1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2: %% Trim Size: 9.75in x 6.5in

3: %% Text Area: 8in (include Runningheads) x 5in

4: %% ws-mpla.tex   :   1-6-2004

5: %% TeX file to use with ws-mpla.cls written in Latex2E.

6: %% The content, structure, format and layout of this style file is the

7: %% property of World Scientific Publishing Co. Pte. Ltd.

8: %% Copyright 1995, 2002 by World Scientific Publishing Co.

9: %% All rights are reserved.

10: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

11: %%

12: \documentclass[showpacs,showkeys,preprintnumbers,amsmath,aps]{revtex4}

13: \usepackage{epsfig}

14: \begin{document}

15: 

16: \markboth{Zhaoxin Xu} {QMC simulations on AFM-FM random

17: alternating spin chain}

18: 

19: %%%%%%%%%%%%%%%%%%%%% Publisher's Area please ignore %%%%%%%%%%%%%%%

20: %

21: %\catchline{}{}{}{}{}

22: %

23: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

24: 

25: \title{Quantum Monte Carlo Simulations on S=1/2 \\

26: Antiferromagnetic-Ferromagnetic Random Alternating spin chain}

27: 

28: 

29: \author{\footnotesize Peng Zhang,\quad Zhaoxin

30:   Xu\footnote{E-mail:zxxu@zimp.zju.edu.cn},\quad Heping Ying and \quad Jianhui Dai}

31: 

32: \affiliation{Zhejiang Institute of Modern Physics, Zhejiang University,\\

33: Hangzhou, 310027,

34: P.R. China}

35: 

36: 

37: %\maketitle

38: 

39: %\pub{\today}

40: \date{\today}

41: 

42: \begin{abstract}

43: The S=1/2 Heisenberg chain with bond alternation and randomness of

44: antiferromagnetic (AFM) and ferromagnetic (FM) interactions is

45: investigated by quantum Monte Carlo simulations of loop/cluster

46: algorithm. Our results have shown interesting finite temperature magnetic

47: properties of this model. The relevance of our study to the observed

48: results of the material (CH$_3$)$_2$CHNH$_3$Cu(Cl$_x$Br$_{1-x}$)$_3$

49: is discussed.

50: 

51: \keywords{bond randomness; antiferromagnetic; ferromagnetic.}

52: \end{abstract}

53: 

54: \pacs{75.10.Jm, 75.10.Nr, 75.40.Cx, 75.40.Mg}

55: \maketitle

56: 

57: \section{Introduction}

58: Randomness induced quantum phase transitions have attracted

59: intensive interests in the past decades. Putting enough strong bond

60: randomness, analyses of real space renormalization group (RSRG)

61: method have shown that RG flows of spin S=1 and S=1/2 quantum

62: antiferromagnetic Heisenberg chains \cite{Dasgupta,Fisher,Hyman1}

63: go to a stable fixed point called the random singlet (RS) phase in

64: which spins far apart in space randomly form weakly bound singlet

65: pairs. This property induces universal behaviors of ground

66: states and low temperature thermodynamics, e.g. the

67: energy spectrum is gapless, the temporal correlation length

68: $\xi_{\tau}$ and spatial correlation length $\xi_L$ diverge at

69: zero temperature, and there is a non-universal infinite dynamical

70: exponent $z$ which comes from the relation $\xi_{\tau}^z \sim

71: \xi_L$. More important, the uniform

72: susceptibility diverges universally in the RS phase as

73: \begin{equation}

74: \chi \sim \frac{1}{T {\bf ln}^2 (\Omega/T)}

75: \label{eq_RSxu}

76: \end{equation}

77: at low temperature, where $\Omega$ is a non-universal constant.

78: On the other hand, the S=1/2 dimerized AFM chain was found to be extremely stable

79: against strong bond randomness\cite{Hyman2}. In this case, the system is in a

80: quantum Griffiths-McCoy (QG) phase when the bond randomness is strong

81: enough. This phase is characterized by gapless excitations and

82: finite correlation length. In the QG phase, the uniform

83: susceptibility behaves as

84: \begin{equation}

85: \chi \sim T^{-\gamma}

86: \label{eq_QGxu}

87: \end{equation}

88: at low temperature, where $\gamma$ is a non-universal exponent.

89: On the experiments, the bond randomness effects have been found in several

90: antiferromagnetic quasi-1D materials

91: %such as

92: %BaCu$_2$(Si$_{x}$Ge$_{1-x}$)$_2$O$_7$, Mg-doped

93: %PbNi$_2$V$_2$O$_8$, and quasi-1D S=1/2 spin ladder material

94: %Zn-doped SrCu$_2$O$_3$

95:  \cite{Uchiyama,Azuma,Masuda}. Especially, a 

96: recent experiment on BaCu$_2$(Si$_{0.5}$Ge$_{0.5}$)$_2$O$_7$

97: \cite{Masuda} clearly show typical scaling relations of RS phase

98: predicted by theory\cite{Damle}.

99: Moreover, there is another kind of bond randomness whose bonds can be both

100: AFM and FM. In such systems, the RSRG analyses

101: \cite{Westerberg,Frischmuth} predicted a

102: universal fixed point different from the RS phase because the

103: spins correlate to form effective spins whose average size grows

104: with lowering of the energy scale, 

105: %Candidate materials of such

106: %randomness, e.g. Sr$_3$CuPt$_{1-x}$Ir$_x$O$_6$ \cite{Nguyen}, is

107: %considered to contain both AFM and FM bonds whose fraction is

108: %related to the concentration $x$ of Ir.

109:  and the magnetic susceptibility is {\it Curie-like}

110:  $\chi_u \sim 1/T$. Materials with

111:  such randomness is also fabricated \cite{Nguyen}.

112: 

113: Besides the above mentioned two kinds of bond randomness, Manaka and

114: coworkers recently found that the compound

115: (CH$_3$)$_2$CHNH$_3$Cu(Cl$_x$Br$_{1-x}$)$_3$ \cite{Manaka1} can be

116: considered as a bond randomness S=1/2 AFM and FM

117: alternating Heisenberg chains. The isomorphous compounds

118: (CH$_3$)$_2$CHNH$_3$CuCl$_3$ \cite{Manaka2} is regarded as quasi-1D S=1/2

119: FM-AFM alternating material, and

120: (CH$_3$)$_2$CHNH$_3$CuBr$_3$ \cite{Manaka3} is a S=1/2 AFM

121: dimerized material. Mixing these two compounds, they observed that

122: a gapless phase appeared in the regime of the intermediate concentration

123: $0.44<x<0.87$ of FM bonds by measuring magnetic susceptibility and

124: specific heats. In order to describe the properties of this

125: material, Hida \cite{Hida} and Nakamura

126: \cite{Nakamura} suggested a 1D model

127: \begin{equation}

128: H=\sum^N_{i=1}J{\bf S}_{2i-1}\cdot{\bf S}_{2i}+\sum^N_{i=1}J_i{\bf

129:   S}_{2i}\cdot{\bf S}_{2i+1},

130: \label{Hamiltonian}

131: \end{equation}

132: where ${\bf S}_i$ presents a spin S=1/2, $J>0$, $J_i=J_F(<0)$ with

133: a probability $p=x^2$  and $J_i=J_A(>0)$ with $1-p$.

134: This model has two limits: (i) when $p=1$, it is a S=1/2 AFM-FM

135: alternating Heisenberg spin chain. When $\mid J_i/J \mid > 1$ its

136: ground state is the Haldane phase with gapped energy spectrum;

137: (ii) when $p=0$, it is a dimerized AFM Heisenberg spin chain, its

138: ground state is the singlet dimer phase with gapped spectrum.

139: By density matrix renormalization group (DMRG) method, Hida

140: \cite{Hida} considered the case $p \geq 0.6$ with $J=1.0$ and $\mid J_i

141: \mid =2.0$ or $4.0$, and confirmed that there exists QG singularity when $p \leq

142: 0.7$. Nakamura \cite{Nakamura} studied

143: the model by non-equilibrium relaxation analysis of the

144: quantum Monte Carlo (QMC) simulation in the whole parameter space

145: of concentration $0 \leq x \leq 1$ with $J=1.0$ and $\mid J_i \mid

146: =2.0$, and found the gap vanishes in the regime $0.44 <x<0.87$

147: consistent with the experimental results \cite{Manaka1}. However, in

148: all these numerical works, the finite temperature magnetic properties

149: in the whole parameter space of $x$ are absent. In order to directly compared

150: with the experimental results \cite{Manaka1}, we perform the finite

151: temperature QMC simulations extensively on this model in this paper.

152: % as a starting point to

153: %compare with the experiment measurements.

154: 

155: \section{QMC simulation results}

156: We investigate the magnetic and thermodynamic properties of the model

157: defined by eq. (\ref{Hamiltonian}) with $J=1.0$ and $\mid J_i \mid =

158: 2.0$ using QMC simulations of continuous imaginary time loop/cluster algorithm.

159: The weak AFM coupling $J=1.0$ is fixed on all odd position bonds.

160: For even position bonds, the strong AFM or FM bonds $\mid

161: J_i \mid =2.0$ are chosen randomly according to the probability $p$. We perform

162: simulations for $100 \sim 200$  bond arrangement configurations. For

163: each bond configuration, after $2000$ Monte Carlo

164: steps (MCS) for thermalization,  we further update $2000$ MCS for

165: Monte Carlo average. The temporal and spatial periodic boundary

166: conditions are chosen for all simulations. In order to convince us of

167: the code validity, we study S=1/2 dimerized AFM chain in weak dimerization, 

168: and find our results for the energy gap consist well with

169: recent DMRG results \cite{Papenbrock}.

170: 

171: 

172: We first investigate the ground state properties on the system size

173: $L=128$ for temperatures as low as $\beta=1/T=200$. At low

174: temperature, the energy gap $\Delta$ is estimated by

175: \begin{equation}

176: \Delta = \lim_{L \rightarrow \infty}\frac{1}{\xi_{\tau}},

177: \end{equation}

178: where $\xi_{\tau}$ is the imaginary time correlation length

179: obtained by second-moment method \cite{Todo1}. Then the

180: valence-bond-solid (VBS) order parameter \cite{Todo2}

181: \begin{equation}

182: Z_{L} = <{\rm exp}[i\frac{2\pi}{L}\sum^{L}_{j=1}jS^{z}_{j}]>.

183: \end{equation}

184: is measured to distinguish different ground state phases. The

185: results obtained are shown in Fig. \ref{zd_eps}

186: \begin{figure}[ht]

187: \begin{minipage}[t]{4.8cm}

188: \epsfxsize \epsfysize

189: \centerline{\psfig{file=affm-zd.eps,width=2in}} 

190: \vspace*{4pt}

191: \caption{The energy gap $\Delta$ and VBS order parameter $Z_L$

192: versus the probability $p$.}

193: \label{zd_eps}

194: \end{minipage}

195: \hspace*{70pt}

196: \begin{minipage}[t]{5cm}

197: \epsfxsize \epsfysize

198: \centerline{\psfig{file=affm_limit.eps, width=2.4in}}

199: \vspace*{4pt}

200: \caption{The illustration of spin ordered phases with the probability

201:   $p=0$ and $1.0$.}

202: \label{limitsfig}

203: \end{minipage}

204: \end{figure}

205: 

206: It is interesting to find that energy gap $\Delta$ exists for the

207: Haldane phase at $p=1$ point. It gradually approaches to zero at $p

208: \approx 0.7$, and it opens again at very small $p \approx 0.02$.

209: It's hard to locate accurately the vanishing point of $\Delta$ because our $\Delta$ results are

210: not exact zero value due to finite-size effects. But our calculations

211: show that the gap should close at $p_{c1} \approx 0.02$ and $p_{c2}

212: \approx 0.7$. As a result, we presume that the system stays in the

213: gapless phase in the regime of $ p_{c1} < p < p_{c2}$, which is consistent with the

214: previous results \cite{Nakamura}.

215: 

216: In Fig. \ref{zd_eps}, we also observe that $Z_L \approx 1.0$ at

217: the limit $p=1$, keeps at finite in the regime $0.35 < p < 1.0$,

218: changes its sign at $p \approx 0.35$,  and then turns down to $-1.0$ when

219: $p <0.35$. The values of $Z_l \approx \pm 1.0$ characterize the two

220: limits of different ordered phases presented in Fig. \ref{limitsfig}.

221: %\begin{figure}[ht]

222: %\epsfxsize \epsfysize

223: %\centerline{\psfig{file=affm_limit.eps,width=2.0in}}

224: %\vspace*{}

225: %\caption{The illustration of spin ordered phases with the probability

226: %  $p=0$ and $1$.}

227: %\label{limitsfig}

228: %\end{figure}

229: 

230: Recent QMC study \cite{Arakawa} on S=1 random bond-alternating Haldane chain

231:  has shown that the VBS order parameter $Z_L$ is not effected by QG

232:  singularity, and it is an effective parameter to locate the RS

233:  critical point. Combining the results of $Z_L$ and $\Delta$, we

234: find that in the regime of $0.02 < p < 0.35$ where $\Delta$ vanishes and

235: $Z_L$ approaches zero from finite values, the system belongs to a

236:  critical phase. At the point $p \approx 0.35$, as $Z_L$ changes its

237:  sign, a phase transition happens. In the regime of $0.35<p<0.7$,

238:  $\Delta$ vanishes and $Z_l$ increases from zero to finite values.

239:  This fact reveals the system enters to other critical phase. In the

240:  regime of $0.7<p<1.0$, where both $\Delta$ and $Z_L$ are

241: finite, the system keeps in an ordered phase.

242: 

243: In order to distinguish the upper mentioned different phases, we

244: calculate the uniform magnetic susceptibility $\chi_u$ over the whole

245: parameter space $0<p<1.0$ at finite

246: temperatures. The results are summarized as following.

247: 

248: {\it I.} As shown in Fig. \ref{xu04_eps}, 

249: \begin{figure}[ht]

250: \begin{minipage}[t]{5cm}

251: \centerline{\psfig{file=xu_T_04.eps,width=2.0in}} \caption{The

252: uniform susceptibility at $0.02<p<0.35$ versus temperature

253:   $T=1/\beta$.}

254: \vspace*{4pt}

255: \label{xu04_eps}

256: \end{minipage}

257: \hspace*{70pt}

258: \begin{minipage}[t]{5cm}

259: \centerline{\psfig{file=fitRS.eps,width=2.0in}}

260: \vspace*{4pt}

261: \caption{The fitness of $\chi_u$ by $\frac{{\bf ln}^{-2}(\Omega/T)}{T}$}

262: \label{fitRS_eps}

263: \end{minipage}

264: %\vspace*{4pt}

265: 

266: \end{figure}

267: in the regime $0.02<p<0.35$, the system keeps in gapless phase, where $\chi_u$

268: diverges when $T \rightarrow 0$, and every $\chi_u$ curve appears a

269: valley, that is the typical feature of eq. (\ref{eq_RSxu}) for

270: denoting RS phase. We fit the curves of $p=0.04$ and $p=0.15$ by eq.

271: (\ref{eq_RSxu}), and find they can be very well fitted as plotted in

272: Fig. \ref{fitRS_eps}. Thus we believe that the phase in this regime

273: belongs to RS phase.

274: 

275: 

276: {\it II.} The regime of $0.35<p<0.7$ is also a gapless regime, where

277: $\chi_u$ curves diverge too, but they are obviously different from

278: those in the regime of $0.02<p<0.35$. Instead, as plotted in

279: Fig. \ref{xu0407_eps}, these curves are very similar as those in the

280: QG phase, where the typical feature is described by eq. (\ref{eq_QGxu}).

281: \begin{figure}[ht]

282: \begin{minipage}[t]{5cm}

283: \centerline{\psfig{file=xu_T_0407.eps,width=2.0in}}

284: \vspace*{4pt}

285: \caption{The uniform susceptibility at $0.35<p<0.7$ versus

286: temperature

287:   $T=1/\beta$.}

288: \label{xu0407_eps}

289: \end{minipage}\hspace*{70pt}

290: \begin{minipage}[t]{5cm}

291: \centerline{\psfig{file=fitQG.eps,width=2.0in}}

292: \vspace*{4pt}

293: \caption{The fitness of $\chi_u$ by $T^{-\gamma}$}

294: \label{fitQG_eps}

295: \end{minipage}

296: \end{figure}

297:  In Fig.\ref{fitQG_eps}, we fit our low temperature results of

298: $\chi_u$ by $T^{-\gamma}$, and find again the fitness are

299: quite good. It is interesting to note that the behavior of $\chi_u$ is

300: not {\it Curie-like}, so one can believe the phase is not belong to the

301: universal class of AFM and FM bonds randomness \cite{Westerberg}. We

302: thus conjecture that the system is now in QG phase. 

303: 

304: 

305: {\it III.} For the regime of $0.7<p<1.0$, the system enters to a gapped

306: phase because all $\chi_u$ curves appear the tendencies going to zero

307: when $T \rightarrow 0$. Our results are

308: plotted in Fig. \ref{xu0710_eps}.

309: \begin{figure}[ht]

310: \begin{minipage}[t]{5cm}

311: \centerline{\psfig{file=x08.eps,width=2.0in}}

312: \vspace*{4pt}

313: \caption{The uniform susceptibility at $0.7<p<1.0$ versus

314: temperature

315:   $T=1/\beta$.}

316: \label{xu0710_eps}

317: \end{minipage}

318: \hspace*{70pt}

319: \begin{minipage}[t]{5cm}

320: \centerline{\psfig{file=xu_limit.eps,width=2.0in}}

321: \vspace*{4pt}

322: \caption{The uniform susceptibility at $p=0$ and $p=1.0$ versus

323: temperature

324:   $T=1/\beta$.}

325: \label{xulimit_eps}

326: \end{minipage}

327: \end{figure}

328: 

329: {\it IV.} At last, we consider the two limit cases $p=0$ and $p=1$,

330: and our results are plotted in Fig. \ref{xulimit_eps}. Obviously, the

331: system belong to the gapped Haldane phase and

332: dimerized AMF phase on these two points, respectively.

333: 

334: \section{Conclusion and discussion}

335: From our QMC calculations, we can conclude that this S=1/2 AFM-FM

336: alternating bond randomness chain has four different phases with

337: respect to the probability $p$: ({\it i}) $p=0$, the

338: system is a dimerized AFM chains with gapped energy spectrum;

339: ({\it ii}) in the regime of $0.02<p<0.35$, the system enters to the RS

340: phase, whose energy spectrum is gapless and the uniform magnetic

341: susceptibility $\chi_u$ obeys the eq. (\ref{eq_RSxu}); ({\it iii})

342: in the regime of $0.35<p<0.7$, the system turns to the QG phase where the

343: energy gap vanishes and the curves of $\chi_u$ consist with eq. (\ref{eq_QGxu});

344: ({\it iv}) in the regime of $0.7<p \le 1.0$, the system is again in a

345: gapped phase. Finally, the case $p=1$ corresponds to the gapped AFM-FM alternating

346: spin chain.

347: 

348: Consequently, there should be three phase boundaries between these different phases:

349: ({\it i})$\rightarrow$({\it ii}), because there is no effective

350: quantities to locate the exact position of this boundary, we 

351: only say that the transition from the dimerized phase to the RS phase

352: happens at very small $p \approx 0.02$; ({\it ii})$\rightarrow$({\it

353:   iii}), the phase boundary between RS phase and the QG phase resides at

354: $p=0.35$, where both the results of VBS order parameter $Z_L$ and

355: susceptibility $\chi_u$ consist reciprocally. $Z_L$ changing its sign

356: at the point $p=0.35$ also implies that it is a good quantity to indicate

357: the transition from RS phase to other phase, which confirm the

358: previous argument \cite{Arakawa}; ({\it iii})$\rightarrow$({\it

359: iv}), the location of the boundary between QG phase and the gapped phase at

360: $p \approx 0.7$ is hard to be determined by results of energy gap $\Delta$ because

361: of the finite-size-effects, but it can be extracted from the

362: behaviors of $\chi_u$ at different temperature. In the QG

363: phase, $\chi_u$ diverges as $T \rightarrow 0$, but in a gapped

364: phase, $\chi_u$ approach zero when $T \rightarrow 0$. Thus there

365: should be a cross point of $\chi_u$ at different temperature which

366: correspond to the boundary. We investigate the case of

367: same system size $L=128$ at different temperature $1/T=\beta

368: =10,50,100,200$, and analyze the results by finite-size-scaling of

369: imaginary time \cite{Sachdev}.

370: \begin{figure}[ht]

371: \centerline{\psfig{file=scalex.eps,width=2.0in}}

372: \vspace*{4pt}

373: \caption{Finite size scaling by $T=1/\beta$ of the uniform susceptibility near $p_c=0.71$.}

374: \label{scalex_eps}

375: \end{figure}

376: Our results, plotted in Fig. \ref{scalex_eps}, have shown that the

377:  gap should close at $p_c=0.71 \pm 0.01$, this result consist with

378: previous DMRG results \cite{Hida}.

379: 

380: This model can be described under the scenario for the study of

381: S=1 AF bond random chain by R. Hyman and K. Yang \cite{Hyman1}, and 

382: our result give a positive evidence to support their conclusion. More

383: important, our QMC results of $\chi_u$ are consistent with the experiment

384: results \cite{Manaka1} of quasi-1D (CH$_3$)$_2$CHNH$_3$Cu(Cl$_x$Br$_{1-x}$)$_3$

385: in the two limits $p=0,1$, as found in Fig.\ref{xulimit_eps}. In the

386: gapless regime, $0.02<p<0.71$, where $\chi_u$ diverge when $T

387: \rightarrow 0$, the behavior are similar with the experiments

388: observation that $\chi_u(T)$ diverges also for $0.56<x<0.83$ at low

389: temperature. Especially, we find the critical point $p_{c2}=0.71$ corresponding

390: to $x=0.84$ is very close to the experimental $x=0.83$. As the critical point

391: $p_{c1} \approx 0.02$ corresponding to $x \approx 0.14$ is different from the

392: experimental $x=0.56$, it implies the current single chain model

393: eq. (\ref{Hamiltonian}) is failed to describe the results of

394: experiment for small $p$, and some additional terms such as the weak

395: inter-chain coupling should be taken into account. Furthermore, we

396: believe that the behavior of $\chi_u(T)$ in the regime $0.56< x

397: <0.87$ for the experiment should successively exhibit first

398: the QG-type divergence $T^{-\gamma}$, then the RS-type divergence

399: ${\bf ln}^{-2}(\Omega/T)/T$ at low temperatures. Our further

400: simulations are under consideration, and interesting results are expected in the

401: near future.

402: 

403: \section*{Acknowledgments}

404: Xu thanks for valuable discussion with Dr. H. Huang and Dr. P. Crompton.

405:  This work was supported in part by the NNSF of China and NSF of Zhejiang

406: province.

407: 

408: 

409: \begin{thebibliography}{0}

410: \bibitem{Dasgupta} C. Dasgupta and S.-K. Ma, Phys. Rev. B {\bf 22} 1305(1980).

411: 

412: \bibitem{Fisher} D.S. Fisher, Phys. Rev. B, {\bf 50},3799(1994).

413: 

414: \bibitem{Hyman1} R.A. Hyman and K. Yang, Phys. Rev. Lett. {\bf 78},

415:   1783(1997).

416: 

417: \bibitem{Hyman2} R.A. Hyman, K. Yang, R.N. Bhatt and S.M. Girvin,

418:   Phys. Rev. Lett. {\bf 76}, 839(1996).

419: 

420: \bibitem{Uchiyama} Y. Uchiyama, Y. Sasago, I.Tsukada, K. Uchinokura,

421:   A. Zheludev, T. Hayashi, N. Miura, and P. Boni, Phys. Rev. Lett.

422:   {\bf 84}, 632(1999).

423: 

424: \bibitem{Azuma} M. Azuma, Y. Fujishiro, M. Takano, M. Nohara and

425:   H. Takagi, Phys. Rev. B {\bf 55}, R8658(1997).

426: 

427: \bibitem{Masuda} T. Masuda, A. Zheludev, K. Uchinokura, J.-H. Chung

428:   and S. Park, cond-matt/0404688.

429: 

430: \bibitem{Damle} O. Motrunich, K. Damle and D.A. Huse, Phys. Rev. B

431: {\bf 63}, 134424(2001).

432: 

433: \bibitem{Westerberg} E. Westerberg, A. Furusaki, M. Sigrist and

434:   P.A. Lee, Phys. Rev. Lett. {\bf 75}, 4302(1997).

435: 

436: \bibitem{Frischmuth} B. Frischmuth, M. Sigrist, B. Ammon and

437:   M. Troyer, cond-matt/9808027.

438: 

439: \bibitem{Nguyen} T.N. Nguyen, P.A. Lee, and H.-C. zur Loye, Science

440:   {\bf 271}, 489(1996).

441: 

442: \bibitem{Manaka1} H. Manaka, I. Yamada and H.A. Katori, Phys. Rev. B

443:   {\bf 63}, 104408(2001).

444: 

445: \bibitem{Manaka2} H. Manaka, I. Yamada and K. Yamaguchi,

446:   J. Phys. Soc. Jpn. {\bf 66}, 564(1997).

447: 

448: \bibitem{Manaka3} H. Manaka, I. Yamada, J. Phys. Soc. Jpn. {\bf 66},

449:   1908(1997).

450: 

451: \bibitem{Hida} H. Hida, J. Phys. Soc. Jpn. {\bf 72}, 688(2003).

452: 

453: \bibitem{Nakamura} T. Nakamura, J. Phys. Soc. Jpn. {\bf 72},

454:   789(2003).

455: 

456: \bibitem{Todo1} S. Todo and K. Kato, Phys. Rev. Lett. {\bf 87}, 047203(2001).

457: 

458: \bibitem{Todo2} M. Nakamura and S. Todo, Phys. Rev. Lett. {\bf 88},

459:   167208(2002).

460: 

461: \bibitem{Papenbrock} T. Papenbrock, T. Barnes, D.J. Dean,

462:   M.V. Stoitsov, and M.R. Strayer, Phys. Rev. B {\bf 68},

463:   024416(2003).

464: 

465: \bibitem{Arakawa} T. Arakawa, S. Todo and H. Takayama,

466:   cond-mat/0410755.

467: 

468: \bibitem{Sachdev} S. Sachdev, {\it Quantum Phase Transition}(1999),

469:   Cambridge University Press.

470: 

471: \end{thebibliography}

472: 

473: \end{document}

474: