1: \documentclass [aps,twocolumn] {revtex4}
2: \input epsf
3: %\topmargin -0.3in
4: %\textwidth 6.5in
5: %\textheight 8.5in
6:
7: %\oddsidemargin 0in
8:
9: \newcommand{\beq}{\begin{equation}}
10: \newcommand{\eeq}{\end{equation}}
11: \newcommand{\beqs}{\begin{eqnarray}}
12: \newcommand{\eeqs}{\end{eqnarray}}
13:
14: \begin{document}
15:
16: %\baselineskip 8.0mm
17:
18: \title{Simulations of a classical spin system with competing superexchange and
19: double-exchange interactions}
20:
21: \author {Shan-Ho Tsai\thanks{email: tsai@hal.physast.uga.edu}
22: %\and
23: and D. P. Landau\thanks{email: dlandau@hal.physast.uga.edu}}
24:
25: \affiliation{
26: Center for Simulational Physics, University of Georgia, Athens, GA 30602}
27:
28:
29: \begin{abstract}
30:
31: Monte-Carlo simulations and ground-state calculations have been used
32: to map out the phase diagram of a system of classical spins, on a simple
33: cubic lattice, where nearest-neighbor pairs of spins are coupled via
34: competing antiferromagnetic superexchange and ferromagnetic double-exchange
35: interactions. For a certain range of parameters, this model is
36: relevant for some magnetic materials, such as doped manganites, which
37: exhibit the remarkable colossal magnetoresistance effect.
38: The phase diagram includes two regions in which the two sublattice
39: magnetizations differ in magnitude.
40: Spin-dynamics simulations have been used to compute the time- and
41: space-displaced spin-spin correlation functions, and their Fourier
42: transforms, which yield the dynamic structure factor $S(q,\omega)$
43: for this system. Effects of the double-exchange interaction on the
44: dispersion curves are shown.
45:
46: \end{abstract}
47: \maketitle
48:
49: %\twocolumn
50:
51: \section{Introduction}
52:
53: Doped rare-earth manganites with general chemical formula Re$_{1-x}$A$_x$MnO$_3$
54: (where Re denotes a rare-earth and A is typically Ca or Sr)
55: have been shown to exhibit colossal magnetoresistance, and a wide range of other
56: physical properties, which are strongly dependent upon doping \cite{exp1}.
57: The physical mechanisms responsible for the
58: unusual magnetoresistance properties of these materials have not been understood
59: yet, and it has been suggested that double-exchange interactions \cite{zener}
60: between Mn$^{+3}$ and Mn$^{+4}$ ions are important. However,
61: there is much ongoing debate as to whether double-exchange alone can indeed give
62: rise to colossal magnetoresistance or if further mechanisms have to be considered
63: \cite{debate}.
64: It is thus important to further understand properties associated with the
65: double-exchange theory.
66:
67: Equal valence manganese ions in the parent compounds interact via a superexchange
68: mechanism, whereas the different valence manganese ions, introduced through doping,
69: are coupled via a double-exchange interaction \cite{exp1,zener,wollan}.
70: We investigate a simple model of classical spins in which nearest-neighbor spins
71: interact via a competing antiferromagnetic superexchange and a ferromagnetic
72: double-exchange interaction, where the ratio of these two interactions mimic
73: the effects of the doping $x$.
74: Properties of the pure Heisenberg model (only superexchange interaction) are
75: well-known \cite{heis} and
76: recently a high-accuracy Monte-Carlo study obtained the paramagnet-ferromagnet
77: transition temperature and associated static critical exponents of the pure
78: double-exchange model \cite{alvaro}. Recent studies of the phase diagram of
79: double-exchange systems using a different approach from ours are given in
80: Ref.\cite{arovas}. We use a combination of Monte-Carlo
81: simulations and ground-state calculations to determine the phase diagram of
82: the model with varying ratios of the superexchange and double-exchange interaction
83: strengths. We also use spin-dynamics techniques to study the dynamic structure
84: factor and obtain the dispersion curve of the double-exchange model.
85:
86: \section{Model and Methods}
87:
88: The model considered here can be described by the Hamiltonian
89: \beq
90: {\cal H}=-J_{AF}\sum_{<{\bf r},{\bf r'}>}{\bf S_r}\cdot {\bf S_{r'}}
91: -J_{DE}\sum_{<{\bf r},{\bf r'}>}\sqrt{1+{\bf S_r}\cdot {\bf S_{r'}}},
92: \label{ham}
93: \eeq
94: where ${\bf S_r}=({S_{\bf r}}^x,{S_{\bf r}}^y,{S_{\bf r}}^z)$ is a
95: three-dimensional classical spin of unit length at site ${\bf r}$. The first
96: term, with $J_{AF}<0$, is the antiferromagnetic superexchange interaction and
97: the second term describes a ferromagnetic double-exchange term with coupling
98: constant $J_{DE}>0$ between nearest-neighbor pairs of spins. We consider
99: $L\times L\times L$ simple cubic lattices with periodic boundary conditions, and
100: divide the system into two sublattices, $A$ and $B$, with magnetizations ${\bf m}_A$
101: and ${\bf m}_B$, respectively. The scalar product is defined as
102: ${\bf m}_A \cdot {\bf m}_B= |{\bf m}_A| |{\bf m}_B| \cos{\theta}\equiv y$, where clearly
103: $-1\leq y \leq 1$.
104:
105: In the ground state, the condition of energy minimization
106: leads to the expression
107: \beq
108: y=(J_{DE})^2/(4|J_{AF}|^2)-1,
109: \label{yeq}
110: \eeq
111: with $|J_{AF}|/J_{DE}\geq 1/\sqrt{8}=0.35355...$,
112: and for $|J_{AF}|/J_{DE} < 0.35355...$ the ground state is ferromagnetic.
113: Eq.(\ref{yeq}) admits the
114: possibility of non-colinear phases and/or sublattice magnetizations with different
115: magnitudes. Note that $y=0$ for $|J_{AF}|/J_{DE}=1/2$, which means that $\theta=\pi/2$
116: and/or that the magnetization in at least one sublattice is zero in this case.
117:
118: For $T>0$ we use Monte-Carlo simulations to measure thermodynamic quantities of
119: the model in order to determine the phase diagram. We used the Metropolis algorithm,
120: lattice sizes $L=12$ and $24$, and typically we discarded 10000 Monte-Carlo steps for
121: thermalization and used 10000-30000 steps in computing averages.
122:
123: The dynamics of the spins are governed by the equations of motion
124: \beq
125: \frac{d}{dt}{\bf S_r}={\bf S_r} \times (-\vec{\bigtriangledown}_{\bf r}{\cal H}),
126: \label{eqmotion}
127: \eeq
128: where the effective field
129: \beq
130: -\vec{\bigtriangledown}_{\bf r}{\cal H}=J_{AF}\sum_{\bf r'}{\bf S_{r'}}+\frac{J_{DE}}{2}
131: \sum_{\bf r'}\frac{\bf S_{r'}}{\sqrt{1+{\bf S_r}\cdot {\bf S_{r'}}}}
132: \eeq
133: and the sums are over ${\bf r'}$ nearest-neighbor to ${\bf r}$. These equations were
134: solved numerically using a method based on second-order Suzuki-Trotter decomposition
135: of exponential operators \cite{krech}, to a maximum time of $t_{max}=440/J_{DE}$.
136:
137: The dynamic structure factor $S({\bf q},\omega)$ for momentum transfer
138: ${\bf q}$ and frequency transfer $\omega$, observable in neutron scattering
139: experiments, is given by
140: \beq
141: S^k({\bf q},\omega)=\sum_{{\bf r,r'}} \exp[i {\bf q}\cdot ({\bf r}-{\bf r'})]
142: \int_{-\infty}^{+\infty} \exp(i\omega t) C^k({\bf r} - {\bf r'},t) \frac{dt}{\sqrt{2\pi}},
143: \eeq
144: where $C^k({\bf r} - {\bf r'},t)$ is the space-displaced, time-displaced spin-spin correlation
145: function defined, with $k=x, y,$ or $z$, as
146: $
147: C^k({\bf r} - {\bf r'},t) =\langle {S_{{\bf r}}}^k(t){S_{{\bf r'}}}^k(0)\rangle-
148: \langle {S_{{\bf r}}}^k(t)\rangle\langle {S_{{\bf r'}}}^k(0)\rangle.
149: $
150: The displacement ${\bf r}$ is in units of the lattice unit cell length $a$. More
151: details on the spin-dynamics methods are given in Refs. \cite{kun,alex}.
152:
153: \section{Results}
154:
155: Our preliminary results show that $|{\bf m}_A|=|{\bf m}_B|$ at all temperatures for
156: $|J_{AF}|/J_{DE}<1/\sqrt{8}$, decreasing smoothly
157: from $|{\bf m}_A|=|{\bf m}_B|=1$ at low $T$, towards zero as $T$ increases. For
158: $|J_{AF}|/J_{DE}>1/\sqrt{8}$ we find that at low temperatures $|{\bf m}_A|\ne |{\bf m}_B|$,
159: as illustrated in Fig. 1 for $|J_{AF}|/J_{DE}=0.9$.
160: For this range of interaction strengths
161: our results indicate that $|{\bf m}_A|=|(J_{DE})^2/(4|J_{AF}|^2)-1|$ and $|{\bf m}_B|=1$ at
162: $T=0$, and Eq.(\ref{yeq}) then implies that $\theta =0$ for $|J_{AF}|/J_{DE}<1/2$,
163: $\theta =\pi$ for $|J_{AF}|/J_{DE}>1/2$, and the sublattice
164: magnetizations reduce to $|{\bf m}_A|=0$ and $|{\bf m}_B|=1$ for $|J_{AF}|/J_{DE}=1/2$.
165: The phase-transition temperatures are obtained from analyses of the sublattice
166: magnetizations for $L=24$, hence there are some uncertainties in these
167: estimates due to finite-size effects.
168: \begin{figure}
169: \centering
170: \leavevmode
171: \epsfxsize=8.0cm
172: \epsffile{mag24s0p9.eps}
173: %\vspace{-3cm}
174: \caption{Sublattice magnetization for $|J_{AF}|/J_{DE}=0.9$, and lattice size $L=24$.}
175: \label{mags0p9}
176: \end{figure}
177:
178: The phase diagram that we obtained from the analysis of the sublattice magnetizations, shown in
179: Fig. 2, has five different regions: (i) ferromagnetic (F) for low $T$ and
180: small values of $|J_{AF}|/J_{DE}$; (ii)
181: antiferromagnetic (AF) for moderate $T$ and high values of $|J_{AF}|/J_{DE}$;
182: (iii) paramagnetic (PM) at high $T$; and two low
183: temperature phases, which we label as (iv) region I for $1/\sqrt{8}<|J_{AF}|/J_{DE}<1/2$,
184: and (v) region II for $|J_{AF}|/J_{DE}>1/2$.
185: Regions (i)-(iii) are characterized by $|{\bf m}_A|=|{\bf m}_B|$, with ${\bf m}_A={\bf m}_B$
186: in the ferromagnetic
187: phase and ${\bf m}_A=-{\bf m}_B$ in the antiferromagnetic phase. In contrast, in regions
188: I and II we find that $|{\bf m}_A|\ne |{\bf m}_B|$ and although our simulations suggest that
189: $\theta =0$ and $\pi$ in regions I and II, respectively, we cannot rule out possible small
190: deviations from the aligned and anti-aligned sublattice spin configurations in the respective
191: regions.
192: The question marks (?) in the phase diagram (see Fig. 2) indicate that the
193: points at which the F-PM and AF-PM transition lines join other phase boundaries have not
194: been determined accurately yet.
195: For the pure double-exchange ferromagnetic model ($J_{AF}=0$), the critical temperature
196: (F-PM transition point) is ${T_c}^{DE}=0.74515(7)J_{DE}$ \cite{alvaro}, whereas
197: for the pure Heisenberg model $T_c^{SE}=1.442929(77)|J_{AF}|$ \cite{heis}
198: for either ferromagnetic or antiferromagnetic interaction.
199: \begin{figure}
200: %\vspace{-10cm}
201: \centering
202: \leavevmode
203: \epsfxsize=8.0cm
204: \epsffile{phase_r2.eps}
205: %\vspace{-2cm}
206: \caption{Phase diagram. When not shown, the error bars are smaller than or of the size of the points and the solid lines simply join the data, guiding the eyes.}
207: \label{phasediag}
208: \end{figure}
209:
210: Fig. 3 shows a comparison of the dispersion relations for the pure
211: Heisenberg ferromagnet and for the pure double-exchange ferromagnetic model.
212: The results at $T=0$ correspond to the linear spin-wave theory, and data at the
213: respective critical temperatures are from our simulations.
214: \begin{figure}
215: \centering
216: \leavevmode
217: %\vspace{-10cm}
218: \epsfxsize=8.0cm
219: \epsffile{wqdese_r.eps}
220: %\vspace{-2cm}
221: \caption{Dispersion relations for the Heisenberg ferromagnet (pure superexchange model, labeled as SE) and the pure ferromagnetic double-exchange model (DE).}
222: \label{disprel}
223: \end{figure}
224:
225:
226: \section{Conclusions}
227:
228: We studied a system of classical spins coupled via competing antiferromagnetic
229: superexchange and ferromagnetic double-exchange interactions. As expected, when the
230: former is much weaker than the latter, the system is ferromagnetic and in the
231: reverse case the system is antiferromagnetic. In addition, we found two other low
232: temperature phases, in which the sublattice magnetizations differ in magnitude and
233: are close to being aligned or anti-aligned, depending on the ratio of the two
234: interaction strenghts. At high temperature the system is paramagnetic.
235: Comparing the dispersion curves we see that spin-waves in the pure double-exchange
236: model occur at lower frequency transfer than in the pure Heisenberg ferromagnet.
237:
238: \begin{center}
239: {\bf Acknowledgments}
240: \end{center}
241: This research was supported in part by NSF grant No. DMR-9727714.
242:
243: %\newpage
244:
245: \begin{thebibliography}{99}
246:
247: \bibitem{exp1}{See for example: P. Schiffer, A. P. Ramirez, W. Bao, and S.-W. Cheong,
248: Phys. Rev. Lett. {\bf 75}, 3336 (1995);
249: J. W. Lynn, R. W. Erwin, J. A. Borchers, Q. Huang, A. Santoro, J.-L. Peng,
250: and Z. Y. Li, Phys. Rev. Lett. {\bf 76}, 4046 (1996);
251: T. G. Perring, G. Aeppli, S. M. Hayden, S. A. Carter, J. P. Remeika,
252: and S.-W. Cheong, Phys. Rev. Lett. {\bf 77}, 711 (1996);
253: J. W. Lynn, R. W. Erwin, J. A. Borchers, A. Santoro, Q. Huang,
254: J.-L. Peng, and R. L. Greene, J. Appl. Phys. {\bf 81}, 5488 (1997);
255: L. Vasiliu-Doloc, J. W. Lynn, A. H. Moudden, A. M de Leon-Guevara,
256: and A. Revcolevschi, J. Appl. Phys. {\bf 81}, 5491 (1997). For a recent review,
257: see for example: J. M. D. Coey, M. Viret, and S. von Moln\'ar, Adv. in Phys. {\bf 48},
258: 167 (1999).}
259:
260: \bibitem{zener}{C. Zener, Phys. Rev. {\bf 82}, 403 (1951);
261: P. W. Anderson and H. Hasegawa, Phys. Rev. {\bf 100}, 675 (1955);
262: P.-G. de Gennes, Phys. Rev. {\bf 118}, 141 (1960).}
263:
264: \bibitem{debate}{See for example: A. J. Millis, R. Mueller, and B. I. Shraiman,
265: Phys. Rev. B {\bf 54}, 5405 (1996); and references therein. }
266:
267: \bibitem{wollan}{E. O. Wollan and W. C. Koehler, Phys. Rev. {\bf 100}, 545 (1955).}
268:
269: \bibitem{heis}{K. Chen, A. M. Ferrenberg, and D. P. Landau, Phys Rev. B {\bf 48},
270: 3249 (1993).}
271:
272: \bibitem{alvaro}{A. Caparica, A. Bunker, and D. P. Landau, submitted for publication.}
273:
274: \bibitem{arovas}{D. P. Arovas and F. Guinea, Phys. Rev. B {\bf 58}, 9150 (1998).}
275:
276: \bibitem{krech}{M. Krech, A. Bunker, and D. P. Landau, Comput. Phys. Comm. {\bf 111}, 1
277: (1998); J. Frank, W. Huang, and B. Leimkuhler, J. Comp. Phys. {\bf 133}, 160 (1997).}
278:
279: \bibitem{kun}{K. Chen and D. P. Landau, Phys. Rev. B {\bf 49}, 3266 (1994).}
280:
281: \bibitem{alex}{A. Bunker, K. Chen, and D. P. Landau, Phys. Rev. B {\bf 54}, 9259
282: (1996).}
283:
284: \end{thebibliography}
285: \end{document}
286:
287: %\newpage
288: \begin{center}
289: {\bf FIGURE CAPTIONS}
290: \end{center}
291:
292: \begin{itemize}
293: \item
294: {Fig. 1:} Sublattice magnetization for $|J_{AF}|/J_{DE}=0.9$, and lattice size $L=24$.
295:
296: \item
297: {Fig. 2:} Phase diagram. When not shown, the error bars are smaller than or of the size
298: of the points and the solid lines simply join the data, guiding the eyes.
299:
300: \item
301: {Fig. 3:} Dispersion relations for the Heisenberg ferromagnet (pure superexchange model,
302: labeled as SE) and the pure ferromagnetic double-exchange model (DE).
303:
304: \end{itemize}
305:
306:
307: \newpage
308: Fig. 1 \hspace{2cm} TSAI and LANDAU \hspace{2cm} Journal of Applied Physics
309:
310: \vspace{2cm}
311:
312: \begin{figure}
313: \centering
314: \leavevmode
315: \epsfxsize=8.5cm
316: \epsffile{mag24s0p9.ps}
317: \vspace{-3cm}
318: %\caption{}
319: \label{mags0p9}
320: \end{figure}
321:
322: \newpage
323: Fig. 2 \hspace{2cm} TSAI and LANDAU \hspace{2cm} Journal of Applied Physics
324:
325: \vspace{2cm}
326:
327: \begin{figure}
328: %\vspace{-10cm}
329: \centering
330: \leavevmode
331: \epsfxsize=8.5cm
332: \epsffile{phase_r2.ps}
333: \vspace{-2cm}
334: %\caption{}
335: \label{phasediag}
336: \end{figure}
337:
338: \newpage
339: Fig. 3 \hspace{2cm} TSAI and LANDAU \hspace{2cm} Journal of Applied Physics
340:
341: \vspace{2cm}
342:
343: \begin{figure}
344: \centering
345: \leavevmode
346: %\vspace{-10cm}
347: \epsfxsize=8.5cm
348: \epsffile{wqdese_r.ps}
349: \vspace{-2cm}
350: %\caption{}
351: \label{disprel}
352: \end{figure}
353:
354: \end{document}
355:
356: