1: %
2: % ZnBeSe by SIESTA
3: % version of Dec 16, 2004
4: %
5: %\documentclass[preprint,aps,showpacs,a4paper]{revtex4}
6: \documentclass[prb,aps,showpacs,twocolumn,a4paper,floatfix]{revtex4}
7: %\documentclass[prb,aps,showpacs,a4paper]{revtex4}
8: \usepackage{epsfig}
9: \usepackage{dcolumn}
10: \begin{document}
11: \title{Lattice dynamics of mixed semiconductors (Be,Zn)Se
12: from first-principles calculations}
13: \author{A.~V.~Postnikov}
14: \email{apostnik@uos.de}
15: \affiliation{Institute of Metal Physics, S. Kowalewskoj 18,
16: Yekaterinburg 620219, Russia, and
17: Universit\"at Osnabr\"uck --
18: Fachbereich Physik, D-49069 Osnabr\"uck, Germany}
19: \author{Olivier Pag\`es, Joseph Hugel}
20: \affiliation{Universit\'e de Metz - Institut de Physique,
21: 1 Bd Arago, F-57078 Metz cedex 3, France}
22: \date{\today}
23: %
24: \begin{abstract}
25: Vibration properties of Zn$_{1-x}$Be$_x$Se, a mixed II-VI semiconductor
26: characterized by a high contrast in elastic properties of its pure
27: constituents, ZnSe and BeSe, are simulated by first-principles calculations
28: of electronic structure, lattice relaxation and frozen phonons.
29: The calculations within the local density approximation has been done
30: with the {\sc Siesta} method, using norm-conserving pseudopotentials and
31: localized basis functions; the benchmark calculations for pure endsystems
32: were moreover done also by all-electron WIEN2k code.
33: An immediate motivation for the study was to analyze, at the microscopic level,
34: the appearance of anomalous phonon modes early detected in Raman spectra
35: in the intermediate region (20 to 80\%) of ZnBe concentration.
36: This was early discussed on the basis of a percolation phenomenon,
37: i.e., the result of the formation of wall-to-wall --Be--Se-- chains
38: throughout the crystal. The presence of such chains was explicitly allowed
39: in our simulation and indeed brought about a softening and splitting off
40: of particular modes, in accordance with experimental observation, due to
41: a relative elongation of Be--Se bonds along the chain as compared to those
42: involving isolated Be atoms. The variation of force constants with interatomic
43: distances shows common trends in relative independence on the short-range
44: order.
45: \end{abstract}
46: %
47: \pacs{
48: 63.20.-e, % Phonons in crystal lattices
49: 63.50.+x, % Vibrational states in disordered systems
50: 71.15.-m, % Methods of electronic structure calculations
51: 71.55.Gs, % II-VI semiconductors
52: 78.30.-j % Infrared and Raman spectra
53: }
54: \maketitle
55: %
56: \section[#1]{Introduction}
57:
58: (Be,Zn)Se is an example of mixed II-VI semiconductor system
59: whose electronic properties and, consequently, elastic characteristics
60: vary with concentration in a non-trivial way. A considerable effort
61: has been spent on characterizing the optical band gap, which drops
62: slightly from intermediate concentrations
63: and undergoes a change from direct to indirect character underway
64: from ZnSe to BeSe \cite{PRB61-5332}.
65: Not less interesting possibilities open in the tuning of elastic properties,
66: because BeSe and ZnSe, similarly to some other easily miscible
67: II-VI constituents, possess quite different elastic constants, which have
68: to be accommodated in a solid solution. (Be,Zn)Se makes, actually,
69: quite an extreme case of elastic contrast between constituents in a mixed
70: system. This was argued to favor the appearance of anomalous lines in the
71: Raman spectra, which split off on the soft side of the BeSe-related TO peak
72: in the $x\sim$0.2--0.7 concentration range of Be$_x$Zn$_{1-x}$Se crystals,
73: as was detected by Pag\`es \emph{at al.} \cite{PRB65-035213}.
74: In order to explain this anomaly, Pag\`es \emph{et al.} \cite{PRB70-155319}
75: brought into discussion an idea of quasi-infinite chains, which are developing
76: in the softer matrix (ZnSe) as the BeSe concentration reaches
77: the percolation threshold of $\sim$20\%.
78: However, the microscopic mechanism relating percolation with
79: vibration properties remained speculative. The present work's aim
80: is to provide a possible complete and reliable first-principles description
81: of internal tensions, structural relaxation, and the impact
82: of the latter on lattice-dynamical properties of a mixed Be$_x$Zn$_{1-x}$Se
83: alloy near the percolation threshold. A concise result of our simulation
84: has been reported earlier as a conference proceeding \cite{Dui03-ZBS}.
85: Here we offer a detailed outline of results obtained for different
86: supercells, with a more lengthy discussion.
87: In particular, we provide an analyzis of bond lengths distribution
88: for a number of compositions, discuss the change of force constants
89: related with that, and finally calulate phonon density of states and
90: provide its the wavevector-resolved decomposition.
91:
92: The electronic properties of pure constituent compounds are well known;
93: they also were probed by \emph{ab initio} methods of respectful accuracy
94: in a number of earlier publications.
95: Simulations of comparable accuracy on mixed alloys are seriously complicated
96: by the necessity to treat large supercell, with a practical loss of
97: any useful symmetry. A certain success has been achieved in molecular
98: dynamics (MD) simulations of an isostructural III-V mixed semiconductor
99: alloy, (Ga,In)As: Branicio \emph{et al.}\cite{APL82-1057} performed
100: parametrized MD calculations for supercells with up to 8000 atoms, and
101: \emph{ab initio} MD calculations on 216-atom supercells.
102: We are not aware of any elastic simulations
103: on mixed II-VI semiconductors but that by Tsai \emph{et al.}
104: \cite{PRB65-235202}, who used multicenter MD method,
105: a formally \emph{ab initio} one but of inferior accuracy to the method
106: applied here. Comparing our results with those of Tsai \emph{et al.},
107: we find an overall good agreement, albeit with quantitative discrepancies
108: on some sensitive issues. A more important difference is, however, that
109: we study more attentively the limit of low Be concentration, particularly
110: the onset of percolation on the Be sublattice. Moreover, we
111: calculate phonon densities of states and discuss them in immediate
112: reference to experimentally obtained Raman spectra.
113:
114: The paper is organized as follows.
115: In Sec.~\ref{sec:method} we outline our \emph{ab initio} calculation approach
116: to electronic structure and vibration properties.
117: Sec.~\ref{sec:pure}, containing the results for pure endsystems ZnSe and BeSe,
118: is concise in view of numerous previous studies, but it is necessary here
119: in order to assess the accuracy of the calculation method.
120: In Sec.~\ref{sec:bonds} we outline our results for lattice parameters
121: and equilibrium bond lengths in relaxed supercells, simulating a broad
122: concentration range.
123: The effect of bond lengths on the force constants between nearest
124: and next-nearest neighbors is discussed in Sec.~\ref{sec:forces}.
125: Finally, the manifestation of the force constants in phonon frequencies
126: is explained in Sec.~\ref{sec:phonons}, along with the discussion
127: on wavevector dependency and vibration patterns in different parts
128: of the phonon spectrum.
129:
130: \section[#2]{Calculation scheme and setup}
131: \label{sec:method}
132: %
133: Dynamical properties are derived from the total energy or forces,
134: which are evaluated \emph{ab initio} in a sequence of
135: density functional calculations. We applied the calculation method,
136: and computer code, {\sc Siesta}\cite{JPCM14-2745},
137: which incorporates norm-conserving pseudopotentials in combination
138: with atom-centered strictly confined numerical basis
139: functions\cite{JPCM8-3859,PRB64-235111}.
140: The pseudopotentials were constructed along the Troullier--Martins
141: scheme\cite{PRB43-1993} for the following valence-states configurations:
142: Be$\,2s^2(2.0)$, Se$\,3d^{10}(1.2)\,4s^2(1.9)\,4p^4(2.0)$,
143: Zn$\,3d^{10}(1.09)\,4s^2(2.28)$. The numbers in brackets stand for
144: pseudoization radii of corresponding states in Bohr.
145: The basis set consisted of double-$\zeta$ functions with polarization
146: orbitals in all $l$ channels on all centers (Be$\,2s2p$, Zn$\,3d4s4p$,
147: Se$\,3d4s4p$), hence also for explicitly included semicore Se$3d$ states.
148: The details on the basis classification and testing in {\sc Siesta}
149: can be found in Refs.~\onlinecite{JPCM8-3859,PRB64-235111}.
150: Our basis generation followed the standard split schema
151: of the {\sc Siesta} code, with the energy shift parameter, responsible
152: for the localization of basis functions, equal to 20 mRy. This resulted
153: in basis functions with maximal extension of 3.17 {\AA} (Be),
154: 3.12 {\AA} (Zn) and 2.82 {\AA} (Se), i.e. well beyond the nearest-neighbor
155: distances but short of direct overlap to next-nearest neighbors.
156: The full structure optimization
157: (of cell parameters and internal coordinates) was performed in all
158: supercells prior to lattice dynamics calculations. In order to get reliable
159: forces on atoms it is essential to accurately perform spatial integration
160: of the (general-form) residual charge density over the unit cell.
161: To this end, we used the mesh cutoff parameter of 250 Ry, that generated
162: a real-space mesh with the step of about 0.1 {\AA} along each Cartesian
163: direction throughout the supercell. Within each mini-cell of this real-space
164: mesh, an averaging of integration results was done over four fcc-type sampling
165: points for better numerical stability. This resulted in the forces
166: summing up, over all atoms in the supercell, to zero within the accuracy of
167: $\pm 0.1$ eV/{\AA}, separately along each Cartesian direction.
168: The thus validated forces were used to calculate the dynamical matrix
169: elements, by introducing small (0.016 {\AA}) deviations of atoms
170: one by one from their equilibrium positions. The details of phonon calculations
171: are discussed in Sec.~\ref{sec:phonons}.
172:
173: All our {\sc Siesta} calculations have been done in the
174: local density approximation (LDA).
175: For the pure ZnSe and BeSe systems, moreover, we used the full-potential
176: augmented plane wave method (see, e.g., Ref.~\onlinecite{Singh_bookPWPP}),
177: as implemented in the WIEN2k code\cite{wien2k}. We apply here this all-electron
178: method of recognized accuracy for benchmark calculations of elastic properties.
179: We provide the WIEN2k results obtained both within the LDA and
180: in the generalized gradient approximation (GGA, Ref.~\onlinecite{PRL77-3865}).
181:
182: \begin {table*}
183: \caption{\label{tab:elastic}
184: Elastic properties of ZnSe and BeSe from experiments and
185: first-principles calculations.
186: }
187: %\begin{center}
188: %\begin{tabular}{l@{\extracolsep\fill}ccccccc}
189: \begin{ruledtabular}
190: %\begin{tabular}{ldddcddd}
191: \begin{tabular}{lccccccc}
192: %\hline
193: \rule[-2mm]{0mm}{7mm}
194: & \multicolumn{3}{c}{ZnSe} && \multicolumn{3}{c}{BeSe} \\
195: \cline{2-4} \cline{6-8}
196: Method & $a$ ({\AA}) & $B$ (Kbar) & $\omega_{\mbox{\tiny TO}}$ (cm$^{-1}$) &&
197: %\rule[-2mm]{0mm}{7mm}
198: $a$ ({\AA}) & $B$ (Kbar) & $\omega_{\mbox{\tiny TO}}$ (cm$^{-1}$) \\
199: \hline
200: WIEN2k (LDA) & 5.568 & 714 & 198 && 5.084 & 920 & 498 \\
201: WIEN2k (GGA) & 5.571 & 727 & 206 && 5.182 & 816 & 523 \\
202: {\sc Siesta} (LDA) & 5.587 & 758 & 203 && 5.078 & 912 & 500 \\
203: %
204: exp.
205: % ---- Exp. for ZnSe:
206: & 5.668\footnotemark[01]
207: & 624\footnotemark[02]
208: & 207,\footnotemark[03]
209: 205\footnotemark[04]
210: % ---- Exp. for BeSe:
211: && 5.137\footnotemark[05]
212: & 920\footnotemark[05]
213: & 501\footnotemark[04] \\
214: %
215: other calc.
216: % ---- Other calc. for ZnSe:
217: & 5.677,\footnotemark[06]
218: 5.638,\footnotemark[07]
219: & 689,\footnotemark[06]
220: 652,\footnotemark[07]
221: & 224\footnotemark[08]
222: % ---- Other calc. for BeSe:
223: && 5.037\footnotemark[09]
224: & 988\footnotemark[09]
225: & 547\footnotemark[10] \\
226: % ---- Other calc. for ZnSe (continued in the next line):
227: & 5.633,\footnotemark[08]
228: 5.636\footnotemark[11]
229: & 811,\footnotemark[08]
230: 649\footnotemark[11] \\
231: %\hline
232: \end{tabular}
233: \end{ruledtabular}
234: \footnotetext[01]{Ref.~\onlinecite{LB17b}}
235: \footnotetext[02]{Ref.~\onlinecite{JAP41-2988}}
236: \footnotetext[03]{Ref.~\onlinecite{APL77-519}}
237: \footnotetext[04]{Ref.~\onlinecite{CrRT38-359}}
238: \footnotetext[05]{Ref.~\onlinecite{PRB52-7058}}
239: \footnotetext[06]{LDA, norm-conserving pseudopotential
240: with Zn$3d$ as valence states, Ref.~\onlinecite{PRB51-10610}}
241: \footnotetext[07]{LDA, norm-conserving pseudopotential with Zn$3d$ and Se $3d$
242: as valence states, Ref.~\onlinecite{PRB52-1459}}
243: \footnotetext[08]{LDA, full-potential LMTO, Ref.~\onlinecite{PRB50-14881}}
244: \footnotetext[09]{LDA, norm-conserving pseudopotential,
245: Ref.~\onlinecite{PRB54-11861}}
246: \footnotetext[10]{LDA, norm-conserving pseudopotential,
247: Ref.~\onlinecite{PRB55-14043}}
248: \footnotetext[11]{full-potential LMTO, Ref.~\onlinecite{PRB64-035206}}
249: \end{table*}
250:
251: \section[#3]{Endsystems, Z\lowercase{n}S\lowercase{e}
252: and B\lowercase{e}S\lowercase{e} }
253: \label{sec:pure}
254:
255: \begin{figure}[b]
256: \centerline{\epsfig{figure=fig01.eps,width=8.5cm}}
257: \caption{\label{fig:elastic}
258: The total energy vs. lattice constant curves (left)
259: and the energy profiles corresponding to the $\Gamma$-TO phonon (right)
260: as calculated for ZnSe and BeSe.
261: $\Delta_x$ indicates the magnitude of the cation (or anion)
262: displacement along [100] from the equilibrium.
263: Open squares: WIEN2k (LDA) results; connected black dots: {\sc Siesta}
264: (LDA) results.
265: }
266: \end{figure}
267: %
268: The electronic structure of zincblende-structure semiconductors
269: ZnSe and BeSe was extensively studied by both experimental means
270: and first-principle calculations.
271: Band dispersions and partial densities of states have been calculated
272: in the LDA by Lee \emph{et al.}\cite{PRB52-1459}
273: and Agrawal \emph{et al.}\cite{PRB50-14881} (for ZnSe, among other
274: Zn chalcogenides) and by Fleszar and Hanke\cite{PRB62-2466} (for BeSe,
275: among other Be chalcogenides). In the latter publication, also the
276: quasiparticle band structure has been calculated in the $GW$
277: approximation, and subsequently discussed in the context of LDA vs.
278: $GW$ and exact exchange for a number of $sp$ zincblende
279: semiconductors\cite{PRB64-245204}. For ZnSe, the quasiparticle
280: band structure has been reported by Luo \emph{et al.}\cite{PRB66-195215}.
281: Our calculations provide the band structures in agreement with
282: good earlier LDA results, so we skip the discussion on this point.
283: Instead we turn to elastic properties and note that the optimized
284: ground-state volume along with the bulk modulus have been reported
285: in many calculations for ZnSe (see Table 1), and also
286: by Mu{\~n}oz \emph{et al.}\cite{PRB54-11861} and Gonz\'ales-D{\'{\i}}az
287: \emph{et al.}\cite{PRB55-14043} for BeSe. The latter calculations
288: largely overestimate the stiffness of the BeSe crystal, probably due to
289: the attribution of the Se$3d$ states to the core. As the treatment of Zn$3d$
290: (and also preferably Se$3d$) as valence states is now recognized as essential
291: in pseudopotential calculations for achieving good description of elastic
292: properties of ZnSe, an overall agreement is established between
293: state-of-art LDA calculations.
294: Fig.~\ref{fig:elastic} shows the total energy profiles, as function
295: of uniform lattice scaling (left panels) and in its dependency
296: on the off-center displacement (along any Cartesian direction) of one
297: sublattice (anions or cations) relative to the other, i.e., the
298: zone-center TO mode (right panels).
299: The energy vs. lattice constant curves from all-electron WIEN2k calculations
300: and from {\sc Siesta} are almost identical. Small kinks in the {\sc Siesta}
301: total energy plot at some values of lattice constant are due to changes
302: of the real-space mesh for spatial integration (as the cell volume varies
303: but the mesh density is maintained about the same).
304: The energy profiles of the anion--cation displacement (TO phonon,
305: see right panels of Fig.~\ref{fig:elastic}) shows no substantial anharmonicity
306: (deviations from parabolic behavior)
307: for both ZnSe and BeSe. The phonon frequencies in Table~\ref{tab:elastic}
308: were calculated from the second-order fit to these data.
309: Phonon frequencies in the following sections result from the diagonalization
310: of dynamical matrices, constructed from the forces induced on atoms
311: by small Cartesian displacements. For pure constituents they agree very well
312: with the results of the second-order total energy fit.
313: It is noteworthy that a big difference in $\Gamma$-TO frequencies
314: of ZnSe and BeSe is only in part due to a large differences
315: in the cation masses. The force constants are larger in BeSe,
316: as is well seen from Fig.~\ref{fig:elastic}, indicating higher stiffness
317: of this material, also manifested in its larger bulk modulus.
318:
319: Some our results listed in Table 1 are slightly refined as compared to
320: those reported earlier in Ref.~\onlinecite{Dui03-ZBS},
321: due to an use of a more recent realization of the all-electron method
322: (WIEN2k instead of WIEN97), and a more extended basis set in {\sc Siesta}.
323:
324: In the discussion of phonon properties below we refer to dispersion curves
325: of ZnSe (obtained from neutron scattering and calculated in the rigid
326: ion model, Ref.~\onlinecite{PhL36A-376}) and BeSe
327: (calculated using self-consistent pseudopotential method and
328: linear response approach, Ref.~\onlinecite{BeSe-Tutuncu}).
329: A recent calculation\cite{PSSB229-563} of phonon dispersion for both systems
330: based on a simple model with central and angular forces
331: yields acceptable results for ZnSe but large difference
332: from \emph{ab initio} phonon dispersion curves for BeSe. This could be
333: a manifestation of more covalent character of bonding in BeSe,
334: as discussed below.
335:
336: \section[#4]{Bond lengths in mixed crystals}
337: \label{sec:bonds}
338: %
339: In a sequence of supercell calculations which represent mixed Be--Zn
340: compositions, we begin by unconstrained relaxation of lattice vectors
341: and internal coordinates.
342: Probably the simplest model of a mixed system with predominantly
343: either Be or Zn at cation sites is an ordered 8-atom superstructure with
344: the cubic primitive cell. Such supercells BeZn$_3$Se$_4$ and Be$_3$ZnSe$_4$
345: are shown in the left-hand side of Fig.~\ref{fig:all_scells}.
346: We emphasize that the BeZn$_3$Se$_4$ superstructure contains Be atoms
347: only in configuration which will be further referred to as ``isolated'' one;
348: i.e., each Se neighbor to a Be atom has only Zn ions to saturate
349: its other bonds. In the Be$_3$ZnSe$_4$ superstructure, on the contrary,
350: each Se ion has but a single Zn neighbor, whereas Be--Se bonds
351: form a continuous framework throughout the crystal.
352:
353: \begin{figure}
354: \centerline{\epsfig{figure=fig02.eps,width=8.5cm}}
355: \caption{\label{fig:all_scells}
356: Cubic supercells used in the calculation. The continuous Be--Se chain
357: and isolated Be atoms are marked in the large supercell.
358: The figure was created with the XCrySDen software \cite{xcrysden}.
359: }
360: \end{figure}
361:
362: In order to better understand the changes of the phonon spectra
363: in presumed dependency on the percolation setup, we performed
364: the lattice dynamics simulation in a larger 2$\times$2$\times$2
365: (64 atoms) supercell. It contained 4 beryllium atoms in a continuous
366: --Be--Se-- chain transversing the crystal; moreover,
367: two other Be atoms were situated in ``isolated'' positions, with
368: only their fifth-nearest neighbors being of the Be type. This supercell
369: is also shown in Fig.~\ref{fig:all_scells}.
370: When fully relaxed, it maintains a nearly cubic shape,
371: with a tiny $z$-compression ($c/a$=0.998) due to the anisotropy of chains.
372: The nominal composition of Be is 18.75 at.\%, i.e. just below the theoretical
373: percolation threshold (0.198) on the fcc lattice\cite{Stauffer-book};
374: however, our structure includes percolation on the Be sublattice
375: by construction. The advantage of our choice
376: of supercell is that, despite its relatively moderate size, it allows
377: to analyze local properties of areas with percolation (chained Be--Se bonds)
378: or without percolation (isolated Be--Se bonds only) on equal footing.
379: The results for this supercell will constitute the bulk of our discussion
380: on the lattice dynamics.
381:
382: We note in passing that possible manifestation of percolation effects in mixed
383: semiconductor systems, notably at the above critical concentration and
384: especially in systems with large contrast in stiffness, has been pointed at
385: by Bellaiche \emph{et al.} already some time ago\cite{PRB54-17568}.
386:
387: Finally, four supercells Be$_n$Zn$_{32-n}$Se$_{32}$ ($n$=1,\dots,4)
388: of the same size and cubic shape as the latest one simulate a gradual
389: aggregation of Be ions neighboring \emph{the same} anion site.
390:
391: \begin{figure}[b]
392: \centerline{\epsfig{figure=fig03.eps,width=8.5cm}}
393: \caption{\label{fig:lattice}
394: Top panel: calculated equilibrium lattice constant in Be$_x$Zn$_{1-x}$Se
395: crystals of different concentration, represented by supercells
396: (see text for details).
397: Bottom panel: equilibrium Be--Se and Zn--Se bond lengths.
398: For $x$=0.19, the bond length values show a scattering due to the presence
399: of different local environments -- see details in the text
400: and in Fig.~\ref{fig:bonds}.
401: }
402: \end{figure}
403:
404: Fig.~\ref{fig:lattice} shows the equilibrium lattice constant
405: (derived from the relaxed supercell volume)
406: and the mean cation--anion bond lengths.
407: The mean lattice parameter exhibits a markedly linear dependence
408: on the concentration. At the same time, the Zn--Se and Be--Se bond lengths
409: tend to remain nearly constant throughout the whole concentration range.
410: A similar conclusion follows from the MD simulations for the (Ga,In)As system
411: by Branicio \emph{et al.}\cite{APL82-1057}, who also cite EXAFS data,
412: agreeing well with their results, and for the (Be,Zn)Se alloy
413: -- by Tsai \emph{et al.}\cite{PRB65-235202}, also based on MD simulations.
414: However, the later publication reports a seemingly too large drop
415: in the Be--Se distance when going from Zn$_{1/4}$Be$_{3/4}$Se to pure BeSe.
416: It should be noted that Tsai \emph{et al.} found much smaller difference
417: between the Zn--Se and Be--Se bond lengths than in our case, moreover
418: their lattice parameters for pure constituents considerably deviate
419: from experiment, probably, due to insufficiency of the basis set
420: (and/or pseudopotential) they used.
421:
422: \begin{figure}
423: \centerline{\epsfig{figure=fig04.eps,width=8.5cm}}
424: \caption{\label{fig:bonds}
425: Bond lengths in relaxed (Be,Zn)Se supercells;
426: A discrete set of 128 nearest-neighbor distances is artificially smeared
427: for better visibility.
428: Top panels: bond lengths in the ``model'' supercell, with four Be atoms
429: in continuous Be--Se chains, and two isolated Be impurities. Distances
430: between different types of anions and cations are shown separately.
431: Bottom panels: Be--Se and Zn--Se interatomic distances in four supercells,
432: containing one to four Be atoms, neighboring the same Se atom.
433: }
434: \end{figure}
435:
436: The diversity of local order in a mixed crystal competes with the
437: tendency to maintain the bond lengths, with the effect that the latter
438: get a certain scattering around their mean values. This is illustrated
439: in Fig.~\ref{fig:lattice} for the Be$_6$Zn$_{26}$Se$_{32}$ supercell,
440: and in more detail for this and other ``large'' supercells --
441: in Fig.~\ref{fig:bonds}. A discrete ``spectrum'' of, in total, 128 bond lengths
442: in each fully relaxed supercell is artificially broadened there
443: for better visibility.
444: We start our discussion from the BeZn$_{31}$Se$_{32}$ supercell.
445: Four Se neighbors experience an inward relaxation to a single Be impurity,
446: shortening the Be--Se bonds to about 2.24 {\AA}, i.e., only slightly larger
447: than in pure BeSe. Simultaneously the bonds to their Zn neighbors extend
448: beyond the average interatomic distance ($\sim$2.41 {\AA}) in the rest
449: of the predominantly-ZnSe supercell. By gradually adding more Be neighbors
450: which share the same Se atom, we end up
451: with a symmetric relaxation pattern in Be$_4$Zn$_{26}$Se$_{32}$
452: with a set of 12 contracted bonds (between Be and their outer Se neighbors)
453: and 4 extended bonds (from each Be to the central Se).
454: This is clearly seen in Fig.~\ref{fig:bonds}.
455: The intermediate cases of 2 and 3 Be atoms in the supercell offer
456: an interpolation between the two discussed cases, introducing
457: a diversity in the Be--Se bond lengths depending on
458: the type of second neighbors. The splitting of the Be--Se bonds into shorter
459: and longer ones, around the mean value of $\sim$2.24 {\AA},
460: leads to the diversification of the Zn--Se bonds into, correspondingly,
461: longer and shorter ones, as compared to the dominating bulk-ZnSe
462: central peak at $\sim$2.41 {\AA}.
463: A more complex distribution of bond lengths comes about in a supercell
464: which contains, along with isolated Be impurities,
465: a continuous --Be--Se-- chain (Fig.~\ref{fig:bonds}, top panels).
466: A Be atom substituting Zn favors an inward relaxation of neighboring Se.
467: Large enough around a zero-dimensional defect (isolated atom),
468: such inward relaxation is even stronger around a one-dimensional defect,
469: such as an extended --Be--Se-- chain. The tendency for
470: shortening the Be--Se bond lengths should obviously exist also along the chain,
471: but the actual contraction is limited by the fact that the chain is
472: quasi-infinite and embedded into the ZnSe lattice, which effectively fixes
473: the chain's step. However, since the repeated cation-anion sequence
474: in the zincblende structure is folded, the shortening of its links can be
475: achieved by accommodating the interbond angles, at the price of
476: stretching the bonds between the in-chain Se atoms and their Zn neighbors
477: in the crystal (shaded area in the top right panel of Fig.~\ref{fig:bonds}).
478: The distribution of bond lengths between Zn cations and the off-chain Se
479: anions resembles roughly that in the chain-free Be$_4$Zn$_{28}$Se$_{32}$
480: supercell.
481:
482: \section[#5]{Force constants}
483: \label{sec:forces}
484: %
485: We look now at how the variations in bond lengths map onto changes in
486: the force constants. As already mentioned, the latter become accumulated
487: as we displace the atoms one by one
488: from their equilibrium positions along three Cartesian directions
489: and analyze the forces induced on all atoms of the supercell.
490: Specifically (in the symmetrized form),
491: %
492: \begin{eqnarray}
493: D^{\alpha\beta}_{ij} = - \frac{1}{2} \left[
494: \frac{
495: F^{\alpha}_i(\{{\bf R}\}+d^{\beta}_j) -
496: F^{\alpha}_i(\{{\bf R}\}-d^{\beta}_j)
497: }{2d^{\beta}_j}
498: \right. \nonumber
499: \\
500: +
501: \left.
502: \frac{
503: F^{\beta}_j(\{{\bf R}\}+d^{\alpha}_i) -
504: F^{\beta}_j(\{{\bf R}\}-d^{\alpha}_i)
505: }{2d^{\alpha}_i}
506: \right]
507: \label{eq:force}
508: \end{eqnarray}
509: %
510: where $F^{\alpha}_i$ is the force on atom $\alpha$ in the direction $i$,
511: and $\{{\bf R}\}$+$d^{\beta}_j$ means that of all atoms, only the atom $\beta$
512: is displaced along $j$ from its equilibrium position, by (in our case)
513: $d$=0.03 Bohr. With all elements of $D^{\alpha\beta}_{ij}$ recovered,
514: the solution of dynamical equation
515: %
516: \begin{equation}
517: \sum_{\beta,k} \left[\,
518: \omega^2 \delta_{\alpha\beta}\delta_{ik} -
519: \frac{D^{\alpha\beta}_{ij}}{\sqrt{M_{\alpha}M_{\beta}}}\right]
520: %\left[\sqrt{M_{\beta}}\,
521: A^{\beta}_k
522: %\right]
523: = 0
524: \label{eq:dynam}
525: \end{equation}
526: %
527: yields zone-center phonon frequencies $\omega$ and eigenvectors
528: $A^{\alpha}_i$ of the supercell.
529: Before turning to the analysis of phonons, we discuss briefly the force
530: constants $D^{\alpha\beta}_{ij}$. For each atomic pair $\{\alpha,\beta\}$,
531: they make a 3$\times$3 matrix which can be diagonalized.
532: The diagonal elements are shown in Fig.~\ref{fig:forceNN} in dependency
533: on the corresponding interatomic distance. In accord with previously
534: discussed scattering of the Be--Se and Zn--Se bond lengths, we have now
535: a scattering of force constants.
536: The major elements (of diagonalized 3$\times$3 matrix) show a remarkably
537: pronounced linear variation with the bond length. Two minor elements
538: are much smaller in magnitude and roughly bondlength independent.
539: One can interpret the smallness of minor elements
540: of $D^{\alpha\beta}$ as a measure of the covalency of corresponding bonds.
541: Indeed, in case of a purely ionic bonding only the central motion
542: of atoms would bring about a change in the force; the tangential component
543: makes a non-zero contribution to the force constant only if
544: tangential displacement induces a re-distribution of charge density,
545: meaning that the covalency of the bond is not negligible.
546: The contrast between strongly anisotropic and hence essentially ionic
547: force constants for the Zn--Se bonds and much more covalent Be--Se
548: interaction is clearly seen in Fig.~\ref{fig:forceNN}. It is remarkable
549: that the central Be--Se interaction is, on the average, smaller
550: than the Zn--Se one. Yet, we have seen that the BeSe crystal possesses
551: larger bulk modulus and a higher elasticity parameter in the
552: $\Gamma$-TO vibration. This can only be explained by a considerable
553: contribution of non-central forces (minor elements of the force constants)
554: in the shaping of such elastic parameters. In other words,
555: whereas in ZnSe the stretching of Zn--Se bonds plays a dominant role
556: in the elasticity, in BeSe the bending of bond angles is nearly as important.
557:
558: \begin{figure}
559: \centerline{\epsfig{figure=fig05.eps,width=8.0cm}}
560: \caption{\label{fig:forceNN}
561: Diagonal elements of force constant matrix between nearest neighbors.
562: See text for details.
563: }
564: \end{figure}
565:
566: \begin{figure}
567: \centerline{\epsfig{figure=fig06.eps,width=8.5cm}}
568: \caption{\label{fig:forceNNN}
569: Diagonal elements of force constant matrix between next nearest neighbors.
570: Left panel: (Be,Zn)--(Be,Zn); right panel: Se--Se.
571: }
572: \end{figure}
573:
574: We find very remarkable a pronounced linear dependence of the force
575: constants on the interatomic distance, in apparent indifference
576: to other aspects of the local order. The obtained dependence can be
577: parametrized and used for fast lattice-dynamics simulations in
578: much larger supercells. The results for similarly diagonalized
579: second-nearest neighbors interaction are shown in Fig.~\ref{fig:forceNNN}.
580: The scattering of the Se--Se distances is twice as large as that of
581: cation-cation ones, and shows a clear splitting into two groups,
582: similar to that reported in Ref.~\onlinecite{APL82-1057} for the
583: As--As pair distribution function in (Ga,In)As.
584: This means that Be substitutes Zn almost at their original
585: (non-displaced) positions, i.e., the cation sublattice remains rather rigid,
586: whereas the Se atoms undergo large displacements,
587: depending on their local environment.
588: We see a pronouncedly different behavior of cation-cation
589: and anion-anion force constants: the former have two almost symmetric
590: (positive/negative) diagonal elements, with the third being nearly zero,
591: independently on both type of pair (Be--Be, Be--Zn or Zn--Zn)
592: and on the distance. The anion-anion coupling has an appreciable
593: central-force element, decreasing with the distance, and two much smaller
594: tangential elements.
595: The variation of the major element with distance is (at least in part)
596: related to the fact that on the left hand side of the plot we deal mostly with
597: Se--Be--Se connections, in which the central Se--Se interaction reflects the
598: effect of bending the strongly covalent Be--Se bonds; on the right hand site,
599: the Se--Zn--Se connections offer a much weaker resistance to such bending,
600: due to a higher bond ionicity.
601:
602: \section[#6]{Phonons}
603: \label{sec:phonons}
604: %
605: Since we deal with relatively large ``disordered'' supercells with no internal
606: (short-scale) periodicity, and no clear-cut phonon dispersion
607: exists in such systems, it does not make sense to Fourier-transform
608: the force constants (\ref{eq:force}) prior to solving the dynamical
609: equation (\ref{eq:dynam}). However, we explain below how to extract
610: some $q$-resolved information about lattice vibrations.
611: The primary characteristic for our discussion
612: will be the phonon density of states (PhDOS), resolved when necessary over
613: freely chosen groups $\aleph$ of atoms $\alpha$:
614: \begin{equation}
615: I_{\aleph}(\omega) = \sum_{\alpha \in \aleph}\sum_i
616: \left|A^{\alpha}_i(\omega)\right|^2\,.
617: \label{eq:phDOS}
618: \end{equation}
619: This is a discrete spectrum of $(3N-3)$ lines (for $N$ atoms in the supercell,
620: with acoustic modes removed) of different intensity, which is in the
621: following figures broadened with the halfwidth parameter of 10 cm$^{-1}$,
622: for better visibility.
623: The vibration modes obtained from the solution of Eq.~(\ref{eq:dynam})
624: correspond to the zone-center of the supercell in question, but they
625: reflect different vibration patterns, also those of non-zone-center
626: character, with respect to the underlying zincblende lattice.
627: Let us discuss this for 8-atom supercells, whose calculated PhDOS
628: is shown in Fig.~\ref{fig:4x-PhDOS}.
629:
630: \begin{figure}
631: \centerline{\epsfig{figure=fig07.eps,width=8.5cm}}
632: \caption{\label{fig:4x-PhDOS}
633: Phonon density of states for the BeZn$_3$Se$_4$ and Be$_3$ZnSe$_4$ supercells,
634: calculated for $q$=0 of the supercell and broadened by 10 cm$^{-1}$.
635: }
636: \end{figure}
637:
638: The $q$=0 vibration modes in a supercell include the ``true''
639: zone-center phonon (relative to the basic zincblende cell), as well as
640: zone-boundary phonons backfolded to a smaller BZ of the superstructure.
641: It is relatively easy to distinguish them by their
642: eigenvectors; certain modes are labeled in Fig.~\ref{fig:4x-PhDOS}.
643: Most obviously, the acoustical branch hits the BZ boundary at
644: 87 cm$^{-1}$ (Zn-rich system) or 140 cm$^{-1}$ (Be-rich system).
645: As ranges of frequencies of Be-related and Zn-related optical modes
646: are well separated, due to a large difference in masses,
647: it is straightforward to recognize the BeSe-type $\Gamma$-TO mode
648: at 494 cm$^{-1}$ (lower panel of Fig.~\ref{fig:4x-PhDOS}), the backfold
649: of the optical branch at the zone boundary at 450 cm$^{-1}$,
650: and the backfold of the zone-boundary longitudinal phonon at 560 cm$^{-1}$ --
651: all in good agreement with the linear-response phonon dispersions
652: calculated for BeSe by Tutuncu \emph{et al.}\cite{BeSe-Tutuncu}.
653: The zone-center LO mode is not present, because we do not include
654: macroscopic electric field in crystal in our calculation.
655: In the Zn-rich crystal, the Be-related TO mode (at 487 cm$^{-1}$) behaves
656: like an impurity mode and exhibits no dispersion.
657: Whereas the Zn-related mode is similarly dispersionless (at 260 cm$^{-1}$)
658: in the Be-rich system, in BeZn$_3$Se$_4$ the ZnSe-related TO branch
659: shows dispersion, manifested by backfolding of zone-boundary phonon
660: and hence additional structure at 200--240 cm$^{-1}$, in agreement
661: with experiments and calculations of Refs.~\onlinecite{PhL36A-376,PSSB229-563}.
662:
663: Even as these 8-atom supercells may be too simplistic to imitate short-range
664: order effects in real quasibinary alloy, they exhibit important features
665: which also persist in more sophisticated representative structures:
666: a difference between dispersionless modes due to ``isolated'' impurities
667: and dispersing modes due to continuous connected chains;
668: a more pronounced dispersion in BeSe-related modes than in much softer
669: ZnSe-related ones.
670:
671: \begin{figure}
672: \centerline{\epsfig{figure=fig08.eps,width=8.5cm}}
673: \caption{\label{fig:32x-PhDOS}
674: Phonon density of states for the Be$_6$Zn$_{26}$Se$_{32}$ supercell,
675: resolved over different groups of atoms,
676: calculated for $q$=0 of the supercell and broadened by 10 cm$^{-1}$.
677: The vertical scaling for different groups is arbitrary.
678: }
679: \end{figure}
680:
681: We turn next to the discussion of phonons in a larger Be$_6$Zn$_{26}$Se$_{32}$
682: supercell. There are now more lines in the spectrum, and, as the Brillouin zone
683: (BZ) becomes smaller, the $\Gamma$ modes of the supercell offer
684: a sampling over more non-zone-center modes in a zincblende lattice.
685: This leads in more features and more ``filled'' form of the PhDOS
686: (Fig.~\ref{fig:32x-PhDOS}).
687:
688: The vibration spectra of both pure systems ZnSe and BeSe are shaped
689: by broad acoustical branches and more narrow optical ones.
690: In ZnSe they nearly overlap, resulting in a characteristic two-peaked
691: structure of the PhDOS -- see, e.g., Hennion \emph{et al.}\cite{PhL36A-376}.
692: Our low-frequency (ZnSe-related, up to 300 cm$^{-1}$) part
693: of the supercell PhDOS provides a fair agreement with this known
694: result for pure ZnSe.
695: The spectrum of BeSe has a large separation between its acoustical
696: and optical parts. The former covers the whole
697: region of the phonon spectrum of ZnSe. Therefore, in a mixed crystal,
698: the partial PhDOS of Se atoms which have four Zn neighbors
699: essentially repeats the shape of the Zn PhDOS, whereas those Se atoms
700: with one or two Be neighbors participate in vibrations at
701: 100 -- 200 cm$^{-1}$, i.e., in the ``pseudogap'' region of ZnSe.
702: These modes have mostly acoustical character.
703:
704: We consider now more attentively the high-frequency optical modes
705: with high participation of Be vibrations.
706: The corresponding PhDOS (Fig.~\ref{fig:32x-PhDOS}, top panel) is separated
707: into contributions from (two) ``isolated Be'' atoms and those (four)
708: in the infinite chain. The ``isolated Be'' contribution is, expectedly,
709: a single narrow line, coming about from several nearly degenerate vibration
710: frequencies. The vibrations of Be in the chains are more diversified:
711: the characteristic frequencies can be found below, within and above
712: the ``isolated Be'' peak. The most remarkable is a clear split off
713: at a soft side, which makes a distinguished peak in the vibration spectrum,
714: similar to that experimentally observed in samples with a presumed
715: percolation on the Be sublattice\cite{APL77-519,PRB65-035213}.
716: The previous discussion on the distribution of bondlengths, and
717: the force constants' dependence on the latter, helps us to understand
718: this behavior. The bonds between the chain Be atoms and
719: \emph{out-of-chain} Se are shorter that those involving isolated Be,
720: and corresponding force constants are larger. On the contrary,
721: the Be--Se bonds \emph{along the chain} are less contracted; the longer
722: bond length implies smaller force constant hence lower vibration frequency.
723: Our explanation of the experimentally observed anomalous (split-off)
724: Be mode is, therefore, the following. As the concentration of Be in ZnSe
725: reaches the percolation limit and infinite chains are built, the Be--Se
726: bonds in these chains are forced to become longer that around isolated Be
727: impurities, or in short chain fragments. This immediately softens the
728: vibration modes which involve such extended bonds.
729:
730: In order to extend this interpretation over experimentally measured
731: Raman lines, which are known to scan primarily the zone-center phonons,
732: one needs yet to extract a wavevector-resolved information about calculated
733: vibrations. As the supercell size increased, the effective sampling over the
734: Brillouin zone (BZ) of the basic zincblende lattice became more fine.
735: In particular, one has not only $X$, $K$ and $L$ zone-boundary phonons
736: sampled, as was the case for the 8-atom cells, but also $q$ points halfway
737: to them from the zone center. We mentioned already that the lack
738: of translation symmetry in the mixed BeSe--ZnSe supercell does not allow
739: to Fourier transform the dynamical matrix and arrive at neat phonon
740: dispersion curves. However, one can project the dispersion patterns,
741: corresponding to different modes of vibration, onto the plane wave with
742: different $q$ values and in such way define $q$-resolved contributions
743: to the phonon DOS:
744: %
745: \begin{equation}
746: I_{\aleph}(\omega,{\bf q}) = \sum_{\alpha \in \aleph}\sum_i
747: \left|A^{\alpha}_i(\omega)\exp(i{\bf q}{\bf R}_{\alpha})\right|^2\,.
748: \label{eq:phon_q}
749: \end{equation}
750: %
751: Since eigenvectors correspond to $q$=0 of the supercell and hence are real,
752: the phase of the plane wave has no importance.
753: One can view $I_{\aleph}(\omega,{\bf q})$ as spectral function, which
754: in principle would show how the phonon dispersion bands of pure constituents
755: get overlapped, distorted and smeared in a mixed crystal. Unfortunately,
756: the results obtained for our -- yet relatively small -- 64-atom supercell
757: do not allow to see anything resembling a continuous displacement of intensity
758: maxima with $q$\cite{note_on_q}.
759: Nonetheless, we present in Fig.~\ref{fig:q-PhDOS} the projected
760: PhDOS for three $q$ values along the (001) direction.
761: The ZnSe-type vibrations make clearly visible the displacement
762: of two acoustical bands, known, e.g., from the measured phonon
763: dispersion curves\cite{PhL36A-376}, as the projection wavevector changes
764: from the zone-center via midpoint to the zone boundary.
765:
766: \begin{figure}
767: \centerline{\epsfig{figure=fig09.eps,width=8.5cm}}
768: \caption{\label{fig:q-PhDOS}
769: Phonon spectral function in the Be$_6$Zn$_{26}$Se$_{32}$ supercell,
770: corresponding to three values of $q$ of the basic zincblende lattice,
771: broadened by 10 cm$^{-1}$.
772: }
773: \end{figure}
774:
775: \begin{figure*}[t!]
776: \centerline{\epsfig{figure=fig10.eps,width=18.0cm}}
777: \caption{\label{fig:modes}
778: Vibration patterns of five selected modes with substantial contribution
779: of Be atoms; the relative displacements of the latter are shown by arrows.
780: Only some atoms of the Be$_6$Zn$_{26}$Se$_{32}$ supercell
781: (Fig.~\ref{fig:all_scells}, right panel) are shown.
782: The figure was created with the XCrySDen software \cite{xcrysden}.
783: The first two modes affect exclusively the continuous --Be--Se-- chains;
784: the last three contain also comparable contribution of
785: impurity Be atoms. See text for details.
786: }
787: \end{figure*}
788:
789: For the Be-related modes, because of small amount of Be atoms in the supercell,
790: the dispersion is generally less pronounced, yet noticeable in the (001)
791: direction, which is the general direction of the chains.
792: Being almost non-interacting linear defects,
793: the chains do not give rise to noticeable dispersion in perpendicular
794: directions. The three acoustical modes, removed in the previous
795: PhDOS figures, are retained in Fig.~\ref{fig:q-PhDOS},
796: in order to show how the spectral weight of the $\omega$=0 modes
797: disappears for $q$ values away from zero. A notable difference is
798: the PhDOS of isolated Be, which is nearly identical for $q$=0 and
799: $q$=(001). This is because the supercell contains two isolated Be
800: atoms. They interact only weakly but create a superstructure that
801: effectively halves the BZ in the $z$ direction.
802:
803: The $q$-resolving of the PhDOS singles out the modes which
804: are likely to dominate the Raman spectra. The separation
805: into $q$=0 and non-$q$=0 modes is however not very clear-cut --
806: in part due to physical reasons (mixed crystal, loss of periodicity)
807: and in part due to technical limitations (small size of supercell
808: and hence non-vanishing $q$=0 projection of many modes due to
809: numerical noise). One can expect better contrast in selecting $q$=0
810: modes in a larger supercell, due to better sampling.
811: Nevertheless, it is clear that the separation into ``isolated Be''
812: and ``Be in the chains'' modes occurs not only on the average
813: over the BZ, but is actually even more pronounced in the zone-center modes.
814: Therefore, our explanation of anomalous Raman lines will hold.
815:
816: Most of the modes are of very mixed character, and involve
817: stretching and bending of bonds in a complicated manner. However, we selected
818: several ones, involving the --Be--Se-- chains,
819: whose vibration pattern can be described with relative
820: simplicity. Two of these modes (441 and 452 cm$^{-1}$)
821: lie in the ``anomalous'' Raman line, where the vibrations of isolated
822: Be atoms are negligible. Three other selected modes,
823: at 485, 495 and 502 cm$^{-1}$, involve simultaneous vibration of all
824: Be atoms, in the chains as well as isolated ones.
825: Fig.~\ref{fig:modes} gives a snapshot of these five modes.
826: Displacement patterns in some other modes are shown in Fig.~5
827: of Ref.~\onlinecite{Dui03-ZBS}.
828: The vibrations of predominantly ``impurity Be'' character,
829: leaving the chain Be atoms silent, occur at nearly 490 cm$^{-1}$,
830: in between the third and the fourth of depicted modes.
831:
832: The most general observation is that the ``anomalous'' modes
833: with reduced frequencies involve nearly in-plane displacement
834: of Be with its own equilibrium position and its two Se neighbors.
835: Such motion stretches or shortens the Be--Se
836: bonds but roughly preserves the Be--Se--Be angles.
837: The 441 cm$^{-1}$ mode is a wave of correlated Be displacement along
838: the chain, periodic with the chain step. This is a predominantly $\Gamma$ mode.
839: The 452 cm$^{-1}$ mode has similarly ``longitudinal'' character but
840: the doubled translation length: two Be atoms approach, or depart from,
841: their common Se neighbor symmetrically. Consequently this is not
842: a clear zone-center vibration, however due to the curvature of the chain and
843: a complexity of detailed displacement pattern it provides a ${\bf q}$=0
844: contribution in the analysis according to Eq.~(\ref{eq:phon_q}).
845:
846: The harder modes are related to out-of-plane Be vibrations and hence bending of
847: covalent Be--Se bonds. Somehow simplifying, in the ``anomalous modes''
848: there are primarily central Be--Se forces at work, so the frequency is lower
849: than that involving an isolated Be impurity -- in accord with longer
850: anion-cation distance along the chain, as illustrated
851: by Fig.~\ref{fig:forceNN}. In the latter case, even as interatomic distances
852: remain the same, the off-center contributions due to the bond bending
853: effectively increase the force constants and harden the resulting
854: frequency beyond those for an isolated Be.
855:
856: \section[#7]{Conclusions}
857:
858: We simulated the equilibrium structure and lattice dynamics
859: from first principles in a series of supercells simulating
860: mixed (Be,Zn)Se crystals. We found a nearly linear dependence
861: of equilibrium lattice constant with composition, but a relative
862: independence of (average) Be--Se and Zn--Se distances on concentration.
863: When incorporated in a mixed crystal, the individual interatomic distances
864: develop a certain scattering around their mean values, maintaining
865: however a clear separation between the Be--Se and Zn--Se bond lengths.
866: Specifically, an isolated Be atom may much more efficiently shorten the bonds
867: to an Se neighbor whose other three bonds are Zn-terminated, than to
868: an Se atom shared by other Be neighbors. In particular, a continuous
869: -Be--Se- chain is caracterized by longer link lengths than the average
870: Be--Se distance, for a given concentration.
871: We demonstrated further that this has effect on force constants and
872: vibration frequencies.
873: As follows from our direct calculation of interatomic force constants,
874: the principal values of the latter, taken as diagonal elements
875: of a 3$\times$3 matrix relating two selected atoms, show a linear
876: decrease as a function of cation-anion distance between nearest neighbors.
877: The calculations manifest a primarily ionic character of the Zn--Se
878: interaction and a remarkable level of covalency in the Be--Se bonds.
879: A decrease of force constants with larger bond distance is the
880: microscopic origin of the formation of a softened ``anomalous'' mode,
881: which appears on the low-frequency side of BeSe-related TO mode.
882: Our wavevector analyzis of vibration patterns in the supercell indicates
883: that the ``anomalous'' mode has a strong zone-center contribution,
884: that is consistent with its clear experimental observation in the Raman spectra.
885: We confirmed therefore on the microscopic level the early presumed relation
886: between the onset of Be percolation threshold on the cation sublattice
887: and the appearance of ``anomalous'' mode, including specific vibrations
888: of continuous --Be--Se-- chains transversing the crystal.
889:
890: \section*{Acknowledgments}
891: A.V.P. thanks the Universit\'e Metz for invitation for a research stay there
892: and the hospitality of the Institut de Physique.
893: Useful comments by N.~E.~Christensen, discussions with V.~Vikhnin
894: and travel support from the NATO project CLG 980378 are greatly appreciated.
895:
896: %\bibliography{Global,update,my_publist,zbs-add}
897: %\bibliography{v08}
898: %------------------------------------------------------------------
899: \begin{thebibliography}{37}
900: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
901: \expandafter\ifx\csname bibnamefont\endcsname\relax
902: \def\bibnamefont#1{#1}\fi
903: \expandafter\ifx\csname bibfnamefont\endcsname\relax
904: \def\bibfnamefont#1{#1}\fi
905: \expandafter\ifx\csname citenamefont\endcsname\relax
906: \def\citenamefont#1{#1}\fi
907: \expandafter\ifx\csname url\endcsname\relax
908: \def\url#1{\texttt{#1}}\fi
909: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
910: \providecommand{\bibinfo}[2]{#2}
911: \providecommand{\eprint}[2][]{\url{#2}}
912:
913: \bibitem[{\citenamefont{Chauvet et~al.}(2000)\citenamefont{Chauvet,
914: Tourni{\'e}, and Faurie}}]{PRB61-5332}
915: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Chauvet}},
916: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Tourni{\'e}}},
917: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.-P.}
918: \bibnamefont{Faurie}}, \bibinfo{journal}{Phys.~Rev.~B}
919: \textbf{\bibinfo{volume}{61}}, \bibinfo{pages}{5332} (\bibinfo{year}{2000}).
920:
921: \bibitem[{\citenamefont{Pag{\`e}s et~al.}(2002)\citenamefont{Pag{\`e}s, Ajjoun,
922: Bormann, Chauvet, Tourni{\'e}, and Faurie}}]{PRB65-035213}
923: \bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Pag{\`e}s}},
924: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Ajjoun}},
925: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Bormann}},
926: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Chauvet}},
927: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Tourni{\'e}}},
928: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~P.}
929: \bibnamefont{Faurie}}, \bibinfo{journal}{Phys.~Rev.~B}
930: \textbf{\bibinfo{volume}{65}}, \bibinfo{pages}{035213}
931: (\bibinfo{year}{2002}).
932:
933: \bibitem[{\citenamefont{Pag{\`e}s et~al.}(2004)\citenamefont{Pag{\`e}s, Ajjoun,
934: Tite, Bormann, Tourni{\'e}, and Rustagi}}]{PRB70-155319}
935: \bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Pag{\`e}s}},
936: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Ajjoun}},
937: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Tite}},
938: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Bormann}},
939: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Tourni{\'e}}},
940: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{K.~C.}
941: \bibnamefont{Rustagi}}, \bibinfo{journal}{Phys.~Rev.~B}
942: \textbf{\bibinfo{volume}{70}}, \bibinfo{pages}{155319}
943: (\bibinfo{year}{2004}).
944:
945: \bibitem[{\citenamefont{Postnikov et~al.}(2004)\citenamefont{Postnikov,
946: Pag{\`e}s, Tite, Ajjoun, and Hugel}}]{Dui03-ZBS}
947: \bibinfo{author}{\bibfnamefont{A.~V.} \bibnamefont{Postnikov}},
948: \bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Pag{\`e}s}},
949: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Tite}},
950: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Ajjoun}}, \bibnamefont{and}
951: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Hugel}},
952: \bibinfo{journal}{to be published in Phase Transitions}
953: (\bibinfo{year}{2005}),
954: \bibinfo{note}{{\tt http://xxx.lanl.gov/abs/cond-mat/0404359}}.
955:
956: \bibitem[{\citenamefont{Branicio et~al.}(2003)\citenamefont{Branicio, Kalia,
957: Nakano, Rino, Shimojo, and Vashishta}}]{APL82-1057}
958: \bibinfo{author}{\bibfnamefont{P.~S.} \bibnamefont{Branicio}},
959: \bibinfo{author}{\bibfnamefont{R.~K.} \bibnamefont{Kalia}},
960: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Nakano}},
961: \bibinfo{author}{\bibfnamefont{J.~P.} \bibnamefont{Rino}},
962: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Shimojo}}, \bibnamefont{and}
963: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Vashishta}},
964: \bibinfo{journal}{Appl.~Phys.~Lett.} \textbf{\bibinfo{volume}{82}},
965: \bibinfo{pages}{1057} (\bibinfo{year}{2003}).
966:
967: \bibitem[{\citenamefont{Tsai et~al.}(2002)\citenamefont{Tsai, Peiris, Lee, and
968: Furdyna}}]{PRB65-235202}
969: \bibinfo{author}{\bibfnamefont{M.-H.} \bibnamefont{Tsai}},
970: \bibinfo{author}{\bibfnamefont{F.~C.} \bibnamefont{Peiris}},
971: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Lee}}, \bibnamefont{and}
972: \bibinfo{author}{\bibfnamefont{J.~K.} \bibnamefont{Furdyna}},
973: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{65}},
974: \bibinfo{pages}{235202} (\bibinfo{year}{2002}).
975:
976: \bibitem[{\citenamefont{Soler et~al.}(2002)\citenamefont{Soler, Artacho, Gale,
977: Garc{\'{\i}}a, Junquera, Ordej{\'o}n, and S{\'a}nchez-Portal}}]{JPCM14-2745}
978: \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{Soler}},
979: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Artacho}},
980: \bibinfo{author}{\bibfnamefont{J.~D.} \bibnamefont{Gale}},
981: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Garc{\'{\i}}a}},
982: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Junquera}},
983: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Ordej{\'o}n}},
984: \bibnamefont{and}
985: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{S{\'a}nchez-Portal}},
986: \bibinfo{journal}{J.~Phys.:~Condens.~Matter} \textbf{\bibinfo{volume}{14}},
987: \bibinfo{pages}{2745} (\bibinfo{year}{2002}).
988:
989: \bibitem[{\citenamefont{S{\'a}nchez-Portal
990: et~al.}(1996)\citenamefont{S{\'a}nchez-Portal, Artacho, and
991: Soler}}]{JPCM8-3859}
992: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{S{\'a}nchez-Portal}},
993: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Artacho}}, \bibnamefont{and}
994: \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{Soler}},
995: \bibinfo{journal}{J.~Phys.:~Condens.~Matter} \textbf{\bibinfo{volume}{8}},
996: \bibinfo{pages}{3859} (\bibinfo{year}{1996}).
997:
998: \bibitem[{\citenamefont{Junquera et~al.}(2001)\citenamefont{Junquera, Paz,
999: S{\'a}nchez-Portal, and Artacho}}]{PRB64-235111}
1000: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Junquera}},
1001: \bibinfo{author}{\bibfnamefont{{\'O}.}~\bibnamefont{Paz}},
1002: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{S{\'a}nchez-Portal}},
1003: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Artacho}},
1004: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{64}},
1005: \bibinfo{pages}{235111} (\bibinfo{year}{2001}).
1006:
1007: \bibitem[{\citenamefont{Troullier and Martins}(1991)}]{PRB43-1993}
1008: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Troullier}} \bibnamefont{and}
1009: \bibinfo{author}{\bibfnamefont{J.~L.} \bibnamefont{Martins}},
1010: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{43}},
1011: \bibinfo{pages}{1993} (\bibinfo{year}{1991}).
1012:
1013: \bibitem[{\citenamefont{Singh}(1994)}]{Singh_bookPWPP}
1014: \bibinfo{author}{\bibfnamefont{D.~J.} \bibnamefont{Singh}},
1015: \emph{\bibinfo{title}{Planewaves, pseudopotentials and the {LAPW} method}}
1016: (\bibinfo{publisher}{Kluwer Academic Publishers}, \bibinfo{address}{Boston},
1017: \bibinfo{year}{1994}).
1018:
1019: \bibitem[{\citenamefont{Blaha et~al.}(2001)\citenamefont{Blaha, Schwarz,
1020: Madsen, Kvasnicka, and Luitz}}]{wien2k}
1021: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Blaha}},
1022: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Schwarz}},
1023: \bibinfo{author}{\bibfnamefont{G.~K.~H.} \bibnamefont{Madsen}},
1024: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Kvasnicka}},
1025: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Luitz}},
1026: \emph{\bibinfo{title}{{WIEN2k}, {V}ienna {U}niversity of {T}echnology}}
1027: (\bibinfo{year}{2001}), \bibinfo{note}{improved and updated Unix version of
1028: the original copyrighted WIEN-code, which was published by P.~Blaha,
1029: K.~Schwarz, P.~Sorantin, and S.~B.~Trickey, in Comput. Phys. Commun. 59, 339
1030: (1990)}, \urlprefix\url{http://www.wien2k.at}.
1031:
1032: \bibitem[{\citenamefont{Perdew et~al.}(1996)\citenamefont{Perdew, Burke, and
1033: Ernzerhof}}]{PRL77-3865}
1034: \bibinfo{author}{\bibfnamefont{J.~P.} \bibnamefont{Perdew}},
1035: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Burke}}, \bibnamefont{and}
1036: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Ernzerhof}},
1037: \bibinfo{journal}{Phys.~Rev.~Lett.} \textbf{\bibinfo{volume}{77}},
1038: \bibinfo{pages}{3865} (\bibinfo{year}{1996}).
1039:
1040: \bibitem[{\citenamefont{Lee et~al.}(1995)\citenamefont{Lee, Lee, and
1041: Ihm}}]{PRB52-1459}
1042: \bibinfo{author}{\bibfnamefont{G.-D.} \bibnamefont{Lee}},
1043: \bibinfo{author}{\bibfnamefont{M.~H.} \bibnamefont{Lee}}, \bibnamefont{and}
1044: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Ihm}},
1045: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{52}},
1046: \bibinfo{pages}{1459} (\bibinfo{year}{1995}).
1047:
1048: \bibitem[{\citenamefont{Agrawal et~al.}(1994)\citenamefont{Agrawal, Yadav, and
1049: Agrawal}}]{PRB50-14881}
1050: \bibinfo{author}{\bibfnamefont{B.~K.} \bibnamefont{Agrawal}},
1051: \bibinfo{author}{\bibfnamefont{P.~S.} \bibnamefont{Yadav}}, \bibnamefont{and}
1052: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Agrawal}},
1053: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{50}},
1054: \bibinfo{pages}{14881} (\bibinfo{year}{1994}).
1055:
1056: \bibitem[{\citenamefont{Fleszar and Hanke}(2000)}]{PRB62-2466}
1057: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Fleszar}} \bibnamefont{and}
1058: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Hanke}},
1059: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{62}},
1060: \bibinfo{pages}{2466} (\bibinfo{year}{2000}).
1061:
1062: \bibitem[{\citenamefont{Hellwege and Madelung}(1972)}]{LB17b}
1063: \bibinfo{editor}{\bibfnamefont{K.-H.} \bibnamefont{Hellwege}} \bibnamefont{and}
1064: \bibinfo{editor}{\bibfnamefont{O.}~\bibnamefont{Madelung}}, eds.,
1065: \emph{\bibinfo{title}{Semiconductors. Physics of {II-VI} and
1066: {I-VII} Compounds}}, vol. \bibinfo{volume}{17, Part b} of
1067: \emph{\bibinfo{series}{Landolt-{B\"o}rnstein, New Series, Group {III}}}
1068: (\bibinfo{publisher}{Springer}, \bibinfo{address}{Berlin},
1069: \bibinfo{year}{1972}).
1070:
1071: \bibitem[{\citenamefont{Lee}(1970)}]{JAP41-2988}
1072: \bibinfo{author}{\bibfnamefont{B.~H.} \bibnamefont{Lee}},
1073: \bibinfo{journal}{J.~Appl.~Phys.} \textbf{\bibinfo{volume}{41}},
1074: \bibinfo{pages}{2988} (\bibinfo{year}{1970}).
1075:
1076: \bibitem[{\citenamefont{Pag\`es et~al.}(2000)\citenamefont{Pag\`es, Ajjoun,
1077: Laurenti, Bormann, Chauvet, Tourni\'e, and Faurie}}]{APL77-519}
1078: \bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Pag\`es}},
1079: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Ajjoun}},
1080: \bibinfo{author}{\bibfnamefont{J.~P.} \bibnamefont{Laurenti}},
1081: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Bormann}},
1082: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Chauvet}},
1083: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Tourni\'e}},
1084: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~P.}
1085: \bibnamefont{Faurie}}, \bibinfo{journal}{Applied Physics Letters}
1086: \textbf{\bibinfo{volume}{77}}, \bibinfo{pages}{519} (\bibinfo{year}{2000}).
1087:
1088: \bibitem[{\citenamefont{Szybowicz et~al.}(2003)\citenamefont{Szybowicz,
1089: Kozielski, Firszt, {\L}{\c{e}}gowski, and M{\c{e}}czy{\'n}ska}}]{CrRT38-359}
1090: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Szybowicz}},
1091: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Kozielski}},
1092: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Firszt}},
1093: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{{\L}{\c{e}}gowski}},
1094: \bibnamefont{and}
1095: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{M{\c{e}}czy{\'n}ska}},
1096: \bibinfo{journal}{Crystal Research and Technology}
1097: \textbf{\bibinfo{volume}{38}}, \bibinfo{pages}{359} (\bibinfo{year}{2003}).
1098:
1099: \bibitem[{\citenamefont{Luo et~al.}(1995)\citenamefont{Luo, Ghandehari, Greene,
1100: Ruoff, Trail, and Di{S}alvo}}]{PRB52-7058}
1101: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Luo}},
1102: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Ghandehari}},
1103: \bibinfo{author}{\bibfnamefont{R.~G.} \bibnamefont{Greene}},
1104: \bibinfo{author}{\bibfnamefont{A.~L.} \bibnamefont{Ruoff}},
1105: \bibinfo{author}{\bibfnamefont{S.~S.} \bibnamefont{Trail}}, \bibnamefont{and}
1106: \bibinfo{author}{\bibfnamefont{F.~J.} \bibnamefont{Di{S}alvo}},
1107: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{52}},
1108: \bibinfo{pages}{7058} (\bibinfo{year}{1995}).
1109:
1110: \bibitem[{\citenamefont{Cheong et~al.}(1995)\citenamefont{Cheong, Park, and
1111: Chang}}]{PRB51-10610}
1112: \bibinfo{author}{\bibfnamefont{B.-H.} \bibnamefont{Cheong}},
1113: \bibinfo{author}{\bibfnamefont{C.~H.} \bibnamefont{Park}}, \bibnamefont{and}
1114: \bibinfo{author}{\bibfnamefont{K.~J.} \bibnamefont{Chang}},
1115: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{51}},
1116: \bibinfo{pages}{10610} (\bibinfo{year}{1995}).
1117:
1118: \bibitem[{\citenamefont{Mu{\~n}oz et~al.}(1996)\citenamefont{Mu{\~n}oz,
1119: {Rodr{\'{\i}}guez-Hern{\'a}ndez}, and Mujica}}]{PRB54-11861}
1120: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Mu{\~n}oz}},
1121: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{{Rodr{\'{\i}}guez-Hern{\'a}n%
1122: dez}}}, \bibnamefont{and}
1123: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Mujica}},
1124: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{54}},
1125: \bibinfo{pages}{11861} (\bibinfo{year}{1996}).
1126:
1127: \bibitem[{\citenamefont{{Gonz{\'a}lez-D{\'{\i}}az}
1128: et~al.}(1997)\citenamefont{{Gonz{\'a}lez-D{\'{\i}}az},
1129: {Rodr{\'{\i}}guez-Hern{\'a}ndez}, and Mu{\~n}oz}}]{PRB55-14043}
1130: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Gonz{\'a}lez-D{\'{\i}}az}}},
1131: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{{Rodr{\'{\i}}guez-Hern{\'a}n%
1132: dez}}}, \bibnamefont{and}
1133: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Mu{\~n}oz}},
1134: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{55}},
1135: \bibinfo{pages}{14043} (\bibinfo{year}{1997}).
1136:
1137: \bibitem[{\citenamefont{Muratov et~al.}(2001)\citenamefont{Muratov, Little,
1138: Yang, Cooper, Myers, and Wills}}]{PRB64-035206}
1139: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Muratov}},
1140: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Little}},
1141: \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Yang}},
1142: \bibinfo{author}{\bibfnamefont{B.~R.} \bibnamefont{Cooper}},
1143: \bibinfo{author}{\bibfnamefont{T.~H.} \bibnamefont{Myers}}, \bibnamefont{and}
1144: \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{Wills}},
1145: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{64}},
1146: \bibinfo{pages}{035206} (\bibinfo{year}{2001}).
1147:
1148: \bibitem[{\citenamefont{Fleszar}(2001)}]{PRB64-245204}
1149: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Fleszar}},
1150: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{64}},
1151: \bibinfo{pages}{245204} (\bibinfo{year}{2001}).
1152:
1153: \bibitem[{\citenamefont{Luo et~al.}(2002)\citenamefont{Luo, Ismail-Beigi,
1154: Cohen, and Louie}}]{PRB66-195215}
1155: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Luo}},
1156: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Ismail-Beigi}},
1157: \bibinfo{author}{\bibfnamefont{M.~L.} \bibnamefont{Cohen}}, \bibnamefont{and}
1158: \bibinfo{author}{\bibfnamefont{S.~G.} \bibnamefont{Louie}},
1159: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{66}},
1160: \bibinfo{pages}{195215} (\bibinfo{year}{2002}).
1161:
1162: \bibitem[{\citenamefont{Hennion et~al.}(1971)\citenamefont{Hennion, Moussa,
1163: Pepy, and Kunc}}]{PhL36A-376}
1164: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Hennion}},
1165: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Moussa}},
1166: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Pepy}}, \bibnamefont{and}
1167: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Kunc}},
1168: \bibinfo{journal}{Physics Letters} \textbf{\bibinfo{volume}{36A}},
1169: \bibinfo{pages}{376} (\bibinfo{year}{1971}).
1170:
1171: \bibitem[{\citenamefont{Tutuncu et~al.}(2004)\citenamefont{Tutuncu, Tse, and
1172: Srivastava}}]{BeSe-Tutuncu}
1173: \bibinfo{author}{\bibfnamefont{H.~M.} \bibnamefont{Tutuncu}},
1174: \bibinfo{author}{\bibfnamefont{J.~S.} \bibnamefont{Tse}}, \bibnamefont{and}
1175: \bibinfo{author}{\bibfnamefont{G.~P.} \bibnamefont{Srivastava}}, in
1176: \emph{\bibinfo{booktitle}{Phonons in Condensed Materials}}, edited by
1177: \bibinfo{editor}{\bibfnamefont{S.~P.} \bibnamefont{Sanyal}} \bibnamefont{and}
1178: \bibinfo{editor}{\bibfnamefont{R.~K.} \bibnamefont{Singh}}
1179: (\bibinfo{publisher}{Allied Publishers Pvt. Limited}, \bibinfo{address}{New
1180: Delhi, Mumbai, Kolkata, Lucknow, Chennai, Nagpur, Bangalore, Hyderabad,
1181: Ahmedabad}, \bibinfo{year}{2004}), pp. \bibinfo{pages}{38--42}.
1182:
1183: \bibitem[{\citenamefont{Doyen-Lang et~al.}(2002)\citenamefont{Doyen-Lang,
1184: Pages, Lang, and Hugel}}]{PSSB229-563}
1185: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Doyen-Lang}},
1186: \bibinfo{author}{\bibfnamefont{O.}~\bibnamefont{Pages}},
1187: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Lang}}, \bibnamefont{and}
1188: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Hugel}},
1189: \bibinfo{journal}{Physica Status Solidi (b)} \textbf{\bibinfo{volume}{229}},
1190: \bibinfo{pages}{563} (\bibinfo{year}{2002}).
1191:
1192: \bibitem[{\citenamefont{Kokalj}(1999)}]{xcrysden}
1193: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Kokalj}},
1194: \bibinfo{journal}{Journal of Molecular Graphics and Modelling}
1195: \textbf{\bibinfo{volume}{17}}, \bibinfo{pages}{176} (\bibinfo{year}{1999}),
1196: \bibinfo{note}{code available from {\tt http://www.xcrysden.org}}.
1197:
1198: \bibitem[{\citenamefont{Stauffer and Aharony}(1998, 2001)}]{Stauffer-book}
1199: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Stauffer}} \bibnamefont{and}
1200: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Aharony}},
1201: \emph{\bibinfo{title}{Introduction to Percolation Theory}}
1202: (\bibinfo{publisher}{Taylor \& Francis Ltd.}, \bibinfo{address}{London,
1203: Philadelphia}, \bibinfo{year}{1998, 2001}).
1204:
1205: \bibitem[{\citenamefont{Bellaiche et~al.}(1996)\citenamefont{Bellaiche, Wei,
1206: and Zunger}}]{PRB54-17568}
1207: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Bellaiche}},
1208: \bibinfo{author}{\bibfnamefont{S.-H.} \bibnamefont{Wei}}, \bibnamefont{and}
1209: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Zunger}},
1210: \bibinfo{journal}{Phys.~Rev.~B} \textbf{\bibinfo{volume}{54}},
1211: \bibinfo{pages}{17568} (\bibinfo{year}{1996}).
1212:
1213: \bibitem[{not()}]{note_on_q}
1214: \bibinfo{note}{The reason is that only limited number of vibration frequencies
1215: is represented in the supecell, and only the weight of different modes but
1216: not their frequencies may change in the projection on different $q$. For
1217: instance, there are not enough modes corresponding to the gradual rise of
1218: acoustical branches towards the zone boundary. As a consequence, the phonon
1219: DOS plotted in the ($q,\omega$)-domain shows a step-like rather than smooth
1220: change.}
1221:
1222: \end{thebibliography}
1223:
1224: \end{document}
1225: