1: %%LN6596
2: %%Keywords: crystal field, spin-orbit coupling, Jahn-Teller effect, LaCoO$_3$
3:
4:
5: \documentstyle[amssymb,aps,12pt]{revtex}
6: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
7: %TCIDATA{TCIstyle=Article/art2.lat,aps,revtex}
8:
9: %TCIDATA{OutputFilter=LATEX.DLL}
10: %TCIDATA{LastRevised=Thursday, January 10, 2002 15:38:11}
11: %TCIDATA{<META NAME="GraphicsSave" CONTENT="32">}
12: %TCIDATA{Language=American English}
13:
14: \begin{document}
15: \title{Influence of spin-orbit interactions on the cubic crystal-field
16: states of the $d^{4}$ system*}
17: \author{R.J. Radwanski}
18: \address{Center for Solid State Physics, S$^{nt}$Filip 5,31-150Krakow,Poland%
19: \\
20: Institute of Physics, Pedagogical University, 30-084 Krakow, Poland}
21: \author{Z. Ropka}
22: \address{Center for Solid State Physics, S$^{nt}$Filip 5,31-150Krakow,Poland.%
23: \\
24: e-mail: sfradwan@cyf-kr.edu.pl, http://www.css-physics.edu.pl\\
25: Original submission: 27.01.1997}
26: \maketitle
27:
28: \begin{abstract}
29: It has been shown that for the highly-correlated $d^{4}$ electronic system
30: the spin-orbit interactions produce, even in case of the cubic crystal-field
31: interactions, a singlet ground state. Its magnetic moment grows rapidly with
32: the applied magnetic field approaching 4 $\mu _{B}$ for the $E_{g}$ state,
33: but only 3 $\mu _{B}$ for the $T_{2g}$ state. The applicability of the
34: present results to the Mn$^{3+}$ ion in LaMnO$_{3}$ is discussed.
35: \end{abstract}
36:
37: \pacs{75.30.Gw; 71.70}
38: \date{}
39:
40: 4 $d$ electrons of paramagnetic ions form a highly-correlated electron
41: system 3$d^{4}$ as their orbital and spin movements are correlated within
42: the incomplete outer shell. This Letter has been motivated by the recent
43: enormous increase of interest in La$_{1-x}$Ca$_{x}$MnO$_{3}$ compounds \cite%
44: {1,2,3}. Surely a large magnetoresistance effect, that enables wide
45: technical applications, vivifies interest to La$_{1-x}$Ca$_{x}$MnO$_{3}$
46: compounds. In the parent LaMnO$_{3}$ compound the manganese atoms occur in
47: the trivalent state anticipated from valences La$^{3+}$Mn$^{3+}$O$_{3}^{2-}$%
48: . This compound is an antiferromagnetic insulator with T$_{N}$ of 140 K and
49: the low-temperature magnetic moment of about 3.9 $\mu _{B}$ \cite{4}. The
50: presently discussed electronic structure\cite{1,2,3} of LaMnO$_{3}$
51: completely ignores the effect of spin-orbit interactions and the 5-fold spin
52: degeneracy of the $^{5}D$ term relevant to the 3$d^{4}$ electronic system.
53:
54: The aim of the present Letter is to analyze the influence of spin-orbit
55: interactions on the localized states of the 3$d^{4}$ system produced by
56: crystal-field (CEF) interactions of the cubic symmetry. The 3$d^{4}$ system
57: is realized in paramagnetic ions like Mn$^{3+}$, Cr$^{2+}$, Ru$^{4+}$ (4$%
58: d^{4}$). The Mn$^{3+}$ state is realized in the antiferromagnetic insulator
59: LaMnO$_{3}$ that recently has got large interest as the parent compound of La%
60: $_{2/3}$Ca$_{1/3}$MnO$_{3}$ shows the Ca-induced ferromagnetism and the
61: enormous magnetoresistance effect at the vicinity of room temperature.
62:
63: One can be surprised but the influence of the spin-orbit interactions has
64: not been systematically studied though CEF effect is already known by more
65: than 65 years, after the works of Bethe and Kramers at 1930. It is only
66: generally accepted that in 3$d$ ions spin-orbit interactions are much weaker
67: than CEF interactions - such the statement, however, has led to the
68: ignorance of spin-orbit interactions. Nowadays performed band-structure
69: calculations usually ignore spin-orbit interactions with a (wrong)
70: justification by the quenching-orbital-moment effect. The forgetting of the
71: spin-orbit interactions is, however, very dangerous. It will be shown in
72: this Letter that they produce a number of new physical phenomena like the
73: singlet ground state, for instance.
74:
75: The physical situation of the 3$d^{n}$ system of a 3$d$-transition-metal ion
76: is taken to be accounted for by considering the single-ion Hamiltonian
77: containing the electron-electron d-d interactions H$_{el-el}$, the
78: crystal-field H$_{CEF}$, spin-orbit H$_{s-o}$ and Zeeman H$_{z}$
79: interactions:
80:
81: $\ H_{d}=H_{el-el}+H_{CEF}+H_{s-o}+H_{z}$ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \
82: \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \
83: (1)
84:
85: The electron-electron and spin-orbit interactions are intra-atomic
86: interactions whereas crystal-field and Zeeman interactions account for
87: interactions of the unfilled 3$d$ shell with charge and spin surrounding.
88: These interactions are written in the decreasing-strength succession.
89:
90: In a zero-order approximation the electron-electron correlations among the $%
91: d $ electrons are accounted for by phenomenological Hund's rules that yield
92: for the 3$d^{4}$ electron configuration the term $^{5}D$ with $S$ = 2 and $L$%
93: = 2 as the ground term. Under the action of the crystal field of the cubic
94: symmetry, the term $^{5}D$ splits \cite{5} into the orbital triplet $\Gamma
95: _{5}$, denoted also as $T_{2g}$, and the orbital doublet $\Gamma _{3}$ ($%
96: E_{g}$). Crystal field (CEF) effects of 3$d$ ions the reader is asked to
97: consult.
98:
99: The $^{5}D$ term is 25-fold degenerated. The 5-fold degeneration occurs with
100: respect to the orbital degree of freedom and each orbital state contains the
101: 5 spin-degree of freedom. This situation is accounted for if one considers
102: the single-ion Hamiltonian (1) written in the explicit form for the Hund's
103: rule $|LS\rangle $ term:
104:
105: $H_{d}=B_{4}(O_{4}^{0}+5O_{4}^{4})+\lambda L\cdot S+\mu
106: _{^{B}}(L+g_{e}S)\cdot B_{ext}$ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \
107: \ \ \ \ \ \ (2)
108:
109: The first term is the cubic CEF Hamiltonian with the Stevens operators $%
110: O_{n}^{m}$ that depend on the orbital quantum numbers $L$ and $L_{z}$ and B$%
111: _{4}$ is CEF parameter. The second term is the spin-orbit interactions where
112: $\lambda $ is the intra-atomic spin-orbit coupling. The last term accounts
113: for the influence of the magnetic field, the externally applied in the
114: present case. For simplicity the free-electron $g_{e}$ value is taken as 2.0
115: instead of 2.0023. The calculations of the many-electron states of the 3$%
116: d^{4}$ system have been performed \cite{6} by the diagonalization of a 25x25
117: matrix associated with the Hamiltonian (2) considered in the $%
118: |LSL_{z}S_{z}\rangle $ basis.
119:
120: As a result of the diagonalization we obtain the energies of the 25 states
121: and the eigenvectors containing information e.g. about the magnetic
122: properties. These magnetic characteristics are computationally revealed
123: under the action of the external magnetic field. Due to the spin-orbit
124: coupling the states are no longer purely cubic orbital states $%
125: |L,L_{z}\rangle $ denoted as $|x^{2}-y^{2}\rangle ,$ $|3z^{2}-r^{2}\rangle ,$
126: $|xy\rangle ,$ $|zx\rangle ,$ $|yz\rangle $ states. In case of purely cubic
127: CEF interactions two first states form the orbital-doublet $E_{g}$ state
128: whereas the three last states form the orbital-triplet $T_{2g}$ state. These
129: states are shown in Fig. 1b for an exemplary value B$_{4}$=-10 K. (In
130: reality, B$_{4}$ amounts to 100 K - 300 K; here this small value of B$_{4}$
131: has been taken to show in the same scale the crystal-field and spin-orbit
132: interactions). In fig. 1c the effect of the spin-orbit interactions with $%
133: \lambda $ of +120 K ( +85 cm$^{-1}$, such the value is given for the Mn$%
134: ^{3+} $ ion in ref. \cite{5} on p. 399). Inspecting Fig. 1 one can see that
135: spin-orbit interactions i) largely remove the degeneracies of the 10-fold $%
136: E_{g}$ state and of the 15-fold $T_{2g}$, state \ and yield ii) the singlet
137: ground state. This singlet ground state is produced always, i.e.
138: independently on the sign of the parameter B$_{4}$ as seen in Figs 1 and 2.
139: In Fig. 2 the same results are shown but for a positive value B$_{4}$ = +10
140: K. For a B$_{4}$%
141: %TCIMACRO{\TEXTsymbol{>}}%
142: %BeginExpansion
143: \mbox{$>$}%
144: %EndExpansion
145: 0 the orbital triplet $T_{2g}$ is the CEF ground state whereas for B$_{4}$%
146: %TCIMACRO{\TEXTsymbol{<}}%
147: %BeginExpansion
148: \mbox{$<$}%
149: %EndExpansion
150: 0 the orbital doublet $E_{g}$ is the ground state.
151:
152: The reason for the formation of the singlet ground state is very simple. It
153: is, somehow, the manifestation of the 3$^{rd}$ Hund rule, that becomes valid
154: in case of the strong spin-orbit coupling (the situation realized in
155: rare-earth ions). The strong spin-orbit coupling is realized in Figs 1 and 2
156: by weakening of CEF interactions (going from left to right). The 3$^{rd}$
157: Hund rule says that the ground multiplet of the ($L$=2, $S$=2) term is $%
158: ^{5}D_{0}$ with $J$=$L$-$S$=0 as the incomplete 3$d$ shell is less than half
159: filled. The $^{5}D_{0}$ multiplet is a singlet. Such the singlet $^{5}D_{0}$
160: multiplet occurs for the $d^{4}$ system where $\lambda $%
161: %TCIMACRO{\TEXTsymbol{>}}%
162: %BeginExpansion
163: \mbox{$>$}%
164: %EndExpansion
165: 0. The formation of the localized singlet ground state by the spin-orbit
166: interactions in the $d^{4}$ system, even in case of the doublet $E_{g}$
167: state, is a new and extremely surprising result owing to a general
168: conviction that the orbital-doublet state ($E_{g}$) is unaffected by the
169: spin-orbit coupling \cite{7}. It comes from this Letter that this is true
170: for the $^{2}D$ term applicable to the $d^{l}$ and $d^{9}$ system but not
171: for the $d^{4}$ and $d^{6}$ systems where $S$ = 2. This effect has been
172: signalized by Abragam and Bleaney \cite{5}, p. 435, who wrote that the $%
173: E_{g} $ state can be split by second-order effects, in the perturbation
174: method, of the s-o coupling. Now it has been proved by direct calculations
175: that these second-order effects have to be taken into account from the
176: beginning. For comparison of the effect of the spin-orbit coupling on the
177: cubic CEF states as has been calculated by Abragam and Bleaney and in the
178: present Letter the reader is asked to compare Fig. 1c and Fig. 7.19b on
179: p.404 of ref. \cite{5}. It allows for revealing of the novelty of the
180: present calculations. As mentioned already the Figs 1 and 2 have been made
181: for small values of B$_{4}$ parameters. The CEF splitting amounts usually to
182: $\thicksim $25000 K, the spin-orbit splitting of the $T_{2g}$ state - to 600
183: K (5$\lambda $). Then the splitting of the $E_{g}$ state amounts to 14 K
184: only.
185:
186: It is worth noting that despite of the singlet nature of the ground state
187: its magnetic moment grows rapidly with applying field as is shown in Fig. 3.
188: In case of the $E_{g}$ CEF ground state this induced moment approaches the
189: saturation to an integer value of 4 $\mu _{B}$, Fig. 3, curve 1. Such the
190: value is somehow manifestation of the well-known quenching of the orbital
191: moment in 3$d$ paramagnetic ions as just this value one expects for 4
192: electrons with the spin-moment only. For the $T_{2g}$ state the moment
193: saturates at level of 3 $\mu _{B}$ as one can see from Fig. 3, curves 3 and
194: 4. This strong influence is due to van Vleck term and occurs so easily
195: because of the presence of a large number of low-energy localized states
196: close to the singlet ground state.
197:
198: In conclusion, it has been shown that the spin-orbit interactions produce
199: for the highly-correlated $d^{4}$ system, even in case of the cubic CEF
200: interactions, the fine electronic structure with a singlet ground state both
201: for the octahedral (B$_{4}$%
202: %TCIMACRO{\TEXTsymbol{<}}%
203: %BeginExpansion
204: \mbox{$<$}%
205: %EndExpansion
206: 0) and tetrahedral (B$_{4}$%
207: %TCIMACRO{\TEXTsymbol{>}}%
208: %BeginExpansion
209: \mbox{$>$}%
210: %EndExpansion
211: 0) sites. As spin-orbit interactions are intra-atomic interactions it means
212: that one always deals with the singlet ground state in case of cubic CEF
213: interactions. Despite of its singlet nature the calculations show that its
214: magnetic moment grows rapidly with the applied magnetic field approaching
215: for the $E_{g}$ state (B$_{4}$%
216: %TCIMACRO{\TEXTsymbol{<} }%
217: %BeginExpansion
218: \mbox{$<$}
219: %EndExpansion
220: 0) a value of 4.0 $\mu _{B}$. Such the state is expected \cite{8} to be
221: realized in LaMnO$_{3}$ for which B$_{4}$ is anticipated for -200 K (and the
222: $i$ ratio of 0.05) from the measured $d-d$ separation of 2.0-2.2 eV \cite%
223: {2,3}. The negative sign of B$_{4}$ is also consistent with the octahedral
224: ligand surrounding of the Mn$^{3+}$ ion in LaMnO$_{3}$. From the value of
225: 140 K for T$_{N}$ in LaMnO$_{3}$ a molecular field of about 100 T has been
226: inferred. The present calculations clearly prove that for the meaningful
227: discussion of properties of 3$d$ paramagnetic-ion compounds the spin-orbit
228: and CEF interactions have to be taken into account at the starting point of
229: a physical analysis.
230:
231: * This paper has been submitted to Phys.Rev.Lett. on 27 January 1997 but has
232: been rejected despite of our argument that our electronic structure is
233: basically different from that published in Phys.Rev.Lett., Refs 1-3. A
234: pretty long and intensive discussion has led to a scientific bet for 1
235: million dollars between the authors and the Editors. The reader is asked to
236: see our later papers.
237:
238: References
239:
240: \begin{references}
241: \bibitem{1} M. C. Martin, G. Shirane, Y. Endoh, K. Hirota, Y. Moritomo, and
242: Y. Tokura, Phys.Rev. B {\bf 53,} 14285 (1996).
243:
244: \bibitem{2} D. D. Sarma, N. Shanthi, and P. Mahadevan, Phys.Rev. B {\bf 54},
245: 1622 (1996).
246:
247: \bibitem{3} J. M. D. Coey, M. Viret, and L. Ranno, Phys.Rev.Lett. {\bf 75},
248: 3910 (1995).
249:
250: \bibitem{4} E. O. Wollan and W. C. Koehler, Phys.Rev. {\bf 100}, 545 (1955).
251:
252: \bibitem{5} A. Abragam and B. Bleaney, {\it Electron Paramagnetic Resonance
253: of Transition Ions}, (Clarendon, Oxford) 1970.
254:
255: \bibitem{6} the computer programme is available on the written request.
256:
257: \bibitem{7} one of our papers has been rejected by the Editor of Phys. Rev.
258: B at beginning of 1997 on an argument that our computer program yielding the
259: splitting of the $E_{g}$ subterm must be wrong. Our long explanations did
260: not help and we were forced to propose, as the real responsibility of
261: Phys.Rev.Lett., a scientific bet for 1 million dollars, if similar discrete
262: structures appear in Phys.Rev.Lett..
263:
264: \bibitem{8} R. J.\ Radwanski and Z. Ropka - to be published (Physica B {\bf %
265: 186-188}, 507 (2000)).
266: \end{references}
267:
268: Figure Captions
269:
270: Fig. 1. Influence of spin-orbit interactions on the localized CEF states of
271: the 3$d^{4}$ system, that is expected to be realized in the Mn$^{3+}$, Cr$%
272: ^{2+}$ and Ru$^{4+}$ (4$d^{4}$) ions, placed in the octahedral ligand field,
273: a) The $^{5}D$ term has 25-fold degeneracy; b) The energy level scheme of
274: the 3$d^{4}$ system under the action of CEF interactions of the cubic
275: symmetry, produced by octahedrally arranged O$^{2-}$ ligands, with B$_{4}$%
276: =-10 K, with the orbital-doublet $E_{g}$ ground state. All the levels would
277: have the internal 5-fold spin-degree of freedom that could be revealed in
278: the paramagnetic state by applying external fields; c) the effect of the
279: spin-orbit interactions with $\lambda $= +120 K on the localized states of
280: the 3$d^{4}$ system; the states are labelled with the degeneracy - a singlet
281: ground state should be noticed; d) the same as c but for smaller $|$B$_{4}|$%
282: , i.e. B$_{4}$=-2 K; e) the multiplet structure obtained for the extremely
283: large spin-orbit coupling realized here by the absence of CEF interactions; $%
284: i$ denotes the ratio of the overall spin-orbit splitting (=10$\lambda $) and
285: of the overall CEF splitting (=120 $|$B$_{4}|$). In real LaMnO$_{3}$, $i$
286: amounts to $\simeq $0.05, i.e. CEF\ interactions largely prevail the
287: spin-orbit coupling.
288:
289: Fig. 2. The same as Fig. 1 but for the tetrahedral cubic CEF interactions,
290: i.e. B$_{4}$%
291: %TCIMACRO{\TEXTsymbol{>}}%
292: %BeginExpansion
293: \mbox{$>$}%
294: %EndExpansion
295: 0, that lead to the orbital-triplet $T_{2g}$ ground state.
296:
297: Fig. 3. The magnetic-field induced magnetic moment of the 3$d^{4}$ system at
298: 0 K with the $E_{g}$ (B$_{4}$%
299: %TCIMACRO{\TEXTsymbol{<}}%
300: %BeginExpansion
301: \mbox{$<$}%
302: %EndExpansion
303: 0, curves 1 and 2) and the $T_{2g}$ (B$_{4}$%
304: %TCIMACRO{\TEXTsymbol{>}}%
305: %BeginExpansion
306: \mbox{$>$}%
307: %EndExpansion
308: 0, curves 3 and 4) ground state calculated for $\lambda $=+120 K. The
309: dashed-point lines show results for $|$B$_{4}|$ = 100 K (curves 2, 4) and
310: the dashed line for $|$B$_{4}|$= 200 K (curves 1, 3). The induced-moment
311: somehow saturates to 2, 3 and 4 $\mu _{B}$ even in the case of magnetic
312: fields so large as 200 T. In LaMnO$_{3}$ the molecular field at 0 K is
313: anticipated for 100 T.
314:
315: \end{document}
316: