1: \documentstyle[twocolumn,aps,epsfig,floats]{revtex}
2: %\documentstyle[preprint,aps,epsfig]{revtex}
3:
4: \draft
5:
6: \begin{document}
7: %==============================================================================
8: \twocolumn[\hsize\textwidth\columnwidth\hsize\csname @twocolumnfalse\endcsname
9: \title{Unquenched large orbital magnetic moment in NiO}
10: \author{ S. K. Kwon and B. I. Min }
11: \address{ Department of Physics,
12: Pohang University of Science and Technology,
13: Pohang 790-784, Korea }
14: \date{February 22, 2000}
15:
16: \maketitle
17:
18: \begin{abstract}
19: Magnetic properties of NiO are investigated by incorporating the
20: spin-orbit interaction in the LSDA + $U$ scheme.
21: It is found that the large part of orbital moment remains unquenched in NiO.
22: The orbital moment contributes about $\mu_L = 0.29\mu_B$
23: to the total magnetic moment of $M = 1.93 \mu_B$,
24: as leads to the orbital-to-spin angular momentum ratio of $L/S = 0.36$.
25: The theoretical values are in good agreement
26: with recent magnetic X-ray scattering measurements.
27: \end{abstract}
28:
29: \pacs{PACS number: 71.20.Be, 71.70.Ej, 75.10.Lp, 75.50.Ee}
30: %{transition metal & alloy, spin-orbit coupling, band and itinerant model,
31: % Antiferromagnetics}
32: ]
33:
34: %\narrowtext
35: %==============================================================================
36: %\section{INTRODUCTION}
37: Electronic and magnetic structures of late $3d$ transition metal (TM)
38: mono-oxides (MnO, FeO, CoO, and NiO) have been extensively investigated
39: over the last decades \cite{Mattheiss}.
40: Conventional band theory with the local spin density approximation
41: (LSDA) fails to describe the electronic structures
42: of the compounds. In the LSDA calculations,
43: the energy gaps of MnO and NiO are underestimated\cite{Terakura}.
44: Even worse, FeO and CoO are predicted to be metallic
45: which are, on the contrary, large gap insulators in nature.
46: The main problem in the LSDA is the use of mean field-type
47: exchange-correlation functional
48: which is improper to describe localized $3d$ electrons.
49:
50: There have been several theoretical efforts
51: to cure the deficiencies in the LSDA, for example,
52: the self-interaction correction (SIC) scheme \cite{Cowan,Zunger,Svane},
53: the GW approximation (GWA) \cite{Hedin,Aryasetiawan},
54: and the LSDA + $U$ method \cite{Anisimov,Liechtenstein}.
55: The SIC-LSDA is in the line of the extended LSDA
56: by removing unphysical electron interaction with itself.
57: In the GWA, the quasi-particle energy is obtained
58: through the self-energy calculation to the lowest order in
59: the screened Coulomb interaction $W$.
60: The GWA method applied to NiO gives a rather good description
61: of the energy gap size \cite{Aryasetiawan}.
62: Computational load, however, is very heavy in the GWA.
63: The LSDA + $U$ method overcomes
64: the failure of the LSDA by incorporating the on-site Coulomb correlation $U$
65: of the multiband Hubbard model-like.
66: In this method, localized $3d$ electrons are treated
67: separately from delocalized $sp$ electrons.
68: As a result, all the $3d$ TM mono-oxides in the LSDA + $U$ are
69: obtained as insulators with well developed energy gaps which are comparable
70: to experimental values \cite{Anisimov}.
71: In this way, the energy gap problem in $3d$ TM mono-oxides
72: is considered to be solved by various calculational methods.
73:
74: Due to the outermost characteristics of TM $3d$ electrons,
75: atomic $3d$ orbitals are greatly deformed in solids
76: by the crystal field and/or band hybridization effects.
77: Hence, the orbital moment of $3d$ TM ion is
78: usually quenched in solids, because it originates from atomic nature of
79: involved atomic elements\cite{Kittel}.
80: For example, all the $3d$ ferromagnetic transition metals
81: of Fe, Co, and Ni show negligible
82: orbital magnetic moments in the range of $\mu_L \lesssim 0.1 \mu_B$ \cite{Min}.
83: It is, however, expected that the strong Coulomb correlation
84: in $3d$ TM mono-oxides preserves
85: the orbital moment of localized $3d$ electrons
86: by reducing the ligand crystal field effects at metal ion sites.
87:
88: In fact, for CoO, it is easily conceived
89: that the orbital moment is only partially quenched
90: because the measured magnetic moment $M = 3.4 \mu_B$ \cite{Khan}
91: simply exceeds the spin magnetic moment alone and
92: the minority $t_{2g}$ band is occupied only two-thirds of its available states.
93: Therefore, the existence of orbital moment in CoO has been stressed
94: many times \cite{Terakura,Svane,Shishidou}.
95: It is also shown that the LSDA + $U$ method gives
96: a good description of magnetic structure of CoO
97: with a large orbital moment of $\mu_L \sim 1 \mu_B$ \cite{Solovyev}.
98: On the other hand, in the case of NiO,
99: measured magnetic moments are in the range
100: of $M = 1.77 \sim 2.2\pm0.2 \mu_B$ \cite{Fender,Cheetham,Fernandez},
101: which are comparable to the spin only moment of isolated Ni$^{2+}$ ion.
102: Therefore, the magnetic moment in NiO has been fully attributed
103: to the spin moment and the orbital moment is
104: expected to be completely quenched.
105:
106: However, recent magnetic X-ray scattering measurement \cite{Fernandez}
107: indicates that the orbital-to-spin angular momentum ratio in NiO is
108: as large as $L/S = 0.34$, far from fully quenched orbital moment.
109: In the nonresonant magnetic X-scattering,
110: the separation of spin and orbital moment is possible,
111: because the spin and orbital moment densities have different geometrical
112: prefactors in the scattering cross section that can be adjusted
113: by changing either the scattering geometry or the X-ray polarization.
114: In Ref. \cite{Fernandez}, the orbital moment in NiO is extracted
115: by the polarization analysis of nonresonant magnetic-scattering intensities.
116: This method has evidenced a large contribution of the orbital moment
117: to the total magnetic moment.
118: They also found that the spin and orbital moments in NiO are collinear.
119:
120: Magnetic moment is one of the basic ground state quantities
121: which should be provided by an appropriate band method.
122: If the orbital moment is so large in NiO,
123: all the previous attempts \cite{Anisimov,Towler,Shick}
124: which tried to directly compare calculated spin moments
125: with experimental moments are under a mistake.
126: The linearized muffin-tin orbital (LMTO) calculation for NiO
127: with an orbital-polarization (OP) correction (LSDA + OP)
128: in a crystal-field basis yields
129: the spin and orbital magnetic moments of $\mu_S = 1.43 \mu_B$
130: and $\mu_L = 0.12 \mu_B$, respectively \cite{Norman1}.
131: Although the total magnetic moment is significantly improved
132: to $M = 1.55 \mu_B$ with the OP correction,
133: as compared to $M = 1.23 \mu_B$ from the LSDA,
134: it is still smaller than the experimental measurements
135: and the ratio of $L/S = 0.17$ is only half of
136: the magnetic X-ray scattering value \cite{Fernandez}.
137: The SIC-LSDA method \cite{Svane} yields
138: a large orbital moment of $\mu_L = 0.27\mu_B$,
139: which, however, was suspected as a methodological artifact
140: by other authors \cite{Anisimov,Towler}.
141: In more recent study, the orbital moment is simply neglected \cite{Shick}.
142:
143: %------------------------------------------------------------------------------
144: \begin{figure}[t]
145: \epsfig{file=NiO_fig1.ps,width=8.65cm}
146: \caption{\label{fig1}
147: Magnetic moment behavior with varying the
148: lattice constant in the LSDA. Inset is the behavior
149: of the orbital-to-spin angular momentum ratio $L/S$.
150: Increase of all the magnetic moments is understood
151: by the reduction of the crystal field and the band hybridization strengths
152: with volume expansion.
153: }
154: \end{figure}
155: %------------------------------------------------------------------------------
156: %------------------------------------------------------------------------------
157: \begin{figure}[t]
158: \epsfig{file=NiO_fig2.ps,width=8.65cm}
159: \caption{\label{fig2}
160: Coulomb correlation effects on the magnetic moments obtained
161: by the LSDA + $U$ method.
162: We have used the exchange parameter of $J = 0.89$ eV.
163: Inset is the orbital-to-spin angular momentum ratio $L/S$.
164: The total magnetic moment $M = 1.93 \mu_B$ and the ratio $L/S = 0.36$
165: for $U = 8.0$ eV are in good agreement with experimental values.
166: }
167: \end{figure}
168: %------------------------------------------------------------------------------
169: %==============================================================================
170: %\section{CALCULATIONAL METHOD}
171: To determine the size of orbital moment in NiO,
172: we have performed the LSDA + $U$ calculations
173: using the LMTO band method within the atomic sphere approximation (ASA).
174: The incorporation of the spin-orbit coupling into the LSDA + $U$
175: method is known to give right orbital polarization
176: in strongly correlated electron systems \cite{Solovyev}.
177: The LSDA + $U$ Hamiltonian is given by
178: \begin{equation}
179: {\cal H}_{{\mathrm LSDA} + U} = {\cal H}_{\mathrm LSDA}
180: - {\cal H}_{\mathrm dc} + {\cal H}_{U}
181: \end{equation}
182: where the first term in the right hand side is the LSDA Hamiltonian
183: and the second term is the double counting correction
184: for the third term ${\cal H}_{U}$.
185: With the Coulomb interaction $U$ and exchange interaction $J$ parameters,
186: one can write ${\cal H}_{\mathrm dc}$ and ${\cal H}_{U}$, respectively, as
187: \begin{eqnarray}
188: {\cal H}_{\mathrm dc} &=& \frac{1}{2}UN(N-1)
189: - \frac{1}{2}J\sum_{\sigma} N^{\sigma}(N^{\sigma}-1), \\
190: {\cal H}_{U} &=& \frac{1}{2}\sum_{\{m\},\sigma}V(mm';m''m''')
191: n_{mm''}^{\sigma}n_{m'm'''}^{-\sigma} \nonumber \\
192: &+& \frac{1}{2}\sum_{\{m\},\sigma}
193: \left [ V(mm';m''m''')-V(mm';m'''m'') \right ] \nonumber \\
194: & &~~~~~~~~~\times n_{mm''}^{\sigma} n_{m'm'''}^{\sigma}
195: \end{eqnarray}
196: where $n_{mm'}^{\sigma}$ is the $d$ occupation number matrix
197: of spin $\sigma$ and $N^{\sigma} = {\mathrm Tr}(n_{mm'}^{\sigma})$,
198: $N = N^+ + N^-$.
199: We relate the screened Coulomb interaction $V(mm';m''m''')$
200: with the Slater integral $F^k$;
201: \begin{eqnarray}
202: V(mm';m''m''') = \sum_{k=0}^{2l}c^{k}(lm,lm'')c^{k}(lm',lm''')F^{k},
203: \end{eqnarray}
204: where $c^k(lm,lm')$ is a Gaunt coefficient.
205: For $3d$ electrons, three Slater integrals of $F^0, F^2$, and $F^4$
206: are involved in the calculation.
207: Among those, the ratio of $F^4/F^2$ is known to be constant
208: $\sim 0.625$ for most $3d$ TM atoms \cite{Sawatzky}.
209: Hence, the actual number of parameters is
210: reduced from three $F^k$'s to two of $U$ and $J$,
211: which are given by $U = F^0$ and $J = (F^2 + F^4)/14$, respectively.
212: To determine the orbital moment, the spin-orbit coupling is simultaneously
213: included in the self-consistent variational loop \cite{Min}.
214: We have assumed the antiferromagnetic ordering state of type-II,
215: and the experimental lattice constant of $a=7.893$ a.u. is used.
216:
217: %==============================================================================
218: %------------------------------------------------------------------------------
219: \begin{table*}[t]
220: \caption{\label{table1}
221: Calculated spin ($\mu_S$), orbital ($\mu_L$), and total ($M$) magnetic
222: moments in $\mu_B$ at the experimental lattice constant of $a=7.893$ a.u..
223: $L/S$ is the orbital-to-spin angular momentum ratio. }
224: \begin{tabular}{cccccc}
225: Method &LSDA$^a$&LSDA + OP$^b$&SIC-LSDA$^c$&LSDA + $U^d$&Experiment\\\hline
226: $\mu_S$ & 1.12 & 1.43 & 1.53 & 1.64 & 1.90$\pm 0.2^e~$ \\
227: $\mu_L$ & 0.15 & 0.12 & 0.27 & 0.29 & 0.32$\pm 0.05^e$ \\
228: $M$ & 1.27 & 1.55 & 1.80 & 1.93 & 1.77$^f$, 1.90$^g$,
229: 2.2$\pm$0.2$^e$ \\
230: $L/S$ & 0.27 & 0.17 & 0.35 & 0.36 & 0.34$^e$
231: \end{tabular}
232: $^a$ Present results. \\
233: $^b$ Reference \cite{Norman1}. \\
234: $^c$ Reference \cite{Svane}. \\
235: $^d$ Present results with $U = 8.0$ eV and $J = 0.89$ eV. \\
236: $^e$ Reference \cite{Fernandez}. \\
237: $^f$ Reference \cite{Fender}. \\
238: $^g$ Reference \cite{Cheetham}.
239: \end{table*}
240: %------------------------------------------------------------------------------
241: %\section{RESULTS AND DISCUSSION}
242: In Fig.~\ref{fig1}, magnetic moment behavior is shown
243: with varying the lattice constant in the LSDA.
244: Both the spin and orbital magnetic moments increase monotonically
245: with increasing the lattice constant.
246: This feature is understandable in view of that the crystal field strength
247: at Ni sites and the hybridization between Ni $3d$ and O $2p$ bands
248: are reduced with volume expansion.
249: At the experimental lattice constant of $a=7.893$ a.u.,
250: the spin and orbital magnetic moments are obtained
251: as $\mu_S = 1.12 \mu_B$ and $\mu_L = 0.15 \mu_B$, respectively,
252: which are consistent with existing results \cite{Norman1}.
253: Both the spin and orbital magnetic moments are larger in NiO than in fcc Ni
254: which has $\mu_S = 0.63 \mu_B$ and $\mu_L = 0.06 \mu_B$.
255: This suggests that $3d$ electrons are more localized in NiO
256: than in Ni metal.
257: Although the ratio of $L/S = 0.27$ in the LSDA is only slightly smaller
258: than $L/S = 0.34$ in the experiment (see Table.~\ref{table1}),
259: the spin and orbital polarizations in the LSDA
260: are not large enough because of underestimation
261: of the Coulomb correlation between $3d$ electrons.
262:
263: Figure~\ref{fig2} shows the Coulomb correlation effects
264: on the magnetic moments obtained by the LSDA + $U$ method.
265: We have used the exchange parameter of $J = 0.89$ eV
266: which is comparable to literature value \cite{Anisimov}.
267: Once the strong Coulomb interaction is introduced between Ni $3d$ electrons,
268: the magnetic moments increase substantially.
269: The role of the Coulomb interaction is significant
270: to localize Ni $3d$ electrons.
271: With $U = 8.0$ eV found in Ref. \cite{Anisimov} and Ref. \cite{Norman2},
272: we have obtained the spin and orbital magnetic
273: moments of $\mu_S = 1.64 \mu_B$ and $\mu_L = 0.29 \mu_B$, respectively.
274: The total magnetic moment $M = 1.93 \mu_B$
275: and the ratio $L/S = 0.36$ for $U = 8.0$ eV are
276: in good agreement with the experimental data.
277: The change of the Coulomb parameter $U$ by $1.0$ eV
278: results in increments of both the magnetic moments
279: and the ratio $L/S$ by $\Delta\mu_S \sim \Delta\mu_L \sim 0.01 \mu_B$
280: and $\Delta (L/S) \sim 0.01$, respectively.
281: The overall total magnetic moment difference
282: and the ratio $L/S$ difference are
283: only $\Delta M = 0.12 \mu_B$ and $\Delta (L/S) = 0.05$
284: inbetween $U = 6.0$ and $10$ eV.
285: Thus, all the magnetic moments and the ratio $L/S$ are rather insensitive
286: to the Coulomb interaction $U$ within the employed parameter range.
287:
288: Calculated magnetic moments are summarized in Table.~\ref{table1}
289: in comparison with experimental data and previous results
290: by other calculational methods.
291: As mentioned above,
292: the sizes of magnetic moments are underestimated in the LSDA.
293: The total magnetic moment of $M = 1.27 \mu_B$ in the LSDA is
294: much smaller than $M = 1.77 \sim 2.2\pm0.2 \mu_B$
295: in experiments \cite{Fender,Cheetham,Fernandez}.
296: In the LSDA + OP, the spin moment is significantly
297: improved to $\mu_S = 1.43 \mu_B$ \cite{Norman1},
298: which, however, is still smaller than the experimental values.
299: More serious is the decrease of orbital moment in the LSDA + OP,
300: which unfavorably makes the consistency between the theoretical
301: and experimental $L/S$ ratio become worse.
302: Both the SIC-LSDA ($M = 1.80 \mu_B$) \cite{Svane}
303: and the LSDA + $U$ ($M = 1.93 \mu_B$)
304: give values in agreement with experimental data.
305: But it can be concluded that the LSDA + $U$ results with
306: $U = 8.0$ eV and $J = 0.89$ eV,
307: among various calculations, are especially
308: in best agreement with experimental measurements.
309: The ratio of $L/S = 0.36$ in the LSDA + $U$ is also consistent with
310: $L/S = 0.34$ in the magnetic X-ray scattering measurement \cite{Fernandez}.
311:
312: %==============================================================================
313: %\section{CONCLUSION}
314: In conclusion, we have found using the LSDA + $U$ calculations
315: that the orbital moment in NiO is not fully quenched, which is
316: as large as $\mu_L = 0.29 \mu_B$.
317: Both the total magnetic moment of $M = 1.93 \mu_B$
318: and the orbital-to-spin angular momentum
319: ratio of $L/S = 0.36$ for $U = 8.0$ eV are close to experimental values.
320: The Coulomb correlation and the spin-orbit coupling are
321: crucial to get right magnetic polarization in NiO.
322:
323: %==============================================================================
324: %\section{ACKNOWLEDGEMENTS}
325: Acknowledgements$-$
326: The authors would like to thank K.B. Lee for helpful discussions.
327: This work was supported by the KOSEF (1999-2-114-002-5)
328: and in part by the Korean MOST-FORT fund.
329:
330: %==============================================================================
331: \begin{thebibliography}{99}
332: \bibitem{Mattheiss} L. F. Mattheiss, Phys. Rev. B {\bf 5}, 290 (1972).
333: \bibitem{Terakura} K. Terakura, A. R. Williams, T. Oguchi,
334: and J. K$\ddot{\mathrm u}$bler,
335: Phys. Rev. Lett. {\bf 52}, 1830 (1984);
336: K. Terakura, T. Oguchi, A. R. Williams,
337: and J. K$\ddot{\mathrm u}$bler,
338: Phys. Rev. B {\bf 30}, 4734 (1984).
339: \bibitem{Cowan} R. D. Cowan, Phys. Rev. {\bf 164}, 54 (1967);
340: I. Lindgren, Int. J. Quantum Chem. {\bf 5}, 411 (1971).
341: \bibitem{Zunger} A. Zunger, J. P. Perdew, and G. L. Oliver,
342: Solid State Commun. {\bf 34}, 933 (1980);
343: J. P. Perdew and A. Zunger,
344: Phys. Rev. B {\bf 23}, 5048 (1981).
345: \bibitem{Svane} A. Svane and O. Gunnarsson,
346: Phys. Rev. Lett. {\bf 65}, 1148 (1990).
347: \bibitem{Hedin} L. Hedin and S. Lundqvist, in {\it Solid State Physics},
348: edited by H. Ehrenreich, F. Seitz, and D. Turnbell
349: (Academic, New York, 1969), Vol 23, p. 1.
350: \bibitem{Aryasetiawan} F. Aryasetiawan and O. Gunnarsson,
351: Phys. Rev. Lett. {\bf 74}, 3221 (1995).
352: \bibitem{Anisimov} V. I. Anisimov, J. Zaanen, and O. K. Andersen,
353: Phys. Rev. B {\bf 44}, 943 (1991).
354: \bibitem{Liechtenstein} A. I. Liechtenstein, V. I. Anisimov, and J. Zaanen,
355: Phys. Rev. B {\bf 52}, R5467 (1995).
356: For review, see V. I. Anisimov, F. Aryasetiawan,
357: and A. I. Liechtenstein,
358: J. Phys. : Condens. Matter {\bf 9}, 767 (1997).
359: \bibitem{Kittel} C. Kittel, {\it Introduction to Solid State Physics} 7th ed.,
360: (John Wiley \& Sons, New York, 1996), Chapter 14.
361: \bibitem{Min} B. I. Min and Y.-R. Jang,
362: J. Phys.: Condens. Matter {\bf 3}, 5131 (1991).
363: \bibitem{Khan} D. C. Khan and R. A. Ericksson,
364: Phys. Rev. B {\bf 1}, 2243 (1970).
365: \bibitem{Shishidou} T. Shishidou and T. Jo, J. Phys. Soc. Jpn.
366: {\bf 67}, 2637 (1998).
367: \bibitem{Solovyev} I. V. Solovyev, A. I. Liechtenstein, and K. Terakura,
368: Phys. Rev. Lett. {\bf 80}, 5758 (1998).
369: \bibitem{Fender} B. E. F. Fender, A. J. Jacobson, and F. A. Wegwood,
370: J. Chem. Phys. {\bf 48}, 990 (1968).
371: \bibitem{Cheetham} A. K. Cheetham and D. A. O. Hope,
372: Phys. Rev. B {\bf 27}, 6964 (1983).
373: \bibitem{Fernandez} V. Fernandez, C. Vettier, F. de Bergevin, C. Giles,
374: and W. Neubeck, Phys. Rev. B {\bf 57}, 7870 (1998);
375: W. Neubeck, C. Vettier, V. Fernandez, F. de Bergevin,
376: and C Giles, J. Appl. Phys. {\bf 85}, 4847 (1999).
377: \bibitem{Towler} M. D. Towler, N. L. Allan, N. M. Harrison, V. R. Saunders,
378: W. C. Mackrodt, and E. Apr$\grave{\mathrm a}$,
379: Phys. Rev. B {\bf 50}, 5041 (1994).
380: \bibitem{Shick} A. B. Shick, A. I. Liechtenstein, and W. E. Pickett,
381: Phys. Rev. B {\bf 60}, 10 763 (1999).
382: \bibitem{Norman1} M. R. Norman,
383: Phys. Rev. B {\bf 44}, 1364 (1991).
384: \bibitem{Sawatzky} F. M .F. de Groot, J. C. Fuggle, B. T. Thole, and
385: G. A. Sawatzky, Phys. Rev. B {\bf 42}, 5459 (1990).
386: \bibitem{Norman2} M. R. Norman and A. J. Freeman,
387: Phys. Rev. B {\bf 33}, 8896 (1986).
388: \end{thebibliography}
389: %==============================================================================
390: \end{document}
391: