1: \documentstyle [aps,twocolumn,epsfig]{revtex}
2: \begin{document}
3: \draft
4:
5: \twocolumn[
6:
7: \hsize\textwidth\columnwidth\hsize\csname@twocolumnfalse\endcsname
8: \begin{center}
9: {\Large \bf Orbital ordering in undoped manganites using Jahn-Teller
10: interaction} \\
11:
12: \vspace{0.5in}
13: {\large Sudhakar Yarlagadda and {Manidipa Mitra}$^1$ } \\
14:
15: \vspace{0.2in}
16: {Saha Institute of Nuclear Physics, Salt Lake City, Kolkata, India} \\
17:
18: \vspace{0.2in}
19: {$^1$S. N. Bose National Centre for Basic Sciences, Salt Lake City, Kolkata, India} \\
20:
21: \vspace{0.3in}
22: {Email : sudhakar@cmp.saha.ernet.in }
23: \end{center}
24: \maketitle
25:
26:
27: --------------------------------------------------------------------------------------------------------------------------------------------------------\\
28: {\small
29: {\bf Abstract } : To understand the orbital ordering of $LaMnO_3$ in
30: the ground state, we
31: study it in the A-type spin antiferromagnetic state.
32: We calculate the two-dimensional response functions
33: associated with the Jahn-Teller $Q_2$ and $Q_3$ distortions
34: and find that they diverge at the wavevector ($\pi$,$\pi$). Furthermore,
35: the $Q_2$ response function diverges faster which
36: indicates that perhaps the ordering of the orbitals at low temperatures
37: is dominated by the $Q_2$ distortion. We calculate the ground state energy
38: for the cases when only one of the $Q_2$ and $Q_3$ modes
39: is excited cooperatively and find that the $Q_2$ excited state yields
40: lower energy.\\
41:
42: \noindent
43: {\bf Keywords }: {Orbital density wave, Jahn-Teller interaction. }\\
44:
45: \noindent
46: {\bf PACS No. }:{ 71.45.Lr, 71.38.-k, 75.10.-b } }\\
47: --------------------------------------------------------------------------------------------------------------------------------------------------------\\
48: ]
49:
50: \vspace{0.1in}
51:
52: \noindent
53: {\bf 1. Introduction}
54:
55: \vspace{0.1in}
56:
57: Undoped manganites like $LaMnO_3$ are the parent systems for the colossal
58: magnetoresistive materials. It is well known that orbital ordering occurs
59: around 780 K resulting in a C-type orbital structure with two kinds
60: of orbitals alternating on adjacent sites in the $xy$ plane while
61: like orbitals are stacked in the $z$ direction [1].
62: %\cite{murakami}.
63: As the temperature is further
64: lowered to 140 K an A-type spin antiferromagnetic order sets in wherein
65: the spins are ferromagnetically aligned in the $xy$ plane with the spin
66: coupling in the $z$ direction being antiferromagnetic [2].
67: %\cite{wollan}.
68: To explain the
69: observed order several studies have been reported. These studies fall
70: into two classes, with one class based on Jahn-Teller (JT)
71: interaction [3--6]
72: %\cite{millis,dagotto,allen,satpathy}
73: %,lsheng}
74: being the main cause while the other class corresponds to Coulombic
75: interaction [7--10]
76: %\cite{khomskii,ishihara,maekawa}
77: being the dominant cause for the observed order. However,
78: it must be pointed out that the electron-phonon based studies do not point to
79: a transparent mechanism.
80:
81: The purpose of the present paper is to
82: use the framework developed earlier by one of us
83: %(Ref. \onlinecite{sudhakar})
84: (Ref. [11]) and analyze
85: the orbital ordering by using a generalized Peierls instability approach.
86: However, as compared to the one-dimensional Peierls charge density wave (CDW)
87: approach, our higher dimensional orbital density wave (ODW) analysis is
88: more complicated on account of there being
89: two $e_g$ orbitals (with inter-orbital hopping) and two response functions
90: corresponding to the JT $Q_2$ and $Q_3$ distortions.
91: We begin by assuming that the A-type antiferromagnetic ordering
92: has already set in. Thus, on account of strong Hund's coupling,
93: the transport is restricted to spin polarized electrons in two dimensions only.
94: We assume that the $Mn$ sites are fixed and that only the oxygen
95: atoms between the $Mn$ sites can move along the axis joining the neighboring
96: $Mn$ sites.
97: We find that the non-interacting response functions
98: $\chi_{2,3}$ corresponding to the $Q_{2,3}$ modes diverge at
99: wavevector $\vec{Q}=(\pi, \pi)$. Thus one can expect local minima in
100: energy for states corresponding to ODW being excited at
101: wavevector $\vec{Q}$ by the cooperative $Q_{2}$ and $Q_{3}$ modes.
102: We find that $\chi_2 ( \pi,\pi)$ diverges faster than
103: $\chi_3 ( \pi,\pi)$ and as expected the energy for the $Q_2$ related
104: ODW state has lower energy than the $Q_3$ related ODW state.
105:
106: \vspace{0.18in}
107:
108: \noindent
109: {\bf 2. Mean-field approach to ODW}
110:
111: \vspace{0.1in}
112:
113: We will now consider manganite systems with two $e_g$ orbitals per
114: site and ignore spin. The Hamiltonian consists of
115: the kinetic term, the ionic term, and the electron-ion interaction term.
116: The kinetic term in momentum space is given by
117: \begin{equation}
118: H_1 = \sum _{\vec{p}}{\bf B}^{\dagger}_{\vec{p}}\cdot{\bf T}
119: \cdot {\bf B}_{\vec{p}} ,
120: \end{equation}
121: where $ {\bf B}^{\dagger}_{\vec{p}} \equiv
122: (b^{\dagger}_{1\vec p} , b^{\dagger}_{2 \vec p})$ with
123: $b_1$ and $b_2$
124: corresponding to the destruction operators
125: for electrons with the orthonormal wavefunctions
126: $\psi_{x^2-y^2}$ and $\psi_{3z^2 -r^2}$ respectively.
127: Furthermore, ${\bf T}$ is a hermitian matrix with
128: ${\bf T}_{1,1}=-1.5 t[\cos p_x + \cos p_y]$,
129: ${\bf T}_{2,2}=-0.5 t[\cos p_x + \cos p_y]$, and
130: ${\bf T}_{1,2}=0.5 \sqrt{3} t[\cos p_x - \cos p_y ]$.
131: The eigen values of the kinetic energy are given by
132: $\lambda_{n}^{\vec{p}} = -\cos p_x - \cos p_y - (-1)^n \sqrt{\cos^2 p_x
133: + \cos^2 p_y - \cos p_x \cos p_y }$ with $n=1,2$.
134: The Fermi sea corresponding to the lower eigen energy value
135: $\lambda_{2}^{\vec{p}}$ is given by the union of the region
136: $-\pi/2 \le k_x \le \pi/2$ (with all values of $k_y$ allowed) and
137: the region
138: $-\pi/2 \le k_y \le \pi/2$ (with all values of $k_x$ allowed).
139: Whereas the Fermi sea corresponding to the higher eigen energy value
140: $\lambda_{1}^{\vec{p}}$
141: is given by the intersection of the region
142: $-\pi/2 \le k_x \le \pi/2$ and the region
143: $-\pi/2 \le k_y \le \pi/2$. Since the number of electrons is equal to
144: the number of sites, the total area
145: occupied by both Fermi seas is equal to the area of the
146: Brillouin zone ($4 \pi^2$). Furthermore, the Fermi surface corresponds
147: to $\lambda_{n}^{\vec{p}} = 0$.
148:
149: Using mean-field approximation we get,
150: after averaging the Hamiltonian over phonon coordinates,
151: the following effective Hamiltonian
152: %(see Ref. \onlinecite{sudhakar} for details):
153: (see Ref. [11] for details):
154: \begin{eqnarray}
155: \bar{H}= &&
156: H_{1}
157: -
158: 2 \frac{A^2}{\omega_0} \sum_{ j }
159: \left [ (b^{\dagger}_{1j} b_{2j} +
160: b^{\dagger}_{2j} b_{1j})
161: \langle b^{\dagger}_{1j} b_{2j} +
162: b^{\dagger}_{2j} b_{1j} \rangle \right .
163: \nonumber \\
164: &&
165: +
166: \left . (b^{\dagger}_{1j} b_{1j} -
167: b^{\dagger}_{2j} b_{2j})
168: \langle
169: b^{\dagger}_{1j} b_{1j} -
170: b^{\dagger}_{2j} b_{2j}
171: \rangle \right ]
172: \nonumber \\
173: &&
174: +
175: \frac{A^2}{\omega_0} \sum_{ j }
176: \left [ \langle
177: b^{\dagger}_{1j} b_{2j} +
178: b^{\dagger}_{2j} b_{1j}
179: \rangle ^2 +
180: \langle
181: b^{\dagger}_{1j} b_{1j} -
182: b^{\dagger}_{2j} b_{2j}
183: \rangle ^2 \right ] ,
184: \label{2dodw}
185: \end{eqnarray}
186: where $<..>$ implies averaging over the phonon coordinates,
187: $A \sqrt{2 M \omega_0}$ is the electron-JT coupling constant,
188: $2 A \langle b^{\dagger}_{1j} b_{2j} +
189: b^{\dagger}_{2j} b_{1j} \rangle =- \omega_0 \sqrt{2 M \omega_0} \langle
190: Q_{2j} \rangle $, and
191: $2 A \langle b^{\dagger}_{1j} b_{1j} -
192: b^{\dagger}_{2j} b_{2j} \rangle =- \omega_0 \sqrt{2 M \omega_0} \langle
193: Q_{3j} \rangle $.
194:
195: We will now determine the wavevector for long range cooperative ordering
196: of the $Q_{2,3}$ modes. For this we must figure out the values
197: of $q$ that make the susceptibilities $\chi_{2,3} (q,q)$
198: diverge. The expressions for
199: $\chi_{2,3} (\vec{q})$ are given as
200: \begin{eqnarray}
201: && \chi_{l} (\vec{q})=
202: -2 \sum_{\vec{k},n}\frac{\langle c^{n\dagger}_{\vec{k}}
203: c^{n}_{\vec{k}}\rangle}{
204: \lambda^{\vec{k}+\vec{q}}_{n} -\lambda^{\vec{k}}_{n}}
205: \sin^2 \left [ \frac{\theta_{\vec{k}+\vec{q}}
206: +\theta_{\vec{k}}+ l \pi }{2}
207: \right ]
208: \nonumber \\
209: &&
210: +2 \sum_{\vec{k}}\frac
211: {\langle c^{1\dagger}_{\vec{k}+\vec{q}}
212: c^{1}_{\vec{k}+\vec{q}}\rangle -
213: \langle c^{2\dagger}_{\vec{k}} c^{2}_{\vec{k}}\rangle}
214: { \lambda^{\vec{k}+\vec{q}}_{1} -\lambda^{\vec{k}}_{2}}
215: \cos^2 \left [ \frac{\theta_{\vec{k}+\vec{q}}
216: +\theta_{\vec{k}}+l \pi}{2}
217: \right ] ,
218: \end{eqnarray}
219: where $( c^{1\dagger}_{\vec{k}}, c^{2\dagger}_{\vec{k}})=
220: (b^{\dagger}_{1 \vec{k}}, b^{\dagger}_{2 \vec{k}}) \cdot {\bf M}$,
221: ${\bf M}$ is the diagonalizing matrix for
222: the kinetic matrix ${\bf T}$ with
223: ${\bf M}_{1,1}=\sin(\theta_{\vec{k}}/2)$,
224: ${\bf M}_{2,2}=-\sin(\theta_{\vec{k}}/2)$, and
225: ${\bf M}_{1,2}=\cos(\theta_{\vec{k}}/2)$.
226: At 0 K the second term on the right hand side of both $\chi_2$ and
227: $\chi_3$ produces a divergence at $\vec{Q}$ because of the fact that
228: $ \lambda^{\vec{k}+\vec{Q}}_{1} = -\lambda^{\vec{k}}_{2}$ and that
229: the Fermi energy is zero. Furthermore the ratio of $\chi_2 / \chi_3$
230: at 0 K and wavevector $\vec{Q}$ is expected to be 3.
231:
232: \vspace{0.18in}
233:
234: \noindent
235: {\bf 3. Results and discussion}
236:
237: \vspace{0.1in}
238:
239: In Fig. 1 we plot $\chi_{2,3} (\vec{Q})$ as a function of
240: temperature (with hopping term $t$ set equal to 0.1 eV)
241: and notice that they diverge as $T \rightarrow 0$
242: with $\chi_2$ diverging faster than $\chi_3$.
243: In Fig. 2 we plot $\chi_{2,3} (q,q)$ as a function of $q$
244: (with $0 \le q \le \pi$)
245: at fixed temperature T = 0.01 K and hopping term t = 0.1 eV
246: and find that they both peak sharply
247: close to $\vec{Q}$ with $\chi_2$ peaking faster. Then based on the fact that
248: the phonon mode goes soft at wavevector $\vec{Q}$ we compute the
249: ground state energy when only either $Q_2$ mode or $Q_3$ mode gets excited
250: cooperatively in the system. The order parameters are given by
251: $\langle b^{\dagger}_{1j} b_{2j} +
252: b^{\dagger}_{2j} b_{1j} \rangle =
253: c_2 \cos (\vec{Q} \cdot \vec{R_j})$ and
254: $\langle b^{\dagger}_{1j} b_{1j} -
255: b^{\dagger}_{2j} b_{2j} \rangle =
256: c_3 \cos (\vec{Q} \cdot \vec{R_j})$ with $-1 \le c_{2,3} \le 1$
257: %and $\vec R_j$ being the position vector.
258: %Here it should be pointed out that the order parameter
259: %$\langle b^{\dagger}_{1j} b_{2j} +
260: %b^{\dagger}_{2j} b_{1j} \rangle$
261: %corresponds to the density difference of electrons in the
262: %two orbitals
263: % $\psi_X \equiv (\psi_{x^2-y^2} - \psi_{3z^2 -r^2})/\sqrt{2}$
264: % and $ \psi_Y \equiv -(\psi_{x^2-y^2} + \psi_{3z^2 -r^2})/\sqrt{2}$ (see
265: %Ref. [5]).
266: %Ref. \onlinecite{allen}).
267: %From the symmetry of the $\psi_X$ and $\psi_Y$ orbitals it
268: %follows that at each site the total charge is unity.
269: %The unit cell needed to
270:
271: \begin{figure}
272: \resizebox*{3.0in}{2.7in}{\rotatebox{0}{\includegraphics{fig1.eps}}}
273: %\vspace*{0.5cm}
274: %\caption{Plot of $\chi_{2,3} (\pi,\pi)$ as a function of
275: %temperature at $t=0.1~ eV$.}
276: \label{scaling}
277: \end{figure}
278: \noindent
279: {\small{\bf Figure 1. }Plot of $\chi_{2,3} (\pi,\pi)$ as a function of
280: temperature at $t=0.1 ~ eV$.}
281: \vspace*{0.5cm}
282: \begin{figure}
283: \resizebox*{3.0in}{2.7in}{\rotatebox{0}{\includegraphics{fig2.eps}}}
284: %\vspace*{0.5cm}
285: %\caption{Plot of $\chi_{2,3} (q,q)$ for different values of $q$ (in radians)
286: %at T = 0.01 K and $t = 0.1 ~eV$.}
287: \label{fig2}
288: \end{figure}
289:
290: \noindent
291: {\small {\bf Figure 2.} Plot of $\chi_{2,3} (q,q)$ for different values
292: of ($0 ~\le q ~\le ~\pi $) at T = 0.01 K and $t = 0.1 ~eV$.} \\
293:
294: \noindent
295: and $\vec R_j$ being the position vector. Here it should be
296: pointed out that the order parameter
297: $\langle b^{\dagger}_{1j} b_{2j} +
298: b^{\dagger}_{2j} b_{1j} \rangle$
299: corresponds to the density difference of electrons in the
300: two orbitals
301: $\psi_X \equiv (\psi_{x^2-y^2} - \psi_{3z^2 -r^2})/\sqrt{2}$
302: and $ \psi_Y \equiv -(\psi_{x^2-y^2} + \psi_{3z^2 -r^2})/\sqrt{2}$ (see
303: Ref. [5]).
304: %Ref. \onlinecite{allen}).
305: From the symmetry of the $\psi_X$ and $\psi_Y$ orbitals it
306: follows that at each site the total charge is unity.
307: The unit cell needed to
308: compute the ground state energy
309: consists of two adjacent sites with the Brillouin zone being given by
310: $-\pi ~\le (k_x + k_y) ~\le \pi $
311: and $-\pi ~\le (k_x - k_y) ~\le \pi $. We
312: diagonalize
313: a $4 \times 4$ matrix at each momentum and integrate the lowest
314: two eigen energies over the Brillouin zone to obtain the ground state
315: energy.
316: The results of
317: %our calculations are
318: \begin{figure}
319: \resizebox*{3.0in}{2.7in}{\rotatebox{0}{\includegraphics{fig3.eps}}}
320: %\vspace*{0.5cm}
321: %\caption
322: \label{fig3}
323: \end{figure}
324: \noindent
325: {\small
326: {\bf Figure 3.} Dependence on dimensionless polaronic energy
327: ($A^2 /\omega_0 t$) of dimensionless ground state energy per site ($ E/t $)
328: for cooperative $Q_2$ and $Q_3$ modes.}\\
329:
330: \begin{figure}
331: \resizebox*{3.0in}{2.7in}{\rotatebox{0}{\includegraphics{fig4.eps}}}
332: %\vspace*{0.5cm}
333: %\caption{Variation of coefficients $c_{2,3}$ of
334: %ODW order parameters for $Q_2$ and $Q_3$ distortions as a function
335: %of the dimensionless polaronic energy
336: % ($A^2 /\omega_0 t$).}
337: \label{fig4}
338: \end{figure}
339: \noindent
340: {\small {\bf Figure 4.} Variation of coefficients $c_{2,3}$ of
341: ODW order parameters for $Q_2$ and $Q_3$ distortions as a function
342: of the dimensionless polaronic energy
343: ($A^2 /\omega_0 t$).}\\
344:
345: \noindent
346: our calculations are shown in Figs. 3 and 4. From Fig. 3
347: we see that the ground state energy corresponds to the $Q_2$ mode
348: with the difference in energy between the $Q_2$ only state and the $Q_3$
349: only state peaking at intermediate values of the dimensionless polaronic energy
350: ($A^2/\omega_0 t$).
351: For zero values and infinite values of
352: the polaronic energy both modes yield the same energy
353: because zero value implies no phononic coupling effect while infinite value
354: corresponds to localized polarons.
355: Thus for
356: large values of the polaronic energy,
357: the ground state energy is only slightly smaller than the polaronic energy.
358: Furthermore, from Fig. 4 we also see that as the polaronic energy increases
359: the values of $c_{2,3}$ increase and become unity around
360: $A^2/(\omega_0 t) \sim 2$ implying that for the $Q_3$ ($Q_2$) mode
361: $\psi_{x^2-y^2}$ ($\psi_X$) orbital is occupied
362: fully at one site with the $\psi_{3z^2 -r^2}$ ($\psi_Y$)
363: orbital being fully occupied at the adjacent sites.
364:
365:
366: In conclusion, we have studied orbital ordering for the ground state
367: of the undoped manganite systems. We find that the two-dimensional
368: orbital ordering, in the ferromagnetic planes of the observed
369: A-type antiferromagnetic
370: state, is governed by the wavevector $\vec{Q}$ with $Q_2$ JT mode
371: being cooperatively excited in the system. \\
372:
373: \vspace{0.5in}
374:
375: \noindent
376: {\bf References } \\
377: %\begin{thebibliography}{999}
378: %\bibitem{murakami}
379: \begin{itemize}
380: \item[[1]] Y. Murakami, J. P. Hill, D. Gibbs, M. Blume, I. Koyama,
381: M. Tanaka, H. Kawata, T. Arima, Y. Tokura, K. Hirota, and Y. Endoh,
382: Phys. Rev. Lett. {\bf 81},
383: 582 (1998).
384: %\bibitem{wollan}
385: \item[[2]] E. O. Wollan and W. C. Koehler,
386: Phys. Rev. {\bf 100}, 545 (1955).
387: %\bibitem{millis}
388: \item[[3]] A. J. Millis,
389: Phys. Rev. B {\bf 53},
390: 8434 (1996).
391: %\bibitem{dagotto}
392: \item[[4]] T. Hotta, S. Yunoki, M. Mayr, and E. Dagotto,
393: Phys. Rev. B {\bf 60},
394: R15009 (1999).
395: % \bibitem{allen}
396: \item[[5]] P. B. Allen and V. Perebeinos, Phys. Rev. B {\bf 60},
397: 10747 (1999).
398: %\bibitem{satpathy}
399: \item[[6]] Z. Popovic and S. Satpathy,
400: Phys. Rev. Lett. {\bf 84},
401: 1603 (2000).
402: % \bibitem{lsheng}
403: %L. Sheng and D. N. Sheng, Int. J. Mod. Phys. B
404: %{\bf 13}, 1397 (1999).
405: %\bibitem{khomskii}
406: \item[[7]] K. I. Kugel and D. I. Khomskii,
407: Sov. Phys. JETP {\bf 37},
408: 725 (1973).
409: %\bibitem{ishihara}
410: \item[[8]] S. Ishihara, J. Inoue, and S. Maekawa, Phys. Rev. B {\bf 55},
411: 8280 (1997).
412: %\bibitem{maekawa}
413: \item[[9]] L. Sheng and D. N. Sheng, Int. J. Mod. Phys. {\bf B 13}, 1397 (1999).
414: \item[[10]] S. Okamoto, S. Ishihara, and S. Maekawa,
415: Phys. Rev. B {\bf 65},
416: 144403 (2002).
417: %\bibitem{sudhakar}
418: \item[[11]] S. Yarlagadda,
419: Int. J. Mod. Phys. B {\bf 15}, 3529 (2001).
420: \end{itemize}
421: %\end{thebibliography}
422: %\end{references}
423: \end{document}
424:
425: \begin{figure}
426: \caption{Plot of $\chi_{2,3} (\pi,\pi)$ as a function of
427: temperature at $t=0.1 eV$.}
428: \label{fig1}
429: \end{figure}
430: \caption{Plot of $\chi_{2,3} (\pi,\pi)$ as a function of
431: temperature at $t=0.1 eV$.}
432:
433: \begin{figure}
434: \caption{Plot of $\chi_{2,3} (q,q)$ for different values of $q$ (in radians)
435: at T = 0.01 K and t = 0.1 eV.}
436: \label{fig2}
437: \end{figure}
438:
439: \begin{figure}
440: \caption{Dependence on dimensionless polaronic energy ($A^2 /\omega_0 t$)
441: of dimensionless ground state energy per site ($ E/t $)
442: for cooperative $Q_2$ and $Q_3$ modes.}
443: \label{fig3}
444: \end{figure}
445:
446: \begin{figure}
447: \caption{Variation of coefficients $c_{2,3}$ of
448: ODW order parameters for $Q_2$ and $Q_3$ distortions as a function
449: of the dimensionless energy
450: ($A^2 /\omega_0 t$).}
451: \label{fig4}
452: \end{figure}
453:
454: %\end{multicols}
455: \end{document}
456:
457:
458:
459: