1: %\documentstyle[aps,prb,floats,epsfig]{revtex}
2: \documentstyle[aps,floats,epsfig]{revtex}
3: %\documentstyle[aps,preprint,epsfig]{revtex}
4: \begin{document}
5: \draft
6: \twocolumn[\hsize\textwidth\columnwidth\hsize\csname @twocolumnfalse\endcsname
7: \title{Band-theoretical prediction of magnetic anisotropy in uranium
8: monochalcogenides}
9: \author{Tatsuya Shishidou}
10: \address{Japan Science and Technology Corporation, Tokyo 102-0081, Japan}
11: \address{and Department of Quantum Matter, ADSM, Hiroshima University,
12: Higashihiroshima 739-8526, Japan}
13: \author{Tamio Oguchi}
14: \address{Department of Quantum Matter, ADSM, Hiroshima University,
15: Higashihiroshima 739-8526, Japan}
16: \date{\today}
17: \maketitle
18: \begin{abstract}
19: Magnetic anisotropy of uranium monochalcogenides, US, USe and UTe, is
20: studied by means of
21: fully-relativistic spin-polarized band structure calculations within the
22: local spin-density
23: approximation.
24: It is found that the size of the magnetic anisotropy is fairly large
25: ($\simeq$10 meV/unit formula), which is comparable with experiment.
26: This strong anisotropy is discussed in view of a pseudo-gap formation,
27: of which crucial ingredients are the exchange splitting of U $5f$ states and their hybridization
28: with chalcogen $p$ states ($f$--$p$ hybridization).
29: An anomalous trend in the anisotropy is found in the series (US$\gg$USe$<$UTe)
30: and interpreted in terms of competition between localization of the U $5f$
31: states and the $f$--$p$ hybridization.
32: It is the spin-orbit interaction on the chalcogen $p$ states that plays an
33: essential role
34: in enlarging the strength of the $f$--$p$ hybridization in UTe,
35: leading to an anomalous systematic trend in the magnetic anisotropy.
36: \end{abstract}
37: % insert suggested PACS numbers in braces on next line
38: \pacs{71.20.Gj, 75.30.Gw, 75.50.Cc}
39: ]
40:
41: Magnetic moments in solid originate in the spin and orbital components
42: of electrons.
43: Since electronic states responsible for the magnetism are normally
44: localized in a particular atomic region,
45: the moments can be regarded as site-selective quantities.
46: Experimental techniques such as x-ray magnetic circular dichroism (XMCD)
47: combined with
48: the so-called spin and orbital sum rules \cite{Thole,Carra} provide such
49: separable information of the spin and orbital magnetic
50: moments of ferro- and ferri-magnets.
51: X-ray magnetic scattering \cite{Blume} can also
52: give us similar information.
53:
54: Usually $5f$ electrons play major roles in magnetism of uranium compounds.
55: Since the spin-orbit interaction (SOI) of the $5f$ electrons is relatively
56: large,
57: the size of the $5f$ orbital moment is often expected to be greater than
58: the spin counterpart.
59: Unlike the $4f$ orbitals in rare-earth element systems,
60: the $5f$ states are more or less extended and may be possibly
61: affected by their environmental effects, hybridization and crystal field.
62: Thus, the magnetic moment in $5f$ systems must be strongly
63: material-dependent. Magnetic anisotropy is another fundamental quantity
64: in magnetism, which is often even more important for applications.
65: However, the magnetic anisotropy energy is usually very small to be
66: evaluated from first principles
67: and furthermore its microscopic origins have never been clearly
68: understood yet \cite{Jansen}.
69: It is, therefore, quite interesting to study such issues on the magnetism
70: by using a state-of-the-art band-theoretical technique.
71:
72: Uranium monochalcogenides, US, USe and UTe, have NaCl-type cubic crystal
73: structure
74: and show a ferromagnetic order at the Curie temperatures, 177K, 160K and
75: 104K, respectively\cite{Handbook Actinide}.
76: It is well known that the monochalcogenides show strong magnetic
77: anisotropy \cite{Lander},
78: where the [111] ([001]) direction is the easy (hard) axis.
79: Interestingly, the saturation magnetic moment depends on the magnetization
80: axis \cite{Tillwick}.
81: The largest moment along the easy axis, the smallest along the hard axis.
82: The total magnetic moment per uranium atom increases from sulfide through
83: telluride with increasing lattice constant.
84: As long as the sulfide is concerned, the $5f$ electrons are considered to
85: be itinerant from photoemission \cite{Reihl} and other
86: experiments \cite{Schoenes,Huang,Rudigier}.
87:
88: In the present study, mechanism of the magnetic anisotropy in the uranium
89: monochalcogenides is investigated by
90: first-principles calculations and an anomalous trend in the size of the
91: anisotropy in the series is predicted.
92: Our method is based on the local spin-density approximation (LSDA) to the
93: density functional theory.
94: One-electron Kohn-Sham equations are solved self-consistently by using
95: an iterative scheme of the
96: full-potential
97: linear augmented plane wave method \cite{Soler} in a scalar-relativistic
98: fashion.
99: We include SOI as the second variation every self-consistent-field step.
100: The improved tetrahedron method proposed by Bl\"{o}chl \cite{Blochl} is used
101: for the Brillouin-zone integration.
102: More details about our methods and calculated results are
103: published elsewhere \cite{Shishidou2}.
104: % fig_band %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
105: \begin{figure}[htb]
106: \centerline{\epsfig{file=band.epsi,width=8cm}}
107: %\centerline{\epsfxsize=8.0cm\epsfbox{band.epsi}}
108: \vspace{3mm}
109: \caption{Fully-relativistic spin-polarized band structure of ferromagnetic
110: US with the [111]
111: magnetization. The Fermi energy is chosen as the origin.}
112: \label{fig_band}
113: \end{figure}
114: % fig_band %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
115:
116: Figure \ref{fig_band} shows calculated fully-relativistic spin-polarized
117: band structure of ferromagnetic US with the [111] magnetization.
118: Shallow core U $6p$ states form $j=1/2$ and $3/2$ bands split by large SOI.
119: S $3s$ bands are located just above the U $6p_{3/2}$ bands and have
120: relatively large dispersion.
121: S $3p$ bands are situated from 7 to 3 eV below the Fermi energy
122: ($\varepsilon_{\rm F}$).
123: Dispersive bands appearing just below $\varepsilon_{\rm F}$ are made mostly
124: of U $d$ states.
125: Relatively narrow U $5f$ bands are pinned around $\varepsilon_{\rm F}$.
126: It is found that the largest contribution to the magnetic moments comes
127: from U $5f$ states as expected.
128: Calculated $5f$ moments of US are listed in Table \ref{tab1} for
129: three magnetization axes, [001], [110] and [111].
130: %TABLE1%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
131: \begin{table}[htdp]
132: \caption{Calculated $5f$ spin and orbital magnetic moments
133: in Bohr magneton of US for different magnetization directions ($\hat{M}$).
134: $\Delta E$ is a difference in the total energy in meV from the case of the
135: [111] magnetization.}
136: \begin{tabular}{ccccc}
137: $\hat{M}$ & spin & orbital & total & $\Delta E$ \\
138: \hline \\
139: $[001]$\tablenotemark[1] &1.61 & -2.13 & -0.52 & 14 \\
140: $[110]$\tablenotemark[1] &1.60 & -2.24 & -0.64 & 5 \\
141: $[111]$\tablenotemark[1] &1.60 & -2.33 & -0.73 & 0 \\
142: %LAPW & [111] &1.7 & -2.6 & -0.9 & -1.5 & \\
143: %ASW & [111] & 1.5 & -2.6 & -1.1 & -1.7 & \\
144: %LMTO & [111] & 2.1 & -3.2 & -1.1 & -1.5 & \\
145: %
146: $[111]$\tablenotemark[2] & 1.3\tablenotemark[3] & -3.0\tablenotemark[3] &
147: -1.7 & \\
148: \end{tabular}
149: \tablenotetext[1]{Present work.}
150: \tablenotetext[2]{Neutron-scattering experiment (Ref.\ \onlinecite{Wedgwood}).}
151: \tablenotetext[3]{From analysis in Ref.\ \onlinecite{Severin}.}
152: \label{tab1}
153: \end{table}%
154: %TABLE1%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
155: % fig_jzdos %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
156: \begin{figure*}[bht]
157: \centerline{\epsfig{file=jzdos.eps,width=16cm}}
158: \vspace{3mm}
159: \caption{Calculated $j_z$-projected $5f$ density of states ($j=5/2$) of
160: ferromagnetic US
161: with the (a) [111] and (b) [001] magnetization directions.
162: Quantization axis for specifying $j_z$ is taken as the same direction of
163: magnetization.}
164: \label{fig_jzdos}
165: \end{figure*}
166: % fig_jzdos %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
167: The spin moments hardly change with the magnetization axis
168: while the orbital moments show large dependence of the axis.
169: Consequently, the total moments depend strongly on the magnetization axis,
170: being in qualitatively good agreement with experiment \cite{Tillwick}.
171: %The results for the easy axis is consistent with the previously calculated ones.
172: %The discrepancy may be due to the difference in the sphere radius assumed.
173: Comparing with the experimental data, the spin moment is slightly overestimated
174: while the orbital moment is underestimated.
175: Origins of the discrepancy in the magnetic moments have been already
176: discussed in the previous Hartree-Fock type study \cite{Shishidou}.
177: The total-energy difference shows that the [111] direction is the lowest
178: while [001] is the highest.
179: The difference of the total energy between the easy and hard axes is nothing but
180: the magnetic anisotropy energy.
181: The size of the magnetic anisotropy is comparable with experimental
182: data \cite{Lander} and is about $10^3$ times larger than that of
183: $3d$ transition metals.
184: Let us discuss mechanism of the magnetic anisotropy and its relation to the
185: magnetic moments.
186:
187: Because of large SOI of $5f$ electrons, $j=5/2$ and $7/2$ states are
188: roughly well separated
189: and the occupied states are composed mostly of the $j=5/2$ states.
190: The $j_z$-projected density of states (DOS) in the $j=5/2$ states is plotted in
191: Fig.~\ref{fig_jzdos}.
192: One can easily note that a clear pseudo-gap is formed at $\varepsilon_{\rm F}$
193: in the [111] magnetization,
194: while it becomes less apparent in the [001] direction.
195: Relative stability of the [111] magnetization must come from the existence
196: of the pseudo-gap.
197:
198: In the case of [111], the plus and minus components are well separated.
199: This is the exchange splitting of the bands and the pseudo-gap is
200: considered to be a sort of the exchange gap.
201: Only a small amount of mixing can be seen in the $j_{z} = \pm 3/2$ states.
202: On the other hand, the pseudo-gap is almost diminished in the case of [001]
203: because of larger mixing in $j_{z}= \pm 3/2$ and $j_{z}= \pm 1/2$.
204: By counting the number of occupied electrons in each partial DOS,
205: one can get the occupation of each $j_z$ state.
206: For [111] $\pm j_z$ state are well exchange-polarized
207: but for [001] the occupations in the $j_{z}= \pm 1/2$ bands are almost
208: equal due to the mixing.
209:
210: In order to get more intuitive insight into the occupations,
211: a change-density plot of the $5f$ electrons around the U atom is
212: depicted in Fig.~\ref{fig_charge}.
213: % fig_charge %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
214: \begin{figure*}[bht]
215: \centerline{\epsfig{file=charge.epsi,width=12cm}}
216: \vspace{3mm}
217: \caption{Angular dependence of calculated charge density of the $5f$ electrons
218: around the U atom with the (a) [111] and (b) [001] magnetization directions.
219: The spherical component has been reduced to $20\%$ of its value
220: to emphasize nonspherical parts.}
221: \label{fig_charge}
222: \end{figure*}
223: % fig_charge %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
224: In the [111] magnetization, the $5f$ electrons tend to point to
225: the direction of neighboring-U atoms;
226: hexagonal brim stretching out in the (111) U plane
227: and triangular bell around [111] direction are clearly formed.
228: As a result, the $5f$ electrons have less distribution in the neighboring-S
229: directions
230: and they can reduce the hybridization with the S $p$ states and gain the
231: exchange splitting.
232: The small mixing found in $j_{z}=\pm 3/2$ can be understood by the fact
233: that the
234: $j_{z}=\pm 3/2$ orbitals are extended to the S atoms
235: with respect to the polar angle \cite{quantization}.
236: In the [001] magnetization, on the other hand,
237: the $5f$ electrons again try to point to the U-U bonds,
238: twelve [110] directions.
239: In the $xy$ plane gentle depression is formed in the nearest-neighboring-S
240: directions,
241: because the $j_{z}=\pm 5/2$ states which spread in the $xy$ plane
242: can prevent from mixing with the S $p$ states by using their azimuthal
243: degrees of freedom.
244: However, no depression is seen in the [$00\pm 1$] directions.
245: The $j_{z}=\pm 1/2$ states extending to the $z$ direction
246: have no way to refuse the mixing with the S $p$ states
247: because of less azimuthal degrees of freedom in its $m=0$ component.
248: This hybridization of the $j_{z}=\pm 1/2$ states with the neighboring S $p$
249: orbits
250: destroys the pseudo-gap in the corresponding partial band and
251: makes the [001] magnetization unfavorable \cite{S_111}.
252: This is the most important mechanism of the magnetic anisotropy
253: found in the present study and leads to a very interesting variation
254: in the series of the uranium monochalcogenide as we shall discuss below.
255:
256: A variation in the calculated spin and orbital magnetic moments
257: for USe and UTe as well as for US is shown in Fig.~\ref{fig_moment}.
258: % fig_monent %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
259: \begin{figure}[htb]
260: \centerline{\epsfig{file=moment.eps,width=8cm}}
261: \caption{Calculated spin and orbital magnetic moments in $\mu_{\rm B}$ of US, USe and UTe
262: with the [111] magnetization.
263: Filled rectangles, triangles, and circles represent
264: the spin, orbital, and total $5f$ magnetic moments, respectively.
265: Open circles show experimental $5f$ moments \protect\cite{Wedgwood}.}
266: %(Ref.\ \onlinecite{Wedgwood}).}
267: \label{fig_moment}
268: \end{figure}
269: % fig_moment %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
270: The spin and orbital moments increase as one goes from sulfide through
271: telluride.
272: This increase of the moments can be interpreted by narrowing of the U $5f$
273: bands due to increase of the lattice constant.
274: Based on such theoretical observation on the $5f$-electron nature,
275: we can expect a simple trend in the magnetic anisotropy.
276: From sulfide to telluride, the lattice constant increases and the $5f$ band
277: width is reduced.
278: Accordingly the magnetic moments are enhanced, approaching to an atomic limit.
279: Therefore, the magnetic anisotropy may show a {\em monotonic decrease\/}
280: from sulfide through telluride with increasing free-atom nature.
281: But the reality is not so simple and calculated magnetic anisotropy energy,
282: plotted in Fig.~\ref{fig_aniene}, shows an anomalous behavior: {\em
283: decrease and increase\/} with
284: a minimum appearing at USe.
285: % fig_aniene %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
286: \begin{figure}[htb]
287: \centerline{\epsfig{file=aniene.eps,width=8cm}}
288: \caption{Calculated magnetic anisotropy energy of US, USe and UTe, taken from the
289: total-energy difference between the [111] and [001] magnetization directions.}
290: \label{fig_aniene}
291: \end{figure}
292: % fig_aniene %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
293: Calling that the strong magnetic anisotropy in the present compounds
294: originates in the hybridization between the U $5f$ and chalcogen $p$ states
295: ($f$--$p$ hybridization),
296: differences in the chalcogen $p$ states may be a clue to understand such an
297: anomalous feature.
298:
299: Hybridization depends upon spatial extension of the relevant orbitals via
300: transfer integrals.
301: Surprisingly, tails of the chalcogen $p$ orbitals are almost the same
302: if we consider the nearest neighbor distance between the U and chalcogen atoms.
303: The Te $5p_{3/2}$ orbital has the longest tail of the chalcogens
304: but the difference is not so large.
305: In addition to the orbital extension, the orbital energy is another factor
306: to determine the hybridization.
307: Generally, the orbital-energy difference between the U $5f$ and chalcogen
308: $p$ states becomes smaller
309: when going from sulfur through tellurium.
310: Because of the large SOI, the Te $5p_{3/2}$ state raises up and its
311: orbital-energy difference from the
312: U $5f$ decreases substantially.
313: Therefore, the $f$--$p$ hybridization
314: becomes very strong in the case of telluride.
315: Basically, from sulfide through telluride,
316: the U $5f$ electrons tend to be more localized due to enlargement of the
317: lattice constant.
318: But the $f$--$p$ hybridization, which makes the [001] magnetization unfavorable,
319: increases rapidly from selenide to telluride, resulting to large magnetic
320: anisotropy in UTe.
321: It should be noted that experimental data about the magnetic anisotropy
322: have not been available for USe and UTe \cite{Lander2}.
323: Experimental efforts to examine our prediction are strongly desired.
324:
325: In conclusion, we have carried out fully-relativistic LSDA calculations for
326: uranium monochalcogenides, US, USe and UTe.
327: The magnetic anisotropy can be well described by the LSDA calculations.
328: The pseudo-gap formation stabilizes the [111] magnetization.
329: We have emphasized important roles of the $j_{z}=\pm 1/2$ states,
330: which make the [001] magnetization unfavorable significantly.
331: Calculated magnetic moments show the monotonic increase in the series
332: with increasing lattice constant.
333: On the other hand, the magnetic anisotropy shows the anomalous systematic trend.
334: This can be understood by considering the hybridization of
335: the U $5f$ states with the chalcogen $p$, especially the $p_{3/2}$ components.
336: Anyhow, SOI not only on U but also on the chalcogen sites
337: is essential to realize such interesting nature in magnetism of the uranium
338: chalcogenides.
339:
340: We thank Takeo Jo for invaluable discussion.
341: This work was supported in part by Japan Science and Technology Corporation.
342: %
343: \begin{references}
344: \bibitem{Thole}B.~T.~Thole, P.~Carra, F.~Sette, and G.~van der Laan, Phys.\
345: Rev.\ Lett.\ {\bf 28}, 1943 (1992).
346: \bibitem{Carra}P.~Carra, B.~T.~Thole, M.~Altarelli, and X.~Wang, Phys.\
347: Rev.\ Lett.\ {\bf 70}, 694 (1993).
348: \bibitem{Blume}M.~Blume, J.\ Appl.\ Phys.\ {\bf 57}, 3615 (1985).
349: \bibitem{Jansen}See, for example, H.~J.~F.~Jansen,
350: Physics Today {\bf 4}, 50 (1995).
351: \bibitem{Handbook Actinide}J.-M.~Fournier and R.~Tro\'c, in {\it Handbook
352: on the Physics and Chemistry of the Actinides}, edited by A.~J.~Freeman and
353: G.~H.~Lander (North-Holland, Amsterdam, 1985), Vol.~2, p.~29.
354: \bibitem{Lander}G.~H.~Lander, M.~S.~S.~Brooks, B.~Lebech, P.~J.~Brown,
355: O.~Vogt, and K.~Mattenberger, Appl.\ Phys.\ Lett.\ {\bf 57}, 989 (1990).
356: \bibitem{Tillwick}D.~L.~Tillwick, P.~de V.~Plessis, J.\ Magn.\ Magn.\
357: Mater.\ {\bf 3}, 329 (1976).
358: \bibitem{Reihl}B.~Reihl, J.\ Less-Common Met.\ {\bf 128}, 331 (1987).
359: \bibitem{Schoenes}J.~Schoenes, B.~Frick, and O.~Vogt, Phys.\ Rev.\ B {\bf
360: 30}, 6578 (1984).
361: \bibitem{Huang}C.~Y.~Huang, R.~J.~Laskowski, C.~E.~Olsen, and J.~L.~Smith,
362: J.\ Physic {\bf 40}, 26 (1979).
363: \bibitem{Rudigier}H.~Rudigier, H.~R.~Ott, and O.~Vogt, Phys.\ Rev.\ B {\bf
364: 32}, 4584 (1985).
365: \bibitem{Soler}J.~M.~Soler and A.~R.~Williams, Phys.\ Rev.\ B {\bf 40},
366: 1560 (1989).
367: \bibitem{Blochl}P.~E.~Bl\"{o}chl, O.~Jepsen, and O.~K.~Andersen, Phys.\
368: Rev.\ B {\bf 49}, 16 223 (1994).
369: \bibitem{Shishidou2}T.~Shishidou and T.~Oguchi, unpublished.
370: \bibitem{Wedgwood}F.~A.~Wedgwood, J.\ Phys.\ C {\bf 5}, 2427 (1972).
371: \bibitem{Severin}L.~Severin, M.~S.~S.~Brooks, and B.~Johansson, Phys.\
372: Rev.\ Lett.\ {\bf 71}, 3214 (1993).
373: \bibitem{Shishidou}T.~Shishidou, T.~Oguchi, and T.~Jo, Phys.\ Rev.\ B {\bf
374: 59}, 6813 (1999).
375: \bibitem{quantization}Note that $j_z$ is defined by taking the quantization
376: axis
377: as the same direction as the magnetization.
378: \bibitem{S_111}As for the [111] magnetization
379: there also exist the S atoms in the magnetization direction,
380: but they are the 3rd nearest neighbors and the hybridization effect is much
381: weaker.
382: \bibitem{Lander2}G.~H.~Lander: private communication.
383: \end{references}
384: %
385: \end{document}
386: