: \end{equation}
224: where $S_j^z(-i\lambda)$ is the Heisenberg representation
225: of $S_j^z$ and $\langle\cdots\rangle$ is the canonical
226: average. $\chi_{ij}$ is Fourier transformed to the
227: $q$-dependent susceptibility $\chi(q)$, which we calculate
228: by the QMC method; the $q\rightarrow 0$ limit gives the
229: uniform spin susceptibility $\chi(T)$ and the staggered
230: susceptibility is defined as $\chi(q)$ at $q=\pi$.
231: In the following, we use the subscripts ${\rm S}$
232: and ${\rm T}$ as in $\chi_{\rm S}(q)$ and
233: $\chi_{\rm T}(q)$, which stand for the spin and
234: pseudospin degrees of freedom, respectively.
235:
236: The phase diagram of the quantum Ising model ${\cal H}_0$
237: is well known;\cite{sachdev} at $T=0$ there is a long-range
238: order for $g<1$ ($g=1$ is a quantum critical point), which
239: corresponds to the zigzag (or `antiferromagnetic') CO.
240: The calculated staggered susceptibility for pseudospins
241: is shown in Fig.~1, where we find that it shows divergent
242: behavior at $T\rightarrow 0$ for $g<1$.
243: The dispersion relation of the pseudospin excitation
244: observed in the calculated dynamical structure factor
245: (shown in Fig.~2, see below) agrees well with the
246: exact result:\cite{sachdev}
247: \begin{equation}
248: \omega_q=\frac{J_1}{2}\sqrt{1+g^2+2g\cos q}.
249: \end{equation}
250: We find in Fig.~1 that the inclusion of the
251: coupling term ${\cal H}_{\rm ST}$, which introduces the
252: quantum fluctuation via the factor $T_i^+T_j^-$, suppresses
253: the divergence. Thus, we may say that the inclusion of the
254: spin degrees of freedom in the quantum Ising model for pseudospins
255: leads to the unstable long-range CO.
256:
257: %%%%%%%%%%%%%%%%%%%%%%
258: \subsection{Spin and pseudospin excitation spectra}
259: %%%%%%%%%%%%%%%%%%%%%%
260:
261: %%%% FIG.2 %%%%
262: \begin{fullfigure}[t]
263: \vspace{5pt}
264: \begin{center}
265: %\epsfxsize=11.0cm \epsfbox{fig2.eps}
266: \includegraphics[width=11.0cm,clip]{fig2.eps}
267: \caption{Dynamical pseudospin structure factor
268: $S_{\rm T}(q,\omega)$ for the coupled spin-pseudospin
269: model calculated at $k_{\rm B}T=0.1J_2$.
270: The results at $J_2/J_1=0$ are for the quantum Ising
271: model. The peak at $\omega=0$ for $J_2/J_1>0$
272: in the uppermost panel of (a) and (b) is spurious, which
273: is due to the error of the maximum entropy method.}
274: \end{center}
275: \label{fig:2}
276: \end{fullfigure}
277: %%%%%%%%%%%%%%%
278: %%%% FIG.3 %%%%
279: \begin{fullfigure}[t]
280: \vspace{5pt}
281: \begin{center}
282: %\epsfxsize=11.0cm \epsfbox{fig3.eps}
283: \includegraphics[width=11.0cm,clip]{fig3.eps}
284: \caption{Dynamical spin structure factor
285: $S_{\rm S}(q,\omega)$ for the coupled
286: spin-pseudospin model calculated at
287: $k_{\rm B}T=0.1J_2$.}
288: \end{center}
289: \label{fig:3}
290: \end{fullfigure}
291: %%%%%%%%%%%%%%%
292: The dynamical pseudospin structure factor $S_{\rm T}(q,\omega)$
293: is defined as
294: \begin{eqnarray}
295: &&S_{\rm T}(q,\tau)=\frac{1}{N}\sum_{r_1,r_2}e^{-iq(r_2-r_1)}
296: \langle T_{r_1}^z(\tau)T_{r_2}^z(0)\rangle\\
297: &&S_{\rm T}(q,\tau)=\frac{1}{\pi}\int_0^\infty\!{\rm d}
298: \omega\,S_{\rm T}(q,\omega)K(\omega,\tau)\\
299: &&K(\omega,\tau)=e^{-\omega\tau}+e^{-\omega(\beta-\tau)}
300: \end{eqnarray}
301: where $S_{\rm T}(q,\tau)$ is the Fourier transform of the
302: imaginary-time correlation function. We use the maximum
303: entropy method for the inverse Laplace transformation (or
304: analytical continuation) to obtain $S_{\rm T}(q,\omega)$
305: from $S_{\rm T}(q,\tau)$.
306: The dynamical spin structure factor $S_{\rm S}(q,\omega)$ is
307: similarly defined by replacing the pseudospin operator $T^z_r$
308: with the spin operator $S^z_r$.
309:
310: The calculated results for the pseudospin excitation spectra
311: at low temperature ($k_{\rm B}T=0.1J_2$) are shown in Fig.~2,
312: where we find that the spectra are under strong influence of
313: the spin-pseudospin coupling term $J_2$.
314: With increasing the coupling strength $J_2/J_1$, the peak of
315: the pseudospin spectra shifts to higher energies and
316: simultaneously the spectra are broadened.
317: Thus, the lower-energy edge of the peak is not affected
318: strongly by the coupling strength $J_2$, at least when $g$
319: is large. It seems reasonable to suppose that the scattering
320: of the pseudospin excitations due to spin excitations causes
321: the broadening of the spectra.
322:
323: The calculated results for the spin excitation spectra at
324: low temperature are shown in Fig.~3, where we find that,
325: in contrast to the pseudospin spectra, the spin excitation
326: spectra change very little; i.e., the peak position, width,
327: as well as the shape of the spectra are not affected by
328: the parameter $J_1$ when $g\gtrsim 1$. When $g$ is small,
329: however, the peak position is slightly shifted to lower
330: energies with increasing the value of $J_1$ (see Fig.~3 (a)).
331: %%%% FIG.4 %%%%
332: \begin{figure}[t]
333: \vspace{5pt}
334: \begin{center}
335: %\epsfxsize=7.0cm \epsfbox{fig4.eps}
336: \includegraphics[width=7.0cm,clip]{fig4.eps}
337: \caption{Dispersion relations of the spin (open symbols)
338: and pseudospin (solid symbols) excitations calculated
339: at $k_{\rm B}T=0.1J_2$. Note that the same data at
340: $g=2$ (at $g=1$) are plotted in (a) and (b) (in (c) and
341: (d)) in different energy scales $J_1$ and $J_2$.
342: The dotted line in (a) and (c) is the dispersion
343: relation for the quantum Ising model eq.~(5), and
344: that in (b) and (d) is the scaled dispersion
345: relation for the 1D antiferromagnetic Heisenberg
346: model eq.~(9).}
347: \end{center}
348: \label{fig:4}
349: \end{figure}
350: %%%%%%%%%%%%%%%
351:
352: The dispersion relation of the spin and pseudospin excitations
353: calculated at low temperature are summarized in Fig.~4, which
354: are obtained as the momentum dependence of the peak position
355: of the spectra.
356: For comparison, we show the dispersion of the quantum Ising
357: model in Fig.~3 (a) and (c); the gap opens when $g>1$, which
358: is closed at $q=\pi$ when $g\rightarrow 1$, leading to the
359: `antiferromagnetic' long-range order (or zigzag CO).
360: We note that the gap remains open irrespective of the value
361: of $g$ when we include the coupling term $J_2$.
362: In Fig.~4, we present the same dispersion relations in a different
363: energy scales, i.e., $\omega_q/J_1$ and $\omega_q/J_2$.
364: We find that, unless $g$ is small, the spin excitation spectra
365: are always {\em inside} the charge gap, i.e., inside the gap
366: of the pseudospin excitation spectrum; when the charge gap is
367: large, the energy scale of the spin excitations is separated
368: from the high-energy charge excitations. With decreasing $g$,
369: however, the energy of the charge excitation decreases at the
370: momentum $q=\pi$ to couple with the spin excitations.
371: We find in Fig.~4 (b) and (d) that for $g\gtrsim 1$ the dispersion
372: of the spin excitation spectra scales very well with $J_2$; i.e.,
373: it does not depend on the value of $J_1$.
374: The dispersion of the calculated spin excitation spectra is
375: fitted well with the dispersion of the 1D antiferromagnetic
376: Heisenberg model
377: \begin{equation}
378: \omega_q/J_2=0.6\times\frac{\pi}{2}\sin q
379: \end{equation}
380: if we include the factor $0.6$ as in eq.~(9).
381: The factor is independent of $J_1$ for $g\gtrsim 1$ and
382: at low $T$.
383:
384: These results suggest that at low temperatures there is a
385: parameter region where the spin degrees of freedom
386: behaves independently from the pseudospin degrees of
387: freedom; it is when $g\gtrsim 1$ and the gap of the pseudospin
388: excitation spectra is large, inside of which there is
389: a spin excitation spectra.
390: Thus, we suggest the validity of the decoupling of
391: the coupling term of the Hamiltonian as
392: \begin{equation}
393: {\cal H}_{\rm ST}\Rightarrow J_2\sum_i
394: \big<T_i^+T_{i+1}^-+{\rm H.c.}\big>
395: \big({\bf S}_i\cdot{\bf S}_{i+1}-\frac{1}{4}\big)
396: \end{equation}
397: with
398: \begin{equation}
399: \big<T_i^+T_{i+1}^-+{\rm H.c.}\big>\simeq 0.6
400: \end{equation}
401: which leads to the effective Heisenberg-model description
402: of the spin degrees of freedom of our model.
403:
404: %%%%%%%%%%%%%%%%%%%%%%
405: \subsection{Uniform spin susceptibility}
406: %%%%%%%%%%%%%%%%%%%%%%
407:
408: %%%% FIG.5 %%%%
409: \begin{figure}[t]
410: \vspace{5pt}
411: \begin{center}
412: %\epsfxsize=8.5cm \epsfbox{fig5.eps}
413: \includegraphics[width=8.5cm,clip]{fig5.eps}
414: \caption{Temperature dependence of the uniform spin susceptibility
415: $\chi_{\rm S}$ calculated for the coupled spin-pseudospin Hamiltonian.
416: The solid and dotted curves are the uniform susceptibility for
417: the system of noninteracting $S=1/2$ spins and that for the 1D
418: antiferromagnetic Heisenberg model, respectively.}
419: \end{center}
420: \label{fig:5}
421: \end{figure}
422: %%%%%%%%%%%%%%%
423: To see the validity of the effective Heisenberg-model
424: description further, in particular for its temperature
425: dependence, we calculate the temperature dependence of
426: the uniform spin susceptibility for the coupled
427: spin-pseudospin Hamiltonian.
428: The results are shown in Fig.~5, where comparisons are
429: made with the uniform susceptibility for the system of
430: free spins and with that for the 1D antiferromagnetic
431: Heisenberg model.
432: We find that the temperature $k_{\rm B}T/J_2$ at which
433: $J_2\chi_{\rm S}(T)$ shows a maximum is lower than that
434: of the 1D antiferromagnetic Heisenberg model; it becomes
435: lower with decreasing the value of $g$ or with increasing
436: the value of $J_1/J_2$. In other words, the deviation
437: from the Heisenberg model is large when the quantum
438: fluctuation of the pseudospins is small, which occurs
439: when $g$ is small or $J_1$ is large.
440:
441: Now, let us analyze the data more precisely. In order to
442: do this, we fit the results with the temperature dependence
443: of the spin susceptibility of the 1D antiferromagnetic
444: Heisenberg model, the so-called Bonner-Fisher
445: curve;\cite{bonner} i.e., we introduce the $T$-dependent
446: {\em effective} exchange coupling constant $J_{\rm eff}(T)$
447: and we determine the values so as to fit the calculated
448: uniform spin susceptibility $\chi_{\rm S}(T)$.
449: If the values of $J_{\rm eff}$ thus obtained do not depend
450: on $T$, it follows that the spin degrees freedom of our
451: spin-pseudospin model is reduced to a 1D Heisenberg model
452: \begin{equation}
453: {\cal H}_{\rm spin}=J_{\rm eff}\sum_i\big({\bf S}_i
454: \cdot{\bf S}_{i+1}-\frac{1}{4}\big)
455: \end{equation}
456: at least for the response to the uniform magnetic field.
457: The results are shown in Fig.~6.
458: We find that the estimated value of $J_{\rm eff}(T)$
459: is indeed a constant for temperatures below
460: $k_{\rm B}T\lesssim 0.7J_2$ at $g=2$. A crossover
461: temperature $T^*$ $(=0.7J_2)$ is thereby defined.
462: The effective exchange coupling constant takes a value
463: \begin{equation}
464: J_{\rm eff}\simeq 0.6J_2
465: \end{equation}
466: which is consistent with the value estimated from the
467: dispersion relation of the spin excitation spectra
468: (see \S3.2). We find that also at $g=4$ the scaling behavior
469: holds up to a higher temperature ($k_{\rm B}T\lesssim 0.8J_2$),
470: but with the same value of $J_{\rm eff}$ (see Fig.~6 (d)),
471: demonstrating the validity of the effective Heisenberg-model
472: description at $T>T^*$. At $g=1$, however, the temperature
473: region where $J_{\rm eff}(T)$ takes a constant value is
474: already very small, although the value is still
475: $J_{\rm eff}\sim 0.6J_2$ at $T\sim 0$ K, and at $g=0.5$,
476: the value of $J_{\rm eff}$ at $T\sim 0$ K deviates largely
477: from $J_{\rm eff}=0.6J_2$ (or decreases strongly when
478: $J_1/J_2$ is large), where the effective Heisenberg-model
479: description completely fails.
480: %%%% FIG.6 %%%%
481: \begin{figure}[t]
482: \vspace{5pt}
483: \begin{center}
484: %\epsfxsize=8.5cm \epsfbox{fig6.eps}
485: \includegraphics[width=8.5cm,clip]{fig6.eps}
486: \caption{Effective exchange coupling constant $J_{\rm eff}(T)$
487: estimated from the fitting of the calculated uniform spin
488: susceptibility to the Bonner-Fisher curve.\cite{bonner}
489: Note that the same data are plotted as a function of
490: $k_{\rm B}T/J_1$ (left panels) and of $k_{\rm B}T/J_2$
491: (right panels), whereby a scaling behavior is seen in the
492: latter. The arrows indicate the crossover temperature $T^*$.
493: The solid lines are the guide to the eye.}
494: \end{center}
495: \label{fig:6}
496: \end{figure}
497: %%%%%%%%%%%%%%%
498:
499: We note here that the crossover temperature $T^*$ roughly
500: scales with $J_2$ rather than $J_1$, as seen in Fig.~6.
501: One might suppose that it should scale with the size of
502: the charge gap: i.e., up to temperatures corresponding
503: to the energy of the lowest charge excitations, with which
504: the pseudospins can excite, the spin excitations may be
505: written in terms of the 1D antiferromagnetic Heisenberg
506: model. However, as we have discussed in \S3.2, the size
507: of the charge gap shows a rather complicated behavior and
508: does not simply scale with either $J_2$ or $J_1$. The naive
509: picture thus does not hold. It may be said however that,
510: since there is no other excitations available, the
511: deviation from the 1D Heisenberg-model description is
512: necessarily due to the pseudospin excitations.
513:
514: %%%%%%%%%%%%%%%%%%%%%%
515: \section{Summary and Discussion}
516: %%%%%%%%%%%%%%%%%%%%%%
517:
518: We have calculated the spin and pseudospin excitation
519: spectra and the temperature dependence of the uniform
520: spin susceptibility of the coupled spin-pseudospin
521: Hamiltonian by using the QMC method.
522: We have first shown that, when the pseudospin quantum
523: fluctuation is large ($g\gtrsim 1$), the dispersion
524: relation of the spin exitation spectra of our model at
525: low temperatures agrees well with that of the 1D
526: antiferromagnetic Heisenberg model with the renormalized
527: effective exchange coupling constant $J_{\rm eff}=0.6J_2$
528: that is independent of the energy scale of the pseudospin
529: system $J_1$. Here, the spin excitation spectra is
530: well inside the charge gap, and thus the spin degrees
531: of freedom are separated from the charge degrees of
532: freedom.
533: We then have shown that the temperature dependence of
534: the uniform spin susceptibility of our model is well
535: described again by the 1D antiferromagnetic Heisenberg
536: model with the same effective exchange coupling constant
537: $J_{\rm eff}=0.6J_2$. The description is valid up to
538: the crossover temperature $T^*$ that is related to the
539: pseudospin excitations of the system and roughly scales
540: with $J_2$ unless the quantum fluctuation of the
541: pseudospins is small ($g\lesssim 1$).
542: We have thus demonstrated the validity of the effective
543: Heisenberg-model description of the coupled spin-pseudospin
544: model for the quarter-filled ladders.
545: It then follows that the coupling between the spin
546: and pseudospin degrees of freedom, which occurs at
547: $g\lesssim 1$, leads to the {\em anomalous} spin and charge
548: dynamics of the system.
549:
550: Although the real material $\alpha'$-NaV$_2$O$_5$
551: is modeled well as a 2D trellis-lattice system
552: rather than a 1D ladder system and thus we need
553: great caution in the direct application of the
554: present results, it may be interesting to have a
555: rough idea of the values of the physical parameters
556: appropriate for $\alpha'$-NaV$_2$O$_5$;
557: according to ref.~\cite{nishimoto1}, we have
558: $t_\parallel\sim 0.14$ eV,
559: $t_\perp\sim 0.30$ eV,
560: and $V_\parallel\sim V_\perp\sim 0.8$ eV,
561: which lead to
562: $J_1\sim 1.6$ eV,
563: $J_2\sim 0.10$ eV, and
564: $g\sim 0.75$.
565: Thus, the real material may be in the region of
566: $g\lesssim 1$, where the spin degrees of freedom
567: are not separated from the charge degrees of
568: freedom. The anomalous response of the spin
569: degrees of freedom may therefore be expected.
570: We would here point out, e.g., that the value of
571: $J_{\rm eff}$ estimated from the uniform
572: susceptibility observed in experiment (which
573: takes the value $\sim 600-700$ K at $T\sim 0$ K)
574: decreases with increasing temperature,\cite{johnston}
575: which is consistent with the results of our
576: calculation. The reported\cite{ohama3} anomalous
577: temperature dependence of the nuclear spin-lattice
578: relaxation rate $1/T_1$ is also interesting in this
579: respect.
580: To clarify the dynamics of the spin-charge coupled
581: systems near the real CO phase transition, we however
582: need not only to examine the region $g\lesssim 1$
583: in greater detail but also to include the 2D coupling
584: in the present model, which we want to leave for
585: future study.
586:
587: Because the anomalous charge dynamics has been noticed
588: also in other transition-metal oxides\cite{amasaki}
589: and some organic systems\cite{shibata}, we hope that
590: the present study will stimulate further researches on
591: the intriguing interplay between the spin and charge
592: degrees of freedom of strongly correlated electron
593: systems with CO instability.
594:
595: \section*{Acknowledgements}
596: We would like to thank A. W. Sandvik, T. Mutou, and
597: T. Suzuki for useful discussions on the numerical
598: techniques and T. Ohama for enlightening discussion
599: on the experimental aspects.
600: This work was supported in part by Grants-in-Aid for
601: Scientific Research (Nos.~11640335 and 12046216) from
602: the Ministry of Education, Culture, Sports, Science,
603: and Technology of Japan.
604: Computations were carried out at the computer centers
605: of the Institute for Molecular Science, Okazaki, and
606: the Institute for Solid State Physics, University of
607: Tokyo.
608:
609: \begin{thebibliography}{99}
610:
611: \bibitem{smolinski} H. Smolinski, C. Gros, W. Weber,
612: U. Pechert, G. Roth, M. Weiden, C. Geibel:
613: Phys. Rev. Lett. {\bf 80} (1998) 5146.
614: \bibitem{seo} H. Seo and H. Fukuyama:
615: J. Phys. Soc. Jpn. {\bf 67} (1998) 2602.
616: \bibitem{nishimoto1} S. Nishimoto and Y. Ohta:
617: J. Phys. Soc. Jpn. {\bf 67} (1998) 2996.
618: \bibitem{thalmeier} P. Thalmeier and P. Fulde:
619: Europhys. Lett. {\bf 44} (1998) 242.
620: \bibitem{mostovoy1} M. V. Mostovoy and D. I. Khomskii:
621: Solid State Commun. {\bf 113} (2000) 159.
622: \bibitem{isobe} M. Isobe and Y. Ueda:
623: J. Phys. Soc. Jpn. {\bf 65} (1996) 1178.
624: \bibitem{ohama1} T. Ohama, H. Yasuoka, M. Isobe,
625: and Y. Ueda: Phys. Rev. B {59} (1999) 3299.
626: \bibitem{sawa} H. Sawa, E. Ninomiya, T. Ohama, H. Nakao,
627: K. Ohwada, Y. Murakami, Y. Fujii, Y. Noda, M. Isobe,
628: and Y. Ueda: J. Phys. Soc. Jpn. {\bf 71} (2002) 385.
629: \bibitem{johnston} D. C. Johnston, R. K. Kremer, M. Troyer,
630: X. Wang, A. Kl\"umper, S. L. Bud'ko, A. F. Panchula,
631: and P. C. Canfield: Phys. Rev. B {\bf 61} (2000) 9558.
632: \bibitem{ohama2} For a review, see T. Ohama:
633: Bussei Kenkyu (Kyoto) {\bf 74} (2000) 391
634: [in Japanese].
635: \bibitem{ravy} S. Ravy, J. Jegoudez, and A. Revcolevschi:
636: Phys. Rev. B {\bf 59} (1999) R681.
637: \bibitem{nakao} H. Nakao, K. Ohwada, N. Takesue, Y. Fujii,
638: M. Isobe, and Y. Ueda: Physica B {\bf 21-243} (1998) 534.
639: \bibitem{damascelli} A. Damascelli, C. Presura,
640: D. van der Marel, J. Jegoudez, and A. Revcolevschi:
641: Phys. Rev. B {\bf 61} (2000) 2535.
642: \bibitem{presura} C. Presura, D. van der Marel,
643: A. Damascelli, and R. K. Kremer:
644: Phys. Rev. B {\bf 61} (2000) 15762.
645: \bibitem{nishimoto2} S. Nishimoto and Y. Ohta:
646: J. Phys. Soc. Jpn. {\bf 67} (1998) 3679.
647: \bibitem{nishimoto3} S. Nishimoto and Y. Ohta:
648: J. Phys. Soc. Jpn. {\bf 67} (1998) 4010.
649: \bibitem{mostovoy2} M. V. Mostovoy, J. Knoester,
650: and D. I. Khomskii: Phys. Rev. B {\bf 65} (2002) 064412.
651: \bibitem{hemberger} J. Hemberger, M. Lohmann, M. Nicklas,
652: A. Loidl, M. Klemm, G. Obermeier, and S. Horn:
653: Europhys. Lett. {\bf 42} (1998) 661.
654: \bibitem{ohama3} T. Ohama {\it et al.}: unpublished.
655: \bibitem{cuoco} M. Cuoco, P. Horsch, and F. Mack:
656: Phys. Rev. B {\bf 60} (1999) R8438.
657: \bibitem{sa} D. Sa and C. Gros:
658: Eur. Phys. J. B {\bf 18} (2000) 421.
659: \bibitem{nakaegawa} For details, see T. Nakaegawa:
660: {\it Master thesis} (Chiba University, 2002).
661: \bibitem{sachdev} S. Sachdev: {\it Quantum Phase Transitions}
662: (University Press, Cambridge, 1999).
663: \bibitem{bonner} J. C. Bonner and M. E. Fisher:
664: Phys. Rev. {\bf 135} (1964) A640.
665: \bibitem{amasaki} R. Amasaki, Y. Shibata, and Y. Ohta:
666: cond-mat/0110333.
667: \bibitem{shibata} Y. Shibata, S. Nishimoto, and Y. Ohta:
668: Phys. Rev. B {\bf 64} (2001) 235107.
669:
670: \end{thebibliography}
671:
672: \end{document}
673: