cond-mat0103226/PRB.tex
1: %\documentstyle[zfangaps,prb,epsf,twocolumn]{zfang_revtex}
2: \documentstyle[preprint,aps,prb,epsf]{revtex}
3: 
4: \begin{document}
5: \draft
6: \title{Surface Magnetic Phase Diagram of Tetragonal Manganites}
7: 
8: \author{Zhong Fang$^{1}$, and Kiyoyuki Terakura$^{2,3}$}
9: \address{
10: $^{1}$JRCAT, Angstrom Technology Partnership, 1-1-4 Higashi, Tsukuba,
11: Ibaraki 305-0046, Japan \\
12: $^{2}$JRCAT, National Institute for Advanced Interdisciplinary Research,
13:         1-1-4 Higashi, Tsukuba, Ibaraki 305-8562, Japan \\
14: $^{3}$Tsukuba Advanced Computing Center, 1-1-4 Higashi, Tsukuba,
15: Ibaraki 305-8561, Japan
16: }
17: %\date{\today}
18: 
19: \maketitle
20: 
21: \begin{abstract}
22:   To gain insights into the fundamental and characteristic features of
23:   the surface of doped manganites, we constructed a general magnetic
24:   phase diagram of La$_{1-x}$Sr$_{x}$MnO$_3$ (001) surfaces in the
25:   plane spanned by $x$ and the bulk tetragonal distortion $c/a$, from
26:   the first-principles calculations. We found that the surfaces are
27:   quite different from the bulk in the sense that both the (La, Sr)O
28:   and MnO$_2$ terminated surfaces show strong tendency toward
29:   antiferromagnetism (A-type and C-type respectively).  The basic
30:   physics governing the phase diagram can be understood in terms of
31:   the surface orbital polarizations.  It is also found that the strong
32:   surface segregation of Sr atoms is mostly caused by the
33:   electrostatic interaction and will further enhance the tendency to
34:   surface antiferromagnetism.
35: \end{abstract}
36: %\pacs{}
37: %\maketitle
38: 
39: \newpage
40: 
41: \section{Introduction}
42: 
43: Over the last half a decade, the perovskite colossal magnetoresistive
44: (CMR) manganites have attracted intensive attention due to the
45: remarkably rich variety of structural, magnetic, transport and optical
46: properties~\cite{CMR} and are regarded as potentially important
47: materials for the next generation technology.  Such possibility may be
48: a strong motivation for the study of thin films and
49: superlattices~\cite{Izumi} of the manganites and the related
50: materials.  In these systems surfaces and interfaces will play
51: important roles.  However, even without appealing to thin films and
52: superlattices, surfaces and interfaces are involved in various
53: aspects.  The enhanced low-field magnetoresistance has been reported
54: in the polycrystalline samples due to the spin-dependent behaviors
55: across the grain boundaries~\cite{IMR}. The photoemission
56: data~\cite{PES}, which provide evidence for the half-metallicity, and
57: the scanning tunneling spectroscopy (STS)~\cite{STS} or microscopy
58: (STM)~\cite{STM}, which provide evidence for the spatial
59: phase-separation~\cite{phase-separation}, are all surface
60: sensitive. Therefore, the basic understanding of the surfaces of doped
61: manganites is an urgent and challenging problem. It is already
62: established that one of the most important implications in the physics
63: associated with the CMR manganites is the key roles of orbital degrees
64: of freedom (ODF)~\cite{ODF}, which couple strongly with the lattice,
65: charge and magnetic degrees of freedom (MDF).  We expect that the ODF
66: will play even more important roles on the surfaces due to the
67: lowering in symmetry and dimensionality.
68: 
69: Detailed experimental and theoretical studies on the surfaces of CMR
70: manganites have started only very recently~\cite{Ma,Peng,Pickett,Brey,LCMO}
71: and the results obtained so far are not yet sufficient to establish
72: any general picture about the surface phase diagram.  In the
73: experimental side, it is still difficult to prepare well defined
74: surfaces. In the theoretical side, on the other hand, only some
75: limited phase space and limited conditions have been taken into
76: account.  Although it is not possible to consider the complete phase
77: space and all possible complications conceivable in real systems, it
78: is very important to construct a general qualitative picture for the
79: surface phase diagram by taking account of possible phases and some
80: important experimental conditions.  This is what we aim in the present
81: work by taking La$_{1-x}$Sr$_x$MnO$_3$ (LSMO) as a canonical system of
82: doped manganites~\cite{CE}.
83: 
84: First, since it is possible to control the terminations by chemical
85: treatment of the surface~\cite{kawasaki} or by controlling the
86: chemical potential~\cite{BTO}, consideration of two possible
87: terminations of (001) surface is important. It is natural to expect
88: that the surface effect will be strong for the MnO$_2$ termination,
89: but weak for the (La, Sr)O termination, due to the loss of part of the
90: ligand of Mn in former case.  Although the expectation is
91: qualitatively true, the change in the electrostatic potential for SrO
92: termination can also produce significant change in the electronic
93: structure. Second, the lattice deformation induced by the substrates
94: can produce cooperative stabilization of orderings in ODF and MDF,
95: leading to strong anisotropy in the thin film~\cite{Konishi,Phase}.
96: Therefore, in the present work, we will present a general surface
97: phase diagram of LSMO, for both the (La, Sr)O and MnO$_2$ terminated
98: (001) surfaces, as functions of the hole doping $x$ and the bulk
99: tetragonal distortion $c/a$ ratio induced by the substrates.  We found
100: that the (La, Sr)O terminated surface shows strong tendency toward the
101: A-type antiferromagnetic (AF) state, while the surface phases of
102: MnO$_2$ termination are dominated by the C-type AF state. The basic
103: physics behind is the ordering in the surface ODF of Mn $e_g$ states.
104: Third, we found that the Sr atoms should seriously segregate towards
105: the surfaces due to the electrostatic interaction. The general
106: tendency of surface segregation will further favor the surface AF
107: states.
108: 
109: 
110: \section{Calculation Methods}
111: 
112: The present work is based on the first-principles electronic structure
113: calculations, which adopt the PBE version~\cite{PBE} of generalized
114: gradient approximation (GGA). The Kohn-Sham equation is solved by
115: using the pseudopotential technique. The Mn(3d) and the O(2p) states
116: are treated by the ultra-soft pseudopotential~\cite{USPP}, while the
117: norm-conserving scheme~\cite{NCPP} is used for other states.  The
118: cutoff energies for the plane-wave expansions of wavefunction and
119: charge density are 30 Ry and 200 Ry respectively. La$_x$Sr$_{1-x}$ is
120: treated as a virtual atom by the virtual crystal approximation
121: (VCA)~\cite{VCA}. In present systems, the valence electronic states
122: near and below the Fermi level are contributed mostly by Mn and O, the
123: states coming from La and Sr are far above the Fermi level. Therefore,
124: the simple VCA can provide a reasonable description of the
125: systems. The validity of these techniques was well demonstrated in our
126: previous calculations for the bulk~\cite{Phase}. We use a repeated
127: slab geometry, which includes five MnO$_2$ layers with mirror symmetry
128: in the central MnO$_2$ layer.  As the MnO$_2$ layer and the (La, Sr)O
129: layer are stacked alternately along the surface normal, i.e.,
130: $z$-axis, nine and eleven atomic layers in total are included in the
131: unit cell for the MnO$_2$ and (La, Sr)O terminations, respectively.  A
132: vacuum region of $12\sim 13$\AA\ is included to separate the slabs.
133: Several magnetic states, i.e., ferromagnetic (FM), A-type AF and
134: C-type AF states, are calculated in the present study. In the A-type
135: AF state, the magnetic moment of Mn are aligned ferromagnetically in
136: the $ab$-plane (slab plane) and these FM layers are coupled
137: antiferromagnetically along the $c$ direction (surface normal). In the
138: C-type AF state, on the other hand, the FM chains along the $c$
139: direction are coupled antiferromagnetically.  In the surface phase
140: diagrams in Fig.1, the above definition is applied to the slab. A
141: simple schematic description about the spin configurations of surface
142: A- and C-type AF states can be found in Fig.2.  To accommodate the
143: C-type AF states, we use the planar c(2x2) unit cell. The atomic
144: positions and magnetic states are fully optimized for each of given
145: $ab$-plane lattice constants, which define the bulk $c/a$ ratio since
146: the bulk volume is given for each doping $x$.  By choosing the lowest
147: energy state among FM, A-type AF and C-type AF states for each
148: $ab$-plane lattice constant and each doping $x$, we can construct the
149: surface phase diagram.
150: 
151: 
152: \section{Results and discussion}
153: 
154: We show the calculated surface phase diagrams in
155: Fig.~\ref{figure:phase} compared with the bulk phase diagram which was
156: obtained in our previous calculations~\cite{Phase}.  The surface phase
157: diagrams for the (La, Sr)O and MnO$_2$ terminations are indicated by
158: the red and green lines respectively.  Clearly for the MnO$_2$
159: termination, the surface phases are dominated by the C-type AF state
160: for a wide range of doping $x$ and $c/a$ ratio, except a tiny FM
161: region in the very small $c/a$ ratio and low doping case.  On the
162: other hand for the (La, Sr)O termination, the surface shows rich
163: phases similar to the bulk case but with a significant shift.
164: Compared with the bulk phase diagram (indicated by black lines and
165: characters), the main change for the surfaces is the up- and downward
166: shifts of the phase boundaries for (La, Sr)O and MnO$_2$ termination
167: respectively.  It should be noted that, since non-equal numbers of
168: (La, Sr)O and MnO$_2$ layers are included in our unit cell, the
169: effective doping $x_{\rm eff}$ of the central MnO$_2$ layer is
170: slightly different from the formula doping $x$.  By calculating the
171: number of $e_g$ electrons, we estimate that the $x_{\rm eff}$ is
172: smaller (larger) than $x$ by about 0.04 (0.05) for (La, Sr)O (MnO$_2$)
173: termination.  For the sampling point ($x=0.5$ and $c/a=1.0$, indicated
174: by the blue star in Fig.~\ref{figure:phase}), the calculated total
175: energies and some physical parameters are summarized in Tables I and
176: II. It should be also noted here that the presence of interfaces
177: between the surface and bulk regions may modify our calculated phase
178: diagram slightly. If the penetration of the perturbation in the
179: magnetic state on surface is deeper than half of the slab thickness,
180: the surface phase diagram may depend on the slab thickness. In order
181: to check these effects, we performed calculations in which the spin
182: structures of the central MnO$_2$ layer are constrained to be the bulk
183: ones. We found that the spin structures in the first two surface
184: layers (of both sides of the slabs) are little affected by the
185: constraint in the central layer. We should also point out that the
186: possible non-collinear magnetic configurations are not taken into
187: account in the present calculation. However, even if we confine
188: ourselves to collinear spin configurations, we can still gain
189: important valuable insights into the basic trend in surface magnetic
190: states by our calculated phase diagram.
191: 
192: The occupation imbalance between two $e_g$ orbitals, namely $3z^2-r^2$
193: and $x^2-y^2$ orbitals in doped manganites, is the key concept to
194: understand the rich phase diagrams~\cite{CMR,Phase,Igor}. On the
195: surfaces, the orbital polarization is certainly very important because
196: of the change in crystal field, hybridization and surface lattice
197: relaxation. For the (La, Sr)O termination, the environment of the 2nd
198: surface Mn changes only from the 2nd nearest neighbor layer, which is
199: negatively charged MnO$_2$ layer.  Missing of this layer in the vacuum
200: side produces an attractive potential leading to downward shift of the
201: 2nd surface Mn $3d$ bands. However, the strengthened $pd\sigma$
202: hybridization due to the missing Mn atom on top of the surface O atom
203: will cause the upward shift of the $3z^2-r^2$ state compared with the
204: $x^2-y^2$ state.  Although the surface relaxation in the interlayer
205: distance, which is very small, will modify the details quantitatively,
206: it will not affect the qualitative features. For the sampling
207: structure in Fig.~1, the Mn-O bond length between the 1st and 2nd
208: layers changes only by $-0.2\%$ (see Table II). For the MnO$_2$
209: termination, the most dramatic effect is the significant downward
210: shift of the $3z^2-r^2$ state due to the reduction in $pd\sigma$
211: hybridization caused by the missing ontop oxygen.  For other orbitals,
212: missing of the positively charged (La, Sr)O layer produces repulsive
213: potential.  In this case, the surface relaxation is very large because
214: of the loss of ligand oxygen atoms.  However, the atoms are relaxed in
215: such a way that the topmost Mn-O bond length along surface normal
216: direction is elongated to further push down the $3z^2-r^2$ orbital.
217: Therefore the net effect of surface orbital polarization is that the
218: occupation of the $x^2-y^2$ ($3z^2-r^2$) orbital is enhanced for (La,
219: Sr)O (MnO$_2$) termination.  The orbital population ratio $n_{\rm
220:   3z^2-r^2}/n_{\rm x^2-y^2}$ for the sampling structures in Fig.~1 are
221: estimated as 0.74 and 2.16 for the surface Mn sites of (La, Sr)O and
222: MnO$_2$ terminations respectively. The surface orbital polarization of
223: MnO$_2$ termination is much stronger than that of (La, Sr)O
224: termination.  Figure 2 shows the charge distributions (from the Fermi
225: level to 0.8 eV below) for two different terminations of the sampling
226: point. The surface orbital polarizations are clearly seen in each
227: case.
228: 
229: The change in occupation will change the competition between the
230: double exchange (DE) and the superexchange (SE). For the less than
231: half-filled majority-spin $e_g$ bands, the more (less) populated are
232: the orbitals, the stronger (weaker) are the DE interactions among
233: these orbitals.  Therefore, the region of the A-type (C-type) AF state
234: becomes wider for the (La, Sr)O (MnO$_2$) termination because of
235: higher population of the $x^2-y^2$ ($3z^2-r^2$) orbital.
236: A. Filippetti and W. E. Pickett~\cite{Pickett} studied the magnetic
237: properties of MnO$_2$ terminated (001) surface of
238: La$_{1-x}$Ca$_x$MnO$_3$ (LCMO) with particular doping $x$
239: (=0.5). Their observation of the stability of the FM state is due to
240: the use of the (1x1) surface unit cell.  Our calculations for LSMO
241: with the same doping ($x=0.5$, $c/a=1.0$) give the same order of total
242: energy difference between the surface FM and A-type AF states (Table
243: I). However, the surface C-type AF state has much lower energy
244: compared with the above two states.
245: 
246: 
247: On the real surfaces of CMR manganites, the surface segregation on the
248: perovskite A-sites may be another important factor. It was
249: experimentally suggested that the Ca content in the surface layers of
250: LCMO is dramatically enhanced~\cite{LCMO} for both terminations. The
251: basic questions here are: 1)why does surface segregation happen? 2)how
252: will it affect our phase diagram?  In the analysis of surface
253: segregation, we need the bulk part as a particle reservoir.  The
254: present type slab calculation has, therefore, rather severe
255: restriction in this context.  In the following, we will do a simple
256: analysis in order to gain insight into the fundamental aspects of the
257: problem.  We artificially locate a pure SrO layer at different
258: positions in our unit cell keeping the Sr content $x$ in all other
259: (La, Sr)O layers as a given value and calculate the total energy as a
260: function of the SrO layer position. The calculated results shown in
261: Fig. 3 clearly suggest that the stable position of the SrO layer is
262: the surface for the (La, Sr)O termination or the 2nd surface for the
263: MnO$_2$ termination.  The result is qualitatively consistent with the
264: experimental observation for LCMO~\cite{LCMO}.  The main reason for
265: the stability of the surface (or subsurface) SrO layer is the
266: electrostatic interaction.  SrO layer is nominally charge neutral,
267: while the nominal charge of LaO is +1.  In the bulk the ionized object
268: is stabilized by the electrostatic interaction with the counter ions.
269: This stability mechanism is weakened on the surface.  Therefore, the
270: neutral object tends to be located on the surface keeping the charged
271: objects inside the bulk.  For the MnO$_2$ termination, the presence of
272: SrO layer at the 2nd surface will make the surface MnO$_2$ layer
273: nearly neutral also.  In Fig.~3(b), we show that the contribution
274: coming from the electrostatic energy is responsible to the stability
275: of the SrO layer at the surface.  As the nominal doping $x$ increases,
276: the effect of Sr segregation should be weakened as is actually
277: demonstrated in Fig.~3.  Although the present analysis is only for
278: limited configurations in which one particular layer of the system has
279: full Sr segregation.  Nevertheless, the above consideration about the
280: mechanism of the stability of surface SrO layer clearly suggests a
281: rather general tendency of surface segregation of the neutral objects.
282: By taking account of the surface segregation of Sr, the effective
283: doping on the surface becomes larger than the bulk doping. This will
284: again lead to stronger stability of surface AF ordering.
285: 
286: 
287: \section{Summary}
288: 
289: 
290: In summary, we calculated the surface magnetic phase diagram of
291: tetragonal manganite LSMO as functions of hole doping $x$ and the bulk
292: tetragonal distortion $c/a$.  The (001) surfaces of tetragonal
293: manganites show clear tendency towards A-type and C-type AF states for
294: (La, Sr)O and MnO$_2$ terminations respectively though the surface
295: A-type AF state may be hard to distinguish experimentally from the
296: surface FM state.  The basic physics governing the phase diagram is
297: explained in terms of the orbital polarization induced by surface
298: effects. Strong surface segregation of Sr atoms is caused by the
299: electrostatic interaction and further favors the tendency to surface
300: antiferromagnetism.
301: 
302: \section{Acknowledgments}
303: 
304: The authors thank Professors Y. Tokura, E. W. Plummer, M. Kawasaki and
305: R. Matzdorf for valuable comments and for providing us with their
306: experimental data. The present work is partly supported by NEDO.
307: 
308: 
309: \begin{thebibliography}{50}
310: 
311: \bibitem{CMR} See, for example, {\it Colossal-Magnetoresistive
312:       Oxides}, edited by Y.~Tokura (Gordon \& Breach Science Publishers,
313:     1999).
314: 
315:   \bibitem{Izumi} See, for example, M. Izumi {\it et al.},
316:     Phys. Rev. B {\bf 61}, 12187(2000).
317: 
318:   \bibitem{IMR} H. Y. Hwang {\it et al.}, Phys. Rev. Lett. {\bf 77},
319:     2041(1996); S. Lee {\it et al.}, Phys. Rev. Lett. {\bf 82},
320:     4508(1999).
321: 
322:   \bibitem{PES} J. H. Park {\it et al.}, Nature {\bf 392}, 794(1998);
323:     {\it ibid}, Phys. Rev. Lett. {\bf 81}, 1953(1998); D. D. Sarma
324:     {\it et al.}, Phys. Rev. B {\bf 53}, 6873(1996)
325: 
326:   \bibitem{STS} M. F\"{a}th {\it et al.}, Science {\bf 285},
327:     1540(1999).
328: 
329: \bibitem{STM} R. Matzdorf and E. W. Plummer, private communication.
330: 
331: \bibitem{phase-separation} A. Moreo, S. Yunoki and E. Dagotto, Science
332:     {\bf 283}, 2034(1999), and references therein.
333: 
334: \bibitem{ODF} M. Cyrot and C. Lyon-Caen, J. Phys (France) {\bf 36},
335:     253(1975); Y. Ito and J. Akimitsu, J. Phys. Soc. Jpn. {\bf 40},
336:     1333(1976); {\it ibid}, {\bf 40}, 1621(1976).
337: 
338:   \bibitem{Ma} J. X. Ma {\it et al.}, (to be published); T. Kawai,
339:     Bulle. Am. Phys. Soc. {\bf 45}, 879(2000).
340: 
341:   \bibitem{Peng} H. B. Peng {\it et al.}, Phys. Rev. Lett. {\bf 82},
342:     362(1999).
343: 
344: \bibitem{Pickett} A. Filippetti and W. Pickett, Phys. Rev. Lett. {\bf
345:       83}, 4184(1999); {\it ibid}, Cond-Mat/0001373.
346: 
347: \bibitem{Brey} M. J. Calder\'{o}n, L. Brey and F. Guinea, Phys. Rev. B
348:     {\bf 60}, 6698(1999).
349: 
350:   \bibitem{LCMO} J. Choi {\it et al.}, Phys. Rev. B {\bf 59},
351:     13453(1999); H. Duli {\it et al.} Phys. Rev. B (to be published).
352: 
353: \bibitem{CE} Since the tilt of MnO$_6$ octahedra in
354:     La$_{1-x}$Sr$_x$MnO$_3$ is rather small, we can treat as a first
355:     approximation the system as tetragonal. Therefore the complicated
356:     charge ordering state does not occur. It is demonstrated in our
357:     previous study for the bulk that the essential physics with respect
358:     to the ODF can be caught in this simple tetragonal lattice.
359: 
360:   \bibitem{kawasaki} M. Kawasaki {\it et al.}, Science {\bf 266},
361:     1540(1994).
362: 
363:   \bibitem{BTO} J. Padilla {\it et al.}, Phys. Rev. B {\bf 56},
364:     1625(1997).
365: 
366:   \bibitem{Konishi} Y. Konishi {\it et al.}, J. Phys.  Soc. Jpn. {\bf
367:       68}, 3790(1999).
368: 
369:   \bibitem{Phase} Z. Fang {\it et al.}, Phys. Rev. Lett. {\bf 84},
370:     3169(2000).
371: 
372:   \bibitem{Igor} I. Solovyev {\it et al.} Phys. Rev. Lett. {\bf 76},
373:     4825 (1996).
374: 
375:   \bibitem{PBE} J. P. Perdew, K. Burke and M. Ernzerhof,
376:     Phys. Rev. Lett. {\bf 77}, 3865(1996).
377: 
378:   \bibitem{USPP} D. Vanderbilt, Phys. Rev. B {\bf 41}, 7892 (1990).
379: 
380:   \bibitem{NCPP} N. Troullier and J. L. Martins, Phys. Rev. B {\bf
381:       43}, 1993 (1991).
382: 
383:   \bibitem{VCA} J. Lee {\it et al.} J. Korean Phys. Soc. {\bf 33},
384:     55(1998).
385: 
386: \end{thebibliography}
387: 
388: 
389: 
390: \newpage
391: \begin{figure}[t]
392:       \centering
393:       \leavevmode  \epsfxsize=120mm \epsfbox[120 230 450 530]{Fig1.ps}
394:       \caption{The calculated phase diagrams of La$_{1-x}$Sr$_x$MnO$_3$
395:         in a plane of hole doping $x$ and $c/a$ ratio. Here the $c/a$
396:         ratio can be defined by the $ab$-plane lattice constant as the
397:         bulk volume is given for each doping $x$. The red and green
398:         colors correspond to the surface phase diagrams for the (La,
399:         Sr)O and the MnO$_2$ terminations respectively, while the
400:         black color corresponds to the bulk phase diagram. The
401:         denotations F, A , C and G mean FM, A-type AF, C-type AF and
402:         G-type AF states respectively.}
403:       \label{figure:phase}
404: \end{figure}
405: 
406: 
407: \newpage
408: \begin{figure}[t]
409:       \centering
410:       \leavevmode  \epsfxsize=120mm \epsfbox[90 20 880 680]{Fig2.ps}
411:       \caption{The occupied (0.8 eV below the Fermi level) electronic
412:         charge distributions for the two terminations corresponding to
413:         the sampling point in Fig. 1. Red color denotes high charge
414:         density.  Between two neighboring Mn sites, there is small
415:         charge density from oxygen sites. The spin configurations of
416:         Mn sites are indicated by the red arrows.}
417:       \label{figure:charge}
418: \end{figure}
419: 
420: 
421: \newpage
422: \begin{figure}[t]
423:       \centering
424:       \leavevmode  \epsfxsize=120mm \epsfbox[100 100 500 600]{Fig3.ps}
425:       \caption{The calculated total energies as a function of SrO layer
426:         position keeping the Sr content $x$ in all other (La, Sr)O
427:         layers as a given value. The zero in the numbering of layers
428:         corresponds to the center of our unit cell, while MnO$_2$
429:         layers are located on even number of layers.}
430: \end{figure}
431: 
432: 
433: 
434: 
435: \begin{table}[t]
436: \caption{The calculated total energies, per surface (1x1) cell,
437:     for the sampling point ($x=0.5$, $c/a=1.0$) in Fig.~1.}
438: \begin{tabular}{c|c|c|c}
439:     &Surface F   &Surface A  &Surface C  \\  \hline
440: (La,Sr)O termination  &22 meV  &0 meV   &61 meV    \\
441: MnO$_2$ termination &124 meV  &203meV  &0 meV  \\
442: \end{tabular}
443: \end{table}
444: 
445: 
446: \begin{table}[t]
447: \caption{Some calculated parameters corresponding to the sampling point
448:   in Fig.~1 for the two terminations.  $\Delta d_{\rm Mn-O}$ means the
449:   change of the topmost Mn-O bond length along surface normal; $M_{\rm c}$
450:   and $M_{\rm s}$ denote the magnetic moments of Mn sites at the central and
451:   the surface layers respectively; $n_{\rm 3z^2-r^2}/n_{\rm x^2-y^2}$
452:   defines the ratio of occupation numbers for two $e_g$ orbitals of
453:   surface Mn.}
454: \begin{tabular}{c|c|c|c|c}
455:     &$\Delta d_{\rm Mn-O}$ &$M_{\rm c}$ &$M_{\rm s}$ &$n_{\rm 3z^2-r^2}/n_{\rm x^2-y^2}$ \\
456:     \hline (La,Sr)O A-AF &-0.2\% &3.11 &3.12 &0.74 \\ MnO$_2$ C-AF
457:     &+2.7\% &2.87 &3.17 &2.16 \\
458: \end{tabular}
459: \end{table}
460: 
461: 
462: 
463: \end{document}
464: 
465: 
466: