cond-mat0512168/po15.tex
1: \documentclass[aps,prb,twocolumn,showpacs,amssymb]{revtex4}
2: \usepackage{graphicx}
3: %\usepackage{dcolumn}% Align table columns on decimal point
4: \usepackage{bm}% bold math
5: 
6: \begin{document}
7: %==============================================================================
8: 
9: \title{The origin of the stabilized simple-cubic structure in polonium}
10: 
11: \author {B. I. Min$^1$, J. H. Shim$^1$, Min Sik Park$^1$, Kyoo Kim$^1$, 
12: S. K. Kwon$^1$, and S. J. Youn$^2$} 
13: 
14: \affiliation{$^1$Department of Physics, Pohang University of Science 
15:         and Technology, Pohang 790-784, Korea \\
16:              $^2$Department of Physics Education, 
17:         Gyeongsang National University, Jinju 660-701, Korea}
18: 
19: \date{\today}
20: 
21: \begin{abstract}
22: 
23: The origin of the stabilized simple-cubic (SC) structure in Po 
24: is explored by using the first principle band calculations.
25: We have found that the prime origin is the inherent strong spin-orbit (SO) 
26: interaction in Po, which suppresses the Peierls-like structural instability
27: as usually occurs in $p$-bonded systems. 
28: Based on the systematic analysis of electronic structures, 
29: charge densities, Fermi surfaces, and susceptibilities of 
30: Se, Te, and Po, we have proved that the stable crystal structure 
31: in VIA elements is determined by the competition between the SO splitting 
32: and the crystal field splitting induced by the low-symmetry 
33: structural transition. The trigonal structure is stabilized in Se and Te by
34: the larger crystal field splitting than the SO splitting, 
35: whereas in Po the SC structure is stabilized by the large SO splitting.
36: 
37: \end{abstract} 
38: 
39: \pacs{61.66.Bi, 71.70.Ej, 71.20.Gj}
40: \maketitle
41: 
42: %==============================================================================
43: 
44: %\section{Introduction}
45: %\label{sec:intro}
46: 
47: Among the elements in the periodic table,
48: only Po (polonium) is known to crystallize in the simple-cubic (SC) 
49: structure in nature \cite{beamer}. 
50: Radioactive Po, which was discovered by Pierre and Marie Curie in 1898,
51: requires difficult sample preparation, and
52: so not many experimental and theoretical studies have been carried out.
53: Po, which is located in the VIA column of the periodic table,
54: has a valence electron configuration of $6s^26p^4$ ($Z$=84),
55: and exhibits two metallic phases of $\alpha$ and $\beta$
56: which are formed in the SC and the 
57: rhombohedral (trigonal) structure below and above $\sim$ 348$K$,
58: respectively \cite{max,des}.
59: Note that the isoelectronic elements located in the same VIA column,
60: Se (selenium) and Te (tellurium), crystallize in the trigonal 
61: structure (space group $P3_{1}21$) with the helical chain 
62: arrangements of atoms, which run parallel to 
63: the crystallographic $c$ axis of the hexagonal setting.
64: Hence the coordination number is two in trigonal Se and Te 
65: in contrast to six in simple-cubic Po (SC-Po). The ratio of the
66: intrachain ($d_1$) bond length to the interchain ($d_2$) bond length 
67: is $\frac{d_2}{d_1} = 1.50, 1.23$, and $1.0$ for Se, Te, and Po, 
68: respectively. The coordination number of two in trigonal Se and Te can be
69: understood based on the simple octet rule \cite{Gaspard}.
70: 
71: The reason why metallic Po has a stable SC structure
72: has not been fully addressed yet.
73: By using the relativistic parameterized extended H\"{u}ckel method,
74: Lohr \cite{lohr} has found a hint that the helical structure 
75: as a distortion of a SC structure might be 
76: quenched in the case of Po due to its very large spin-orbit (SO) coupling.
77: However, their calculation was not self-consistent and
78: adopted the atomic value of the SO coupling parameter. 
79: They did not study the structural energetics.
80: Recently, using the {\it ab-initio} pseudopotential band method,
81: Kraig {\it et al.} \cite{kraig} showed that 
82: the SC structure is preferred by Po to face-centered or
83: body-centered cubic structure.
84: They argued that the large $s$-$p$ splitting in Po would produce
85: the stable SC structure. However, they could not
86: explain why this happens only in Po but
87: not in other elements with similarly large $s$-$p$ splittings.
88: Moreover, the SO interaction was not taken into account in their
89: band calculations. Lach-hab {\it et al.} \cite{Lach} studied
90: the structural energetics for Po using the tight-binding (TB) band method.
91: After having determined the TB parameters by fitting the TB bands
92: to the semirelativistic full-potential linearized augmented plane-wave 
93: (FLAPW) \cite{wim} bands, they took account of the 
94: SO effect {\it a posteriori} by employing the atomic value of the
95: SO coupling parameter.
96: They found that the SC structure is the most stable among
97: the close-packed structures for both cases
98: with and without the SO effect included. They did not 
99: consider the lower symmetry structures with the internal atomic relaxation,
100: as observed in Se and Te. 
101: 
102: %-----------------------------------------
103: \begin{figure}[b]
104: \includegraphics[scale = 0.33,angle=270]{fig1.ps}
105: \caption{(a) A hexagonal unit cell with tripled volume of 
106: a SC unit cell. 
107: (b) The hexagonal Brillouin zone (BZ) compared to the SC BZ.
108: Note that $\Gamma$-A and $\Gamma$-L in the hexagonal BZ correspond to 
109: $\frac{1}{3}\Gamma$-R and $\Gamma$-X in the SC BZ.
110: (c) Schematic total energy curves for Se, Te, and Po in the hypothetical
111: structural space.
112: T and SC represent trigonal and simple-cubic structures,
113: respectively.
114: Both in the GGA and GGA+SO, trigonal Se and Te are more stable.
115: For Po, trigonal Po is more stable in the GGA, while
116: SC-Po becomes more stable in the GGA+SO.
117: }
118: \label{fig1}
119: \end{figure}
120: %-----------------------------------------
121: 
122: %-----------------------------------------
123: \begin{figure}[t]
124: \includegraphics[scale = 0.55]{fig2.ps}
125: \caption{(a)-(b) The GGA and GGA+SO band structures of SC-Po.
126: (c)-(d) DOSs of SC-Po and trigonal Po. 
127: The latter corresponds to the stable phase
128: obtained after the geometry optimization in the GGA scheme.
129: In each frame, DOSs obtained by the GGA and GGA+SO schemes 
130: are compared.
131: (e)-(f) The GGA and GGA+SO band structures of SC-Po drawn in the
132: frame of the hexagonal BZ. There are several band crossings 
133: at $E_F$ in the GGA, but not in the GGA+SO.
134: }
135: \label{fig2}
136: \end{figure}
137: %-----------------------------------------
138: 
139: A systematic crystallographic transformation can occur from the 
140: SC lattice to the trigonal lattice.
141: As shown in Fig.~\ref{fig1}(a),
142: the SC lattice can be described by a hexagonal lattice
143: in which the hexagonal $a$ axis corresponds to the face diagonal
144: and the hexagonal $c$ axis to the body diagonal of cube.
145: Then the size of the unit cell becomes tripled 
146: with $c/a=\sqrt\frac{3}{2}$. The trigonal structure of $\beta$-Po
147: corresponds to the elongated SC structure simply along the body diagonal
148: direction. In the trigonal structures of Se and Te, the internal atomic 
149: position parameters are also changed in addition to elongation \cite{Kresse}. 
150: 
151: To explore the origin of stabilized SC structure in Po,
152: we have investigated the electronic and structural properties of Se, Te, and
153: Po systematically by using the first-principles FLAPW band method
154: incorporating the SO interaction. 
155: For the calculations, the QMD-FLAPW \cite{wim} and 
156: the WIEN2K \cite{Blaha} codes are utilized. 
157: The band structure results obtained from 
158: both codes are qualitatively coincident with each other.
159: Below we will use mainly WIEN2K results, 
160: since the local orbital basis is implemented in WIEN2K in order
161: to describe the relativistic $6p_{1/2}$ wave function 
162: correctly \cite{Singh}.
163: The exchange-correlation interaction is treated by
164: the generalized gradient approximation (GGA)\cite{Perdew,LDA}.
165: 
166: We have first studied the structural energetics by optimizing
167: the atomic positions for both trigonal
168: and SC structures of Se, Te, and Po 
169: with (GGA+SO) and without (GGA) the SO interaction included.
170: In the GGA scheme, the trigonal structures are
171: more stable than SC structures by 0.092, 0.031, and 0.004 eV/atom
172: for Se, Te, and Po, respectively.
173: The GGA+SO scheme, however, yields different results.
174: The incorporation of the SO coupling reduces the stability of the 
175: trigonal structure.
176: For Se and Te, the trigonal structure is still more stable
177: than the SC structure by 0.090 and 0.021 eV/atom, respectively.
178: In contrast, for Po, the SC structure 
179: becomes more stable than the trigonal structure by 0.016 eV/atom. 
180: Thus the total energy behaves as in Fig.~\ref{fig1}(c).
181: There are two local minima in the hypothetical structural space,
182: the trigonal (T) and the SC structure.
183: The SO coupling effect does not change the global minima 
184: in Se and Te, but changes that in Po from trigonal Po to SC-Po.
185: This finding indicates that the SO interaction plays an essential
186: role in stabilizing the SC structure of Po.
187: 
188: Indeed band structures in Fig.~\ref{fig2}(a)-(b) reveal that the effect of
189: the SO interaction is substantial for SC-Po, as compared to that for Se and Te.
190: The lowest bands in Fig.~\ref{fig2}(a)-(b)
191: correspond to Po $6s$ states, while the bands near the Fermi level ($E_F$) 
192: to Po $6p$ states.
193: Due to the large energy gap between $6s$ and $6p$ states,
194: there will not be a considerable mixing between them,
195: and thus Po can be classified as a $p$-bonded system.
196: When the SO interaction is taken into account, 
197: the degenerate $6p$ bands along $\Gamma$-X, $\Gamma$-M, and $\Gamma$-R are 
198: split so that the doubly degenerate 
199: $6p_{1/2}$ band is separated from the upper half-filled
200: $6p_{3/2}$ band.
201: Due to the SO splitting, the small hole Fermi surface at $\Gamma$ disappears.
202: The SO splitting of $p$ states at $\Gamma$ 
203: for SC-Po is as much as 3.16 eV, which is much larger
204: than those for SC structures of Se and Te, 0.53 and 1.05 eV.
205: 
206: In Fig.~\ref{fig2}(c)-(d), we have compared the density of states (DOS) 
207: of SC-Po and that of trigonal Po, which are obtained 
208: by the GGA and GGA+SO schemes.
209: Trigonal Po here corresponds to the stable phase determined by the 
210: geometry optimization in the GGA scheme.
211: Notable in the GGA-DOS is the pseudogap feature at $E_F$ 
212: for trigonal Po, which would give rise to the semi-metallic behavior. 
213: The pseudogap appears between the nonbonding lone-pair
214: and the antibonding Po $6p$ state,
215: because of the splitting between the bonding
216: and antibonding Po $6p$ state in the chain arrangement of trigonal 
217: Po \cite{joan}. 
218: Hence the DOS at $E_F$, $N(E_F)$, becomes much reduced in trigonal Po 
219: from that in SC-Po.  In fact, the stable trigonal phases of Se and Te
220: result from this gap formation near $E_F$.
221: In the case of Se, the energy splitting between the bonding
222: and the antibonding state is so large that
223: the system becomes insulating, while Te is semimetallic.
224: 
225: In the GGA+SO scheme, the $6p_{1/2}$ state is separated out, 
226: and so the band shape and N($E_F$) are changed a lot. 
227: Occupied band widths are broadened with more
228: weight at low energy side, which produces energy gains 
229: for both SC-Po and trigonal Po.
230: The stable SC-Po in the GGA+SO scheme reflects that 
231: the SO induced energy gain is larger for SC-Po than for trigonal Po.
232: It is because $N(E_F)$ for trigonal Po is rather enhanced 
233: producing some energy loss.
234: Note that, for Se and Te,  both the GGA and GGA+SO schemes
235: yield similar DOS, implying no noticeable energy gain from the SO interaction.
236: These results suggest that the stable crystal structure
237: in VIA elements is mainly determined by the
238: competition between the SO splitting and the crystal field splitting 
239: due to the low symmetry structure.
240: That is, for Se and Te, the larger crystal field splitting 
241: between the intra-chain bonding and antibonding states 
242: stabilizes the trigonal structure,
243: whereas, for Po, the larger SO splitting stabilizes the SC structure.
244: 
245: The stable trigonal structure of Se and Te can be understood in terms of
246: the three dimensional (3D) Peierls distortion of the SC structure
247: \cite{Gaspard,Kresse,Burdett}.
248: In general, $p$-bonded systems would favor the SC structure to maximize the 
249: directional $p_{\sigma}$ bonding. 
250: However, three mutually orthogonal linear chains in the SC structure
251: will be easily distorted by the Peierls mechanism 
252: when the linear chain is partially filled. 
253: In the case of Se and Te, each chain is 2/3-filled, 
254: and so the energy gain can be achieved by the trimerization with 
255: the short-long-long 
256: bond alternation. Decker {\it et al.} \cite{Decker} recently
257: demonstrated this mechanism for Te by comparing the band structures of
258: the undistorted SC and the distorted trigonal structure.
259: The structural distortion from the SC to the trigonal structure
260: induces a splitting of degenerate bands near $E_F$, which yields
261: the energy gain for trigonal Te.
262: 
263: %-----------------------------------------
264: \begin{figure}[t]
265: \includegraphics[scale = 0.4, angle=270]{fig3.ps}
266: \caption{The occupied charge density distributions 
267: of Po $6p$ bonding states in the energy range of $-6.0 \sim -3.0$ eV.
268: (a) SC-Po (GGA), (b) trigonal Po (GGA), (c) SC-Po (GGA+SO), 
269: and (d) trigonal Po (GGA+SO).
270: }
271: \label{fig3}
272: \end{figure}
273: %-----------------------------------------
274: 
275: %-----------------------------------------
276: \begin{figure}[t]
277: \includegraphics[scale = 0.37, angle=270]{fig4.ps}
278: \caption{Fermi surfaces of SC-Po derived from the third (green) 
279: and fourth (blue) bands in the GGA (a) and the GGA+SO (b).
280: (c) Susceptibilities of SC-Po in the GGA and the GGA+SO.
281: }
282: \label{fig4}
283: \end{figure}
284: %-----------------------------------------
285: 
286: We have examined the above scenario for SC-Po.
287: Figure~\ref{fig2}(e)-(f) display the GGA and GGA+SO band structures 
288: of undistorted SC-Po which are projected in the hexagonal BZ.
289: The GGA band of Fig.~\ref{fig2}(e) shows several band crossings at $E_F$ 
290: which are created by the back-folding of bands of
291: the larger SC BZ into the smaller hexagonal BZ (see Fig.~\ref{fig1}(b)).
292: Then, as in Te \cite{Decker}, a Peierls-like structural distortion would
293: split the band crossings at $E_F$ to yield the energy gain.
294: In contrast, the GGA+SO band of Fig.~\ref{fig2}(f) shows the band crossings 
295: not at $E_F$ but at higher or lower energy side far apart from $E_F$.
296: This is because of the substantial band shift by the SO splitting.
297: Accordingly, no energy gain would result from the Peierls-like structural 
298: distortion.
299: This proves how the SO interaction stabilizes the SC structure of Po.
300: When Decker {\it et al.} \cite{Decker} demonstrated
301: this mechanism for Te, they did not take into account the
302: SO interaction. The SO induced band shift in Te, however, is not 
303: large enough to suppress the Peierls-like instability.
304: 
305: The bonding characters can be investigated by the charge density plot.
306: Figure~\ref{fig3} shows the occupied charge density distributions 
307: of Po $6p$ bonding states in the energy range of $-6.0 \sim -3.0$ eV
308: for both SC-Po and trigonal Po.
309: They are plotted on the special plane of hexagonal unit cell
310: to see the bonding character along the helical zig-zag chains 
311: in the $c$-direction.  The GGA charge density for SC-Po
312: (Fig.~\ref{fig3}(a)) shows larger bonding character between Po atoms 
313: than the GGA+SO charge density (Fig.~\ref{fig3}(c)).
314: This implies that the directional bonding between atoms becomes weakened
315: when incorporating the SO interaction, and so the charge density becomes 
316: more isotropic.
317: This feature is more clearly seen in trigonal Po which manifests
318: the prominent chain nature.
319: The GGA charge density for trigonal Po (Fig.~\ref{fig3}(b))
320: shows a pronounced intrachain bonding character along the zig-zag chains.
321: In the GGA+SO charge density (Fig.~\ref{fig3}(d)),
322: the SO interaction weakens the intrachain directional bondings 
323: so that the anisotropic chain nature is suppressed.
324: This is consistent with the enhanced $N(E_F)$ in the
325: GGA+SO DOS in Fig.~\ref{fig2}(d).
326: The charge density plots reveal that the SO interaction weakens 
327: the directional bondings between Po atoms and so suppresses 
328: the 1D chain nature and the corresponding Peierls-like structural instability.
329: 
330: The structural transition can also be studied by analyzing the 
331: behavior of the charge susceptibility $\chi_0(q)$.
332: $\chi_0(q)$'s in Fig.~\ref{fig4}(c) are obtained from the band structures
333: of SC-Po.
334: $\chi_0(q)$ in the GGA scheme shows the highest peak near
335: $\vec{Q} =\frac{2}{3}$$\Gamma$-R, indicating a possible structural instability 
336: at this $\vec{Q}$. 
337: This $\vec{Q}$ vector is coincident with the reciprocal lattice vector
338: connecting the hexagonal BZ boundaries including A (see 
339: Fig.~\ref{fig1}(b)), at which the bands of SC-Po
340: along $\Gamma$-R are folded back.
341: Therefore, this peak is closely related to the trigonal 
342: distortion of the SC structure along the (111) direction.
343: Noteworthy in Fig.~\ref{fig4}(c) is the substantial 
344: reduction of this peak of $\chi_0(q)$ when incorporating the SO interaction.
345: The absence of the structural instability in Po
346: would be closely correlated to this behavior of $\chi_0(q)$.
347: One can also note the peaks in $\chi_0(q)$ 
348: near $\vec{q} =\frac{2}{3}$$\Gamma$-X, $\frac{2}{3}$$\Gamma$-M,
349: $\frac{2}{3}$X-M. These peaks and the peak near $\vec{Q}$
350: are expected to arise from the quasi-1D Fermi surfaces of SC-Po,
351: which are formed by the third and fourth bands in Fig.~\ref{fig2}(a).
352: Namely, each $\vec{q}$ producing the peak in $\chi_0(q)$ corresponds to the 
353: nesting vector of the quasi-1D Fermi surfaces projected in each symmetry plane.
354: The GGA Fermi surfaces in Fig.~\ref{fig4}(a), which are drawn in the (220)
355: plane of the SC BZ, exhibit the clear quasi-1D nature
356: with the corresponding nesting vectors.
357: As shown in Fig.~\ref{fig4}(b), the SO interaction breaks 
358: these quasi-1D Fermi surfaces into pieces, and so reduces the nesting effect.
359: This situation in Po is different from that in Se and Te, 
360: in which the 1D nature is preserved even with the SO interaction, so that 
361: the nesting effect is still active.
362:  
363: Finally, it is worthwhile to discuss the stable structures of 
364: neighboring elements of Po, Bi (Z=83) and 
365: At (Z=85), both of which are also $p$-bonded metallic systems 
366: with the large SO interaction.
367: In nature, Bi crystallizes in the trigonal ($\alpha$-arsenic) 
368: structure \cite{Shick}. As for radioactive At, no structural information is 
369: available because of its too short half-life of only 8.3 hours.
370: $\alpha$-arsenic structure of Bi can be understood as arising from
371: the Peierls-like distortion, as discussed above \cite{Burdett,Gaspard}. 
372: Since the $6p$-bands in Bi are half-filled, the tendency of the
373: Peierls instability is predominating so as to drive the structural distortion
374: despite the large SO interaction.
375: In the case of At, the $6p$-bands are 5/6-filled so that
376: the SO interaction would dominate over the Peierls instability
377: as in Po.
378: Hence the SC structure is expected to be stabilized for At.
379: 
380: In conclusion, we have clarified that the origin of the stabilized 
381: SC structure in Po is its inherent strong SO interaction.
382: The large SO interaction in Po weakens the directional bondings between atoms 
383: so as to suppress the Peierls-like distortion.
384: Our study also reveals that the stable crystal structure in VIA elements is 
385: determined by the competition between the SO splitting and the crystal 
386: field splitting induced by the structural transition to a lower symmetry.
387: For Se and Te, the larger crystal field splitting between the intra-chain
388: bonding and the antibonding state stabilizes the trigonal structure,
389: whereas, for Po, the larger SO splitting stabilizes 
390: the SC structure.
391: 
392: %Acknowledgments $-$
393: This work was supported by the SRC program of MOST/KOSEF
394: and in part by the KRF.
395: 
396: \begin{thebibliography}{99}
397: 
398:    \bibitem{beamer} W. H. Beamer and C. R. Maxwell, 
399:            J. Chem. Phys. {\bf 14}, 569 (1946).
400:    \bibitem{max} C. R. Maxwell, J. Chem. Phys. {\bf 17}, 1288 (1949).
401:    \bibitem{des} R. J. Desando and R. C. Lange, 
402:             J. Inorg. Nucl. Chem. {\bf 28}, 1837 (1966).
403:  \bibitem{Gaspard} J.-P. Gaspard, A. Pellegatti, F. Marinelli, C. Bichara C,
404: 	 Phil. Mag. B {\bf 77}, 727 (1998).
405:    \bibitem{lohr} L. L. Lohr, Inorg. Chem. {\bf 26}, 2005 (1987).
406:    \bibitem{kraig} R. E. Kraig, D. Roundy, and M. L. Cohen,
407:             Solid State Comm. {\bf 129}, 411 (2004).
408:    \bibitem{Lach} M. Lach-hab, B. Akdim, D.A. Papaconstantopoulos, M.J. Mehl, 
409: 	N. Bernstein, J. Phys. Chem. Solid {\bf 65}, 1837 (2004).
410:    \bibitem{Kresse} G. Kresse, J. Furthmuller, and J. Hafner,
411: 	Phys. Rev. B {\bf 50}, 13181 (1994).
412:    \bibitem{wim} E. Wimmer, H. Krakauer, M. Weinert, and A. J. Freeman,
413:             Phys. Rev. B {\bf 24}, 864 (1981);
414: 	M. Weinert, E. Wimmer, and A. J. Freeman, {\it ibid.} {\bf 26}, 
415: 	4571(1982); H. J. F. Jansen and A. J. Freeman {\it ibid.} {\bf 30},
416: 	561 (1984).
417:  \bibitem{Blaha} P. Blaha, K. Schwarz, G. K. H. Madsen, K. Kvasnicka, 
418: 	and J. Luitz, WIEN2K (Karlheinz Schwarz, Technische Universitat Wien,
419: 	 Austria, 2001).
420:  \bibitem{Singh} D. Singh, Plane waves, Pseudopotentials,
421: 	 and the LAPW Method (Kluwer Academic, Boston, 1994).
422:  \bibitem{Perdew} J.P. Perdew and Y. Wang, 
423:             Phys. Rev. B {\bf 45}, 13244 (1992).
424:  \bibitem{LDA} We have found that the GGA describes the ground states of
425: 	Se, Te, and Po better than the LDA.
426:  \bibitem{joan} J. D. Joannopoulos, M. Schl\"{u}ter, and M. L. Cohen,
427:             Phys. Rev. B {\bf 11}, 2186 (1975).
428:  \bibitem{Burdett} J. K. Burdett and S. Lee, J. Amer. Chem. Soc.,
429: 	  {\bf 105}, 1079 (1983).
430:  \bibitem{Decker} A. Decker, G. A. Landrum, and R. Dronskowski,
431: 	Z. Anorg. Allg. Chem., {\bf 628}, 295 (2002).
432:  \bibitem{Shick} A. B. Shick, J. B. Ketterson, D. L. Novikov, 
433: 	and A. J. Freeman, Phys. Rev. B {\bf 60}, 15484 (1999).
434: 
435: %            Nature {\bf 395}, 677 (1998).
436: %            Appl. Phys. Lett. {\bf 74}, 1737 (1999).
437: %            Phys. Rev. Lett. {\bf 74}, 1171 (1995).
438: %            Solid State Comm. {\bf 124}, 77 (2002).
439: %            Rev. Mod. Phys. {\bf 40} 790 (1968).
440: %            Physica C {\bf 341-348}, 785 (2000).
441: 
442: \end{thebibliography}
443: 
444: 
445: \end{document}
446: