1: \documentstyle[prl,aps,multicol,epsf]{revtex}
2: \begin{document}
3: \draft
4: \widetext
5: \title{Ferromagnetic transition in a double-exchange model}
6: \author{Eugene Kogan$^{1}$, Mark Auslender$^2$ and Eran Dgani$^{1}$}
7: \address{$^1$ Jack and Pearl Resnick Institute
8: of Advanced Technology,
9: Department of Physics, Bar-Ilan University, Ramat-Gan 52900,
10: Israel\\
11: $^2$ Department of Electrical and Computer Engineering,
12: Ben-Gurion University of the Negev,
13: P.O.B. 653, Beer-Sheva, 84105 Israel}
14: \date{\today}
15: \maketitle
16: \begin{abstract}
17: \leftskip 54.8pt
18: \rightskip 54.8pt
19: We calculate the temperature of a ferromagnetic transition in a
20: double-exchange model with classical core spins for arbitrary relation
21: between Hund exchange coupling and electron band width by solving
22: the Dynamical Mean
23: Field Approximation equations.
24: \end{abstract}
25: \pacs{ PACS numbers: 75.10.Hk, 75.30.Mb, 75.30.Vn}
26: \begin{multicols}{2}
27: \narrowtext
28:
29: \section{Introduction}
30:
31: The double-exchange (DE)
32: model \cite{zener,anderson,degennes} is one of the basic ones
33: in the theory of magnetism. Magnetic ordering appears in this model due to
34: Hund
35: exchange coupling between the core
36: spins and the mobile carriers. The Hamiltonian of the model is
37: \begin{eqnarray}
38: \label{HamDXM}
39: H = \sum_{nn'\alpha} t_{n-n'} c_{n\alpha}^{\dagger} c_{n'\alpha}
40: -J \sum_{n\alpha\beta} {\bf S}_n\cdot
41: {\bf \sigma}_{\alpha\beta}c_{n\alpha}^{\dagger} c_{n\beta},
42: \end{eqnarray}
43: where $c$ and $c^{\dagger}$ are the electrons annihilation and creation
44: operators, ${\bf S}_n$ is the operator of core spin, $t_{n-n'}$ is the
45: electron hopping,
46: $J$ is Hund
47: exchange
48: coupling between a core spin and a conduction electron,
49: $\hat{\bf \sigma}$ is the vector of the Pauli matrices, and $\alpha,\beta$ are
50: spin indices.
51:
52: The model (the core spins being treated as classical
53: vectors) was thoroughly studied already in the
54: papers \cite{zener,anderson,degennes}. During the last years, because of a
55: general
56: interest in manganites, the model was brought in the
57: focus of attention, and a
58: lot more was achieved (see reviews \cite{coey99,izyumov01,ziese02}
59: and references therein).
60: However,
61: some basic properties of the model are still known only partially.
62: For example, most of
63: the papers dealing with the DE model, starting from classical paper by De
64: Gennes \cite{degennes}, considered the DE Hamiltonian
65: with infinite exchange (and with the addition of the antiferromagnetic
66: superexchange, which is crucial for the explanation of magnetic properties of
67: manganites).
68:
69: The other extremity which was studied is the particular case
70: of weak exchange (much less than the electron
71: bandwidth), when the DE Hamiltonian
72: can be reduced to
73: Ruderman-Kittel-Kasuya-Yosida (RKKY) Hamiltonian (see review \cite{vf} and
74: references therein).
75:
76:
77: In this paper we calculate the temperature of a ferromagnet - paramagnet
78: transition $T_c$ in a
79: double-exchange model for arbitrary relation
80: between Hund exchange coupling and electron band width by solving
81: the Dynamical Mean
82: Field Approximation equations.
83: Note that we treat the
84: core spins as classical
85: vectors. (When the quantum nature of the core spins is taken into account,
86: the Hamiltonian (\ref{HamDXM}),
87: which in this case is often called the periodic Kondo model, becomes much
88: more complicated; only scanty results were obtained for the model up to now.)
89: The problem of classical spins, as we shall see,
90: combines tractability with rich and
91: interesting physics.
92:
93:
94: \section{Hamiltonian and DMFA equations}
95:
96: Like it was said above, we consider spins as classical
97: vectors $ {\bf S}_n = {\bf m}_n$
98: with the normalization $|{\bf m}|^2 = 1$, Thus the DE Hamiltonian
99: in a single electron representation
100: can be presented as
101: \begin{equation}
102: \label{generic}
103: H_{nn'}=t_{n-n'}-J {\bf m}_n\cdot {\bf \sigma}\delta_{nn'}.
104: \end{equation}
105: We have a problem of electron scattered by core spins,
106: the probability of any given core spin
107: configuration depending upon the energy of electron
108: subsystem. To solve the problem we will use the
109: Dynamical Mean Field Approximation (DMFA) (see \cite{DMFA,furukawa2}
110: and references
111: therein).
112:
113: In this approach, first we calculate an averaged,
114: with respect to random
115: orientation of core spins, density of states of
116: electron in a random core spins configuration, treating electron scattering in a
117: single site approximation, and considering the probability of
118: any configuration as given. We introduce Green's function
119: \begin{eqnarray}
120: \label{green}
121: \hat{G}(E)=(E-H)^{-1},
122: \end{eqnarray}
123: In this approximation the averaged locator
124: \begin{eqnarray}
125: \hat{G}_{\rm loc}(E)=
126: \left\langle\hat{G}_{nn}(E)\right\rangle,
127: \end{eqnarray}
128: is expressed through the the local self-energy $\hat{\Sigma}$ by the
129: equation
130: \begin{eqnarray}
131: \label{local}
132: \hat{G}_{\rm loc}(E) =g_0\left(E - \hat{\Sigma}(E)\right),
133: \end{eqnarray}
134: where
135: \begin{eqnarray}
136: \label{g}
137: g_0(E) =\frac{1}{N}\sum_{\bf k}\left(E-t_{\bf k}\right)^{-1}
138: \end{eqnarray}
139: is the bare (in the
140: absence of the exchange interaction) locator. The
141: self-energy satisfies equation
142: \begin{eqnarray}
143: \hat{G}_{\rm loc}(E)=\left\langle \frac{1}
144: {\hat{G}_{\rm loc}^{-1}(E)+\hat{\Sigma} (E)
145: +J{\bf m}\cdot\hat{\bf \sigma}}\right\rangle,
146: \label{cpa}
147: \end{eqnarray}
148: where $\left\langle X({\bf m})\right\rangle \equiv \int X({\bf m})P({\bf m})$,
149: and $P({\bf m})$ is a probability
150: of a given spin orientation (one-site probability). The quantities
151: $\hat{G}$ and $\hat{\Sigma}$
152: are $2\times 2$ matrices in spin space.
153:
154: In PM phase $P({\bf m})={\rm const}$,
155: the averaging in Eq. (\ref{cpa}) can be
156: performed explicitly,
157: $\hat{\Sigma}=\Sigma\hat 1$, $\hat{G}=g\hat 1=g_0(E-\Sigma)\hat 1$,
158: where $\hat 1$ is a unity matrix, and we obtain
159: \begin{eqnarray}
160: g(E)= \frac{1}{2}\sum_{(\pm)} \frac{1}
161: {g^{-1}(E)+\Sigma(E)
162: \pm J}.
163: \label{cpa2}
164: \end{eqnarray}
165:
166: To first approximation of the DMFA, leading to Eq. (\ref{cpa}),
167: has a simple physical meaning. We reduce the problem of electron scattering due
168: to many spins, each with the scattering potential
169: $-J {\bf m}\cdot {\bf\sigma}$,
170: to a problem of a scattering due to a single spin with the effective
171: scattering potential
172: $-J {\bf m}\cdot {\bf \sigma}-\hat{\Sigma}$,
173: embedded in an effective medium,
174: described by the Hamiltonian $t_{\bf k}+\hat{\Sigma}$,
175: and, hence, by locator $\hat{G}_{\rm loc}$.
176:
177: The same MF approach leads to the second
178: DMFA approximation - the approximation for the one-site
179: probability
180: $P({\bf m})$, which allows to perform averaging in FM phase.
181: Consider again a single spin with the effective scattering potential
182: in an effective medium.
183: The change in the number of states of the electron gas due to
184: such spin is \cite{ziman,hewson}
185: \begin{eqnarray}
186: \label{probability4}
187: \Delta N(E,{\bf m})= -\frac{1}{\pi}{\rm Im}
188: \ln {\rm det}\left[1+\left(J{\bf m}\hat{\bf \sigma}
189: +\hat{\Sigma}_+\right)\hat{G}_{\rm loc\;+}\right],
190: \end{eqnarray}
191: where $Y_+\equiv Y(E+i0)$.
192: So the change in thermodynamic potential is \cite{doniach,chat,ak2}
193: \begin{eqnarray}
194: \beta\Delta\Omega({\bf m})=\int f(E)\Delta N(E,{\bf m})dE,
195: \label{probability}
196: \end{eqnarray}
197: where $f(E)$ is the Fermi function,
198: the chemical potential is found from the equation
199: \begin{equation}
200: n=-\frac{2}{\pi}\int_{-\infty}^{\infty}f(E){\rm Im}\;g_+ \; dE,
201: \end{equation}
202: and $n$ is the number of electrons per site.
203: The result for the one-site probability reads:
204: \begin{eqnarray}
205: \label{prob2}
206: P({\bf m})\propto \exp\left[-\beta\Delta\Omega({\bf m})\right].
207: \end{eqnarray}
208:
209:
210: Eqs. (\ref{cpa}) and (\ref{prob2}) are the system of two non-linear (integral)
211: equations for $\hat{\Sigma}(E)$ and $P({\bf m})$, which one should solve
212: to find thermodynamic properties of the model.
213: However, in linear approximation with respect to macroscopic magnetization
214: ${\bf M}$,
215: we can reduce this complicated system
216: to a traditional MF equation for
217: ${\bf M}$ \cite{aus02}:
218: \begin{eqnarray}
219: P({\bf m})\propto \exp\left( -3\beta T_c{\bf M}\cdot{\bf m}\right).
220: \label{probability2}
221: \end{eqnarray}
222: The parameter $T_c$ is formally introduced as a coefficient in the
223: linear term of the expansion of $\Delta \Omega({\bf m})$ with respect to
224: ${\bf M}$ (the reason for the notation we have chosen and for the numerical
225: coefficient 3 will be clear immediately); it is determined by the
226: properties of the system in paramagnetic phase.
227: Non-trivial solution of the MF equation
228: \begin{equation}
229: {\bf M}=\left\langle{\bf m}\right\rangle
230: \end{equation}
231: can exist only for $T<T_{c}$, hence $T_c$ the Curie
232: temperature.
233:
234:
235: \section{$T_c$ for semi-circular DOS}
236:
237: For simplicity consider the semi-circular (SC) bare density of states (DOS)
238: $N_0(\varepsilon)$, the bandwidth being $2W$.
239: Then
240: \begin{eqnarray}
241: \label{gint}
242: g_0(E)=\int \frac{N_0(\varepsilon)d \varepsilon }{E - \varepsilon}
243: =\frac{2}{W}\left[\frac{E}{W}-
244: \sqrt{\left(\frac{E}{W}\right)^{2}-1}\right].
245: \end{eqnarray}
246: For this case
247: \begin{equation}
248: \label{sigma}
249: \hat{\Sigma}(E)=E-2w\hat{G}_{\rm loc}-\hat{G}_{\rm loc}^{-1},
250: \end{equation}
251: where $w= W^2/8$,
252: and Eq. (\ref{cpa}) and (\ref{cpa2}) take respectively the form
253: \begin{eqnarray}
254: \label{sigma2}
255: \hat{G}_{\rm loc}(E)=\left\langle\frac{1}{E-2w\hat{G}_{\rm loc}(E)
256: +J{\bf m}\cdot\hat{\bf \sigma}}\right\rangle\\
257: \label{rq2}
258: g(E)=\frac{1}{2}\sum_{(\pm)}\frac{1}{E-2wg(E)\pm J}.
259: \end{eqnarray}
260: Expanding Eq. (\ref{sigma2}) and then Eq. (\ref{probability4}) with respect to
261: ${\bf M}$, after straightforward algebra, for the $T_c$ we obtain
262: \begin{eqnarray}
263: \label{Theta}
264: T_{\rm c}=\frac{4J^2w}{3\pi}\int_{-\infty}^{\infty}f(E)\nonumber\\
265: \mbox{Im}\left[\frac{g_+^2}
266: {(E_+-2wg_+)(E_+-4wg_+) -\frac{4J^2w}{3}g_+^2}\right]dE.
267: \end{eqnarray}
268: Formally speaking because the
269: integral in Eq. (\ref{Theta})
270: contains Fermi function critical temperature enters also into the r.h.s. of the equation.
271: But in all cases $T_c$ turns out to be much less the chemical potential, so we
272: can consider electron gas as degenerate, and Eq. (\ref{Theta}) is an explicit
273: formula for calculation of $T_c$.
274:
275: Eq. (\ref{Theta}) is the main result of our paper. The analysis of this equation
276: let us start from the limiting case $J=\infty$. In this case integral can be
277: calculated explicitly and we obtain \cite{ak2}
278: \begin{eqnarray}
279: \label{tcw}
280: T_c =\frac{W\sqrt{2}}{4\pi}\left[ \sqrt{1-y^{2}}-\frac{1}{\sqrt{3}}\tan^{-1}
281: \sqrt{3(1-y^{2})}\right],
282: \end{eqnarray}
283: where $y$ is an implicit function of concentration, given by equation
284: $n=\frac{1}{2}-\frac{1}{\pi}\left(\sin^{-1}y+y\sqrt{1-y^2}\right)$.
285: The result coincides with that known previously \cite{furukawa2}.
286:
287: For arbitrary exchange the integral can be calculated only numerically, but
288: before we present the results of calculations we should state that
289: in part of the $J/W-n$ plane Eq. (\ref{Theta})
290: gives $T_c<0$.
291: In fact, like in any MF theory of a second order phase transition,
292: in our calculation of the critical temperature we
293: started from a high temperature paramagnetic phase and decreasing
294: the temperature were looking
295: for an instability of a model with respect to appearance of small
296: spontaneous magnetic moment that is for the appearance
297: of a nontrivial solution of the MF equation. Negative $T_c$ in some
298: part of the
299: $J/W-n$ plane means, that at
300: any temperature, including
301: $T=0$, the paramagnetic phase here
302: is stable with respect to the appearance
303: of small spontaneous magnetic moment even at $T=0$, and excludes ferromagnetism
304: in that region of the plane. The part of the critical surface corresponding to
305: the region of the parameters plane where $T_c\geq 0$ is presented on FIG. 1.
306: \begin{figure}
307: \epsfxsize=3truein
308: \centerline{\epsffile{FIG1.eps}}
309: \label{FIG.1}
310: %\vskip .5cm
311: \caption{$T_c$ as a function of
312: relative strength of the Hund exchange $J/W$ and electron concentration $n$.}
313: \end{figure}
314:
315: \section{Discussion}
316:
317: Let us finally discuss, whether the PM-FM transition observed with decreasing
318: temperature can be preceded by
319: the transition from the the PM phase to some magnetic phase other
320: than FM (say, antiferromagnetic)?
321: We would like to
322: present some heuristic arguments that this is not the case.
323:
324: First, consider the case of weak exchange $J\ll W$. In this case Eq.
325: (\ref{Theta}) takes the form
326: \begin{eqnarray}
327: \label{rkky2}
328: T_{\rm c} = \frac{2J^2}{3}\int_{-\infty}^{\infty}f(E)
329: \left\{\frac{dN_0(E)}{dE}-\frac{1}{\pi }\mbox{Im}\;g_0(E_+)^2\right\}dE.
330: \end{eqnarray}
331: In fact, this equation is the MF approximation
332: \cite{aus02} for the RKKY Hamiltonian \cite{vf}, to which the original
333: Hamiltonian
334: (\ref{HamDXM}) can be reduced to in the case $J\ll W$. So Eq. (\ref{rkky2}) does not
335: involve either the coherent potential approximation (Eq. (\ref{cpa})),
336: or the SC density of states.
337:
338: Anyhow, for the SC density of states we use, Eq. (\ref{rkky2})
339: gives ferromagnetic ground state for $n<0.4$, which qualitatively
340: agrees with the result of
341: numerical calculations,
342: giving FM ground state for
343: $n<0.25$ for the three principal cubic lattices \cite{mattis}.
344:
345: To formulate the second argument,
346: let us compare the energies of the PM state
347: \begin{equation}
348: E_p=-\frac{2}{\pi}\int_{-\infty}^{E_F^{(p)}}E\;{\rm Im}\; g_+\; dE;
349: \end{equation}
350: and of the saturated FM state
351: \begin{eqnarray}
352: E_f=\int_{-\infty}^{E_F^{(f)}}E\;\sum_{(\pm)}N_{0}(E\pm J)\;dE,
353: \end{eqnarray}
354: where $E_F$ is the appropriate Fermi energy.
355: On fig. 2 we plotted simultaneously the curve given by by equation
356: $E_p=E_f$
357: and by equation
358: $T_c=0$,
359: where $T_c$ is obtained from Eq. (\ref{Theta}).
360: \begin{figure}
361: \epsfxsize=2.2truein
362: \centerline{\epsffile{FIG2.eps}}
363: \caption{The FM region boundary
364: in the coordinates of
365: relative strength of the Hund exchange $J/W$ and electron concentration $n$.}
366: \end{figure}
367: The close vicinity of the abovementioned curves supports the belief that
368: the curve $T_c=0$ is the
369: quantum critical line (see \cite{sachdev} and references therein),
370: which bounds {\it ferromagnetic} phase.
371: Also, the boundary of ferromagnetic phase on FIG. 2 agrees with those obtained on the basis
372: of numerical calculations \cite{dag} and
373: from qualitative reasoning \cite{chat}.
374:
375: The destruction of ferromagnetic ground state occurs because finite double
376: exchange between the itinerant electrons and core spins, unlike an
377: infinite one, by itself generates effective antiferromagnetic
378: exchange between the core spins (which was absent in original Hamiltonian).
379: Due to the fact that our main
380: result (equation for the transition temperature) indicates it's own
381: limits of validity, we can find the
382: boundaries of ferromagnetic phase without analyzing what phases
383: are beyond the boundaries.
384:
385: Finally, we would like to mention, that in the DMFA,
386: as one can easily see from Eq. (\ref{cpa}),
387: density of electron
388: states in paramagnetic phase does not depend upon electron concentration. In
389: this case,
390: the derivative of chemical potential with respect to number of electrons
391: is just the inverse density of states at the
392: Fermi level (for the degenerate electron gas), and
393: is always positive. Hence, there is no phase separation in
394: paramagnetic phase.
395:
396: In conclusion, we explicitly formulated the Dynamical Mean Field Approximation
397: equations
398: for the double exchange model with classical spins
399: for arbitrary relation between Hund exchange and
400: the electron bandwidth. Near paramagnetic-ferromagnetic transition critical
401: point, these equations were reduced to a MF equation, describing a
402: single spin in an effective field, proportional to the macroscopic
403: magnetization. The effective exchange interaction,
404: entering into the MF equation was
405: found for the semicircular electron density of states. We thus calculated the
406: transition temperature $T_c$ as a function of Hund exchange interaction and
407: electron density in the whole parameters plane. The results obtained also allow
408: to plot the boundaries of the ferromagnetic region on the model phase diagram.
409:
410: This research was supported by the Israeli Science Foundation administered
411: by the Israel Academy of Sciences and Humanities and BSF.
412:
413: \begin{thebibliography}{99}
414:
415: \bibitem{zener} C. Zener, Phys. Rev. {\bf 82}, 403 (1951).
416:
417: \bibitem{anderson} P. W. Anderson and H. Hasegawa, Phys. Rev. {\bf 100}, 675
418: (1955).
419:
420:
421: \bibitem{degennes} P. G. De Gennes, Phys. Rev. {\bf 118}, 141 (1960).
422:
423:
424: \bibitem{coey99} J. M. D. Coey, M. Viret, S. von Molnar, Adv. Phys. {\bf 48},
425: 167 (1999).
426:
427: \bibitem{izyumov01} Yu. A. Izyumov and Yu. N. Skryabin, Physics-Uspekhi, {\bf
428: 44}, 109 (2001).
429:
430: \bibitem{ziese02} M. Ziese, Rep. Prog. Phys. {\bf 65}, 143 (2002).
431:
432:
433: \bibitem{vf} J. H. Van Vleck, Rev. Mod. Phys. {\bf 34}, 681 (1962).
434:
435:
436: \bibitem{DMFA} A. Georges, G. Kotliar, W. Krauth, and M. J. Rozenberg, Rev. Mod.
437: Phys. {\bf 68}, 13 (1996).
438:
439: \bibitem{furukawa2} N. Furukawa: in {\it Physics of Manganites}, ed. T. Kaplan
440: and S. Mahanti (Plenum Publishing, New York, 1999).
441:
442: \bibitem{ziman}
443: J. M. Ziman, {\it Models of Disorder} (Cambridge University Press, Cambridge,
444: 1979).
445: \bibitem{hewson} A. C. Hewson, {\it The Kondo Problem to Heavy Fermions},
446: (Cambridge University Press,
447: Cambridge, England, 1993).
448:
449: \bibitem{doniach} S. Doniach and E. H. Sondheimer, {\it Green's functions for
450: solid
451: state physicists} (Imperial College Press, London, 1998).
452:
453: \bibitem{chat} A. Chattopadhyay, A. J. Millis, S. Das Sarma, \prb {\bf 61},
454: 10738 (2000).
455:
456: \bibitem{ak2} M. Auslender and E. Kogan,
457: \prb {\bf 65}, 012408 (2002);
458: Physica A {\bf 302}, 345 (2001).
459:
460: \bibitem{aus02} M. Auslender and E. Kogan,
461: EuroPhys. Lett. {\bf 59}, 277 (2002).
462:
463: \bibitem{mattis} D. C. Mattis, {\it The theory of magnetism}, (Springer, Berlin,
464: 1981).
465:
466: \bibitem{sachdev}
467: S. Sachdev, {\it Quantum Phase Transitions} (Cambridge University Press,
468: Cambridge, England, 1999).
469:
470: \bibitem{dag} E. Dagotto, S. Yunoki, A. L. Malvezzi, A. Moreo, J. Hu, S.
471: Capponi, D. Poilblanc, and N. Furukawa, \prb {\bf 58}, 6414 (1998).
472:
473:
474: \end{thebibliography}
475:
476: \end{multicols}
477: \end{document}
478:
479:
480: