1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: \documentclass[12pt,floats]{article}
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4:
5: \setlength{\textheight}{23.0cm}
6: \setlength{\textwidth}{14.0cm}
7: \setlength{\topmargin}{-0.40cm}
8: \usepackage[]{umlaut,latexsym}
9: \usepackage[xdvi]{graphicx}
10: \usepackage{natbib}
11: \usepackage{amssymb}
12:
13: %%%%%%%%%%%%%%%%
14: \begin{document}
15: %%%%%%%%%%%%%%%%
16:
17: \parskip 2mm
18: \renewcommand{\refname}{\normalsize \bf \em References}
19:
20: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
21: \title{\bf ISOLATED MODES AND PERCOLATION IN LATTICE DYNAMICS OF (Be,Zn)Se }
22: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
23: %
24: \author{
25: A.\ V.\ POSTNIKOV,$^a$\footnote{Corresponding author:
26: Tel.: +49 541 969 2377,
27: fax: +49 541 969 2351,
28: email: \mbox{apostnik@uos.de}}~
29: O.\ PAGES,$^b$
30: T.\ TITE,$^b$
31: M.\ AJJOUN,$^b$ \\
32: and J.\ HUGEL$^b$
33: \\*[0.2cm]
34: $^a${\small \it Institute of Metal Physics,
35: S. Kowalewskoj 18, Yekaterinburg 620219, Russia,} \\
36: {\small \it and Universit\"at Osnabr\"uck -- Fachbereich Physik,
37: D-49069 Osnabr\"uck, Germany} \\
38: $^b${\small \it IPEM, Universit\'e de Metz,
39: 1 Bd Arago, F-57078 Metz cedex 3, France}
40: } % author
41: \date{\small \it (Received \today)}
42: \maketitle
43: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
44:
45: %%%%%%%%%%%%%%%%
46: \begin{abstract}
47: %%%%%%%%%%%%%%%%
48: %
49: A mixed II-VI semiconductor Zn$_{1-x}$Be$_x$Se possesses non-trivial vibration
50: properties, because its two constituent compounds, ZnSe and BeSe,
51: show very different degree of covalency and hence high elastic contrast.
52: An anomalous Be-Se vibration line has been observed mostly at
53: intermediate Be content in the Raman spectra of thin (Zn,Be)Se films.
54: In order to explain microscopic origins and the detailed composition
55: of these lines, a first-principles calculation of vibration frequencies
56: in a mixed crystal has been done, with frozen-phonon technique and
57: supercell setup within the density functional theory, by the
58: {\sc Siesta} method, which uses norm-conserving pseudopotentials
59: and strictly localized numerical basis functions. The calculations
60: confirmed an earlier assumption that the anomalous Be-Se line appears
61: due to the formation of continuous chains of a more rigid Be-rich
62: pseudo-continuous phase formed within the more soft Zn-rich host region
63: on crossing the Be-Se bond percolation threshold ($x\sim{0.19}$).
64: Different local deformation in percolated and non-percolated regions affect
65: interatomic elastic interactions and split corresponding vibration lines.
66: Besides confirming the percolation model qualitatively, the calculation
67: provides details about vibration patterns in different phonon modes.
68: %
69: %%%%%%%%%%%%%%
70: \end{abstract}
71: %%%%%%%%%%%%%%
72:
73: \noindent {\it Keywords:}\/ Semiconductors, Density functional theory,
74: Phonons, Percolation, Isolated modes
75:
76: %%%%%%%%%%%%%%%%%%%%%%%%%%
77: \section*{1. INTRODUCTION}
78: %%%%%%%%%%%%%%%%%%%%%%%%%%
79:
80: Chalcogenides of Be, characterized by a strong covalent character of their
81: chemical bonding, are outstanding among the II-VI semiconductors
82: which are otherwise largely ionic systems. BeSe makes solid solution
83: with ZnSe in the whole composition region. Beyond providing the possibility
84: to tune the fundamental band gap and dielectric function in broad limits
85: \citep{PRB59-10071}, the coexistence of
86: strongly ionic and strongly covalent bonds in the same crystal makes
87: its elastic and vibrational properties interesting.
88: The frequencies of optical modes
89: in pure ZnSe and BeSe are far separated.
90: This remains basically true in the alloy \citep{APL77-519}. The remarkable
91: point is that an additional Be-Se mode has been detected, mostly in the
92: intermediate concentration range, 20 to 80 at.\% Be.
93: This effect has been tentatively attributed by \citet{PRB65-035213}
94: to the onset of percolation on the Be sites, i.e., creation of
95: infinite --(Be--Se)-- chains in the predominantly ZnSe crystal.
96:
97: In the attempts to understand and describe the lattice dynamics
98: of mixed crystals in terms of simple models, an obvious difficulty is
99: that interaction parameters (force constants) between two given atoms depend
100: on their local environment in a not straightforward way. The elastic
101: couplings are affected by interatomic distances which may differ in an alloy;
102: moreover the combinatorial aspect (how many neighbours of one or another
103: type) may play an unknown role. Therefore it could be difficult to agree
104: on a simple and yet realistic model.
105: First-principles calculations of lattice vibrations, either by linear response
106: \citep{PRB48-3156,PRB50-13035,RMP73-515} or frozen-phonon
107: \citep{PRL69-2799} schemes, work well for perfect crystals
108: \citep{PRB60-15511}
109: or ordered mixed phases \citep{CMS13-232,RMP73-515}
110: but face problems in case of substitutional disorder.
111: In this latter case, the treatment of representative
112: supercells remains a virtually unique alternative, with the obvious
113: shortcoming that the selection of such supercells is limited,
114: and the calculation technically demanding.
115: In the present contribution, we calculate electronic structure,
116: force constants and phonon frequencies in the (Zn,Be)Se mixed system
117: using the first-principles method, and calculation code,
118: {\sc Siesta} \citep{JPCM14-2745}. The calculations
119: were practically done for Zn$_3$BeSe$_4$, ZnBe$_3$Se$_4$ and
120: Zn$_{26}$Be$_6$Se$_{32}$ supercells, the latter simulating both percolating
121: and ``isolated'' Be ions in the lattice.
122:
123: The additional vibrational modes, whose appearance in the calculation
124: could be traced back to the onset of percolated Be chains, closely resemble
125: those observed earlier in experimental spectra by \citet{PSSB229-25}
126: and outlined briefly in Sec.~2. This supports earlier assumptions about
127: the role of percolation in the formation of split-off vibration lines.
128: The results of calculation, which were organized as described in Sec.~3,
129: allow to study the composition of each vibration mode separately, and
130: to explain the underlying microscopics, in terms of variation of
131: interatomic distances. The discussion of results is given in Sec.~4.
132:
133: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
134: \section*{2. EXPERIMENTAL SITUATION}
135: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
136:
137: \begin{figure}[b!]
138: \begin{center}
139: \includegraphics[angle=0,width=11cm]{zbs-fig1.eps}
140: \end{center}
141: \caption {\small
142: Raman spectra of Zn$_{1-x}$Be$_x$Se films grown on the GaAs substrate,
143: measured in the backscattering geometry along the (110) edge axis.
144: The ``regular'' and ``anomalous''
145: (labeled TO$^{\,\mbox{\footnotesize a}}_{\,\mbox{\footnotesize Be-Se}}$)
146: lines coexist in the concentration range 20 to 80 at.\% of Be.
147: See \citet{PRB65-035213} for more detailed discussion of the experiment.
148: \label{fig:exp}
149: }
150: \end{figure}
151:
152: The vibrations spectra of (Zn,Be)Se alloys have been detected
153: by Raman measurements in the backscattering geometry from thin films grown
154: by molecular beam epitaxy on (001) GaAs.
155: The experimental setup, obtained spectra and their discussion have been
156: in detail described earlier \citep{APL77-519,PRB65-035213,JAP91-9187}.
157: A typical selection of spectra,
158: in dependence on concentrations, is shown in Fig.~\ref{fig:exp}.
159: For the sake of our subsequent discussion, we emphasize the region
160: of transverse optical (TO) Be-Se modes, in the range 400--500 cm$^{-1}$.
161: The Zn--Se mode, around 200 cm$^{-1}$, is useless because of strong
162: Fano interference with a disorder-activated acoustical
163: continuum \citep{PRB65-035213}. Incidentally the large frequency gap
164: between the Be--Se and Zn--Se nodes is basically attributed to the large
165: difference in the reduced mass of two bonds, in the ratio 1:4.
166: The increase in the Be concentration shifts the BeSe-like
167: TO line upwards. From $\sim$20\% of Be on, an additional (``anomalous'') peak
168: is formed at $\sim$70 cm$^{-1}$ on the low-frequency side from the main one.
169: Two peaks coexist through the concentration range $\sim$20 to $\sim$80\%
170: of Be, both displacing towards higher frequencies and approximately
171: maintaining their separation. The intensity of the anomalous peak grows
172: at the expense of the normal one, until finally (on the Be-rich side)
173: the anomalous peak remains the only one, and ends up as the narrow
174: TO line in the limit of pure BeSe.
175:
176: The discussion in previous works \citep{PRB65-035213,PSSB229-25}
177: emphasized that such behaviour cannot be accounted for by segregation
178: of crystal into BeSe-rich and ZnSe-rich fractions.
179: The proposed explanation of the anomaly was the following.
180: Above the Be--Se bond percolation threshold ($\sim$19 at.\% Be),
181: a pseudo-infinite wall-to-wall chain of Be--Se bonds is formed spontaneously
182: throughout the alloy. Based on simple local strain considerations,
183: the bond length within the chain was presumed to be larger than the bond
184: length of the Be--Se bonds isolated in the ZnSe-like host matrix, with
185: concomitant impact upon the phonon frequencies. However, microscopic
186: justification of this picture was so far lacking. Our present aim is
187: to construct a representative structure model,
188: compare the vibration properties following from it with experimental
189: observations, and in case of agreement to analyze microscopical
190: origins of the observed anomaly.
191:
192: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
193: \section*{3. CALCULATION METHOD AND SETUP}
194: \label{sec:calc}
195: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
196:
197: First-principles calculations have been performed within the general
198: scope of the density functional theory
199: \citep[see, e.g.,][for a review]{RMP71-1253},
200: using the calculation method {\sc Siesta} \citep{JPCM14-2745}
201: with its compact basis of strictly confined atom-centered functions
202: \citep{JPCM8-3859,PRB64-235111}. Double-$\zeta$ basis functions with
203: polarization orbitals were used for all valence states -- Be $2s$ and $2p$,
204: along with $3d$, $4s$ and $4p$ for Zn and Se. Note that we attributed
205: the semicore Se$3d$ states to the valence band.
206: The norm-conserving pseudopotentials
207: have been generated after \citet{PRB43-1993}. The results discussed below
208: have been obtained using the local density approximation for the
209: exchange-correlation.
210:
211: \begin{figure}[t!] % Fig. 2
212: \begin{center}
213: \includegraphics[angle=0,width=13cm]{zbs-fig2.eps}
214: \end{center}
215: \caption {\small
216: BeZn$_3$Se$_4$, Be$_3$ZnSe$_4$ and Be$_6$Zn$_{26}$Se$_{32}$ supercells
217: used in the calculation. More covalent Be--Se bonds are shown
218: by thick lines. The large supercell contains a continuous
219: (--Be--Se--) chain and two isolated Be atoms (indicated by arrows),
220: who have only Zn atoms as next-nearest neighbours.
221: \label{fig:scells}
222: }
223: \end{figure}
224:
225: The calculations have been done for three different cubic supercells,
226: shown in Fig.~\ref{fig:scells}. The smaller ones
227: included four formula units in the zincblende structure.
228: BeZn$_3$Se$_4$, with the nominal Be concentration (25 at.\%) already
229: beyond the percolation threshold, contained nevertheless only
230: isolated (and artificially ordered) Be impurities, whose second-nearest
231: neighbours are Zn atoms only. The ``symmetric'' case of 75 at.\% Be
232: (Be$_3$ZnSe$_4$) contains, on the contrary, isolated Zn atoms and
233: fully connected Be-Se network.
234: Moreover, a larger supercell Be$_6$Zn$_{26}$Se$_{32}$
235: simulated the onset of percolation on the Be sublattice, i.e.,
236: the formation of continuous (--Be--Se--) chains, along with the presence
237: of isolated Be substitutions. The Be concentration in this case is just
238: above the percolation threshold on the zincblende lattice.
239: In all cases, the calculated forces and strain were used for
240: relaxing the lattice vectors and internal coordinates
241: by the conjugated gradient algorithm, without any symmetry constraints
242: imposed. This allowed us to discuss the variation of the lattice parameters
243: and interatomic distances in their dependence on concentration and
244: local environment.
245: The vibration frequencies have been calculated in the frozen phonon approach,
246: sequentially introducing finite displacements of atoms in the supercell
247: by 0.03 Bohr along three Cartesian directions from their equilibrium
248: positions, and analyzing the forces induced on all atoms.
249: In a single cell, this technique gives access to the zone-center TO
250: phonon only, but in a larger supercell the analysis of phonon eigenvectors
251: allows to make some conclusions about the phonon dispersion.
252:
253: \begin{table}[t!]
254: \caption{\small
255: Equilibrium lattice constant $a$, bulk modulus $B$, and
256: frequency of the $\Gamma$-TO phonon $\omega$ for pure ZnSe and BeSe
257: from the present calculations (WIEN97 and {\sc Siesta}), previous
258: calculations, and experiment.
259: }
260: \begin{center}
261: \begin{tabular*}{\textwidth}{l@{\extracolsep\fill}ccccccc}
262: \hline
263: \rule[-2mm]{0mm}{7mm}
264: & \multicolumn{3}{c}{ZnSe} && \multicolumn{3}{c}{BeSe} \\
265: \cline{2-4} \cline{6-8}
266: Method & $a$ ({\AA}) & $B$ (Kbar) & $\omega$ (cm$^{-1}$) &&
267: \rule[-2mm]{0mm}{7mm}
268: $a$ ({\AA}) & $B$ (Kbar) & $\omega$ (cm$^{-1}$) \\
269: \hline \hline
270: WIEN97 & 5.571 & 727 & 162 && 5.087 & 831 & 456 \\
271: {\sc Siesta} & 5.590 & 778 & 200 && 5.114 & 965 & 429 \\
272: exp. & 5.668 & 624--647 & 207$^a$ && 5.137$^b$ & 920 & 501$^c$ \\
273: other calc. & & 750$^d$ & 216$^e$ & \\
274: \hline
275: \end{tabular*} \\*[4mm]
276: \end{center}
277: \footnotesize
278: $^a$\citet{CrRT38-359};
279: $^b$\citet{PRB52-7058};
280: $^c$\citet{APL77-519};
281: $^d$\citet{PRB32-7988};
282: $^e$\citet{PhL36A-376}
283: \normalsize
284: \end{table}
285:
286: As benchmarks for testing the quality of the pseudopotentials
287: used and of the basis sets, we calculated the equilibrium lattice parameter
288: and the $\Gamma$-TO frequency in pure ZnSe and BeSe. The calculation results
289: are shown in Table I along with the experimental data and the results
290: of another, all-electron calculation, which has been performed
291: using an accurate all-electron full-potential linearized augmented
292: plane wave method WIEN97 \citep{wien97}.
293: The difference between the {\sc Siesta} results and the experiment
294: reflects the shortcomings of the local density approximation, and moreover
295: technical limitations of the pseudopotential generation and the choice
296: of basis functions. A similarly more faithful estimation of some
297: results with {\sc Siesta}, as compared to WIEN97, in view of numerical
298: superiority of the latter method, is probably due to some error cancellation.
299: Anyway, Table I gives an idea of the level of accuracy of our present
300: calculation approach. It is obvious already from comparison the
301: bulk moduli that ZnSe is much softer than BeSe. The same trend holds
302: for the force constants, and mainly accounts for the difference in
303: the $\Gamma$-TO phonon frequencies. The difference in Be and Zn masses
304: further enlarges this difference.
305:
306: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
307: \section*{4. RESULTS AND DISCUSSION}
308: \label{sec:results}
309: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
310:
311: \begin{figure}[t!] % Fig. 3
312: \begin{center}
313: \includegraphics[angle=0,width=11cm]{zbs-fig3.eps}
314: \end{center}
315: \caption {\small
316: Mean lattice constant (top panel) and nearest-neighbor bond lengths (bottom
317: panel), according to calculations on two pure end systems and
318: three mixed-composition supercells. In the Be$_6$Zn$_{26}$Se$_{32}$ supercell,
319: the distribution of Be--Se and Zn--Se bond lengths is shown.
320: \label{fig:bonds1}
321: }
322: \end{figure}
323:
324: Fig.~\ref{fig:bonds1} shows the calculated equilibrium lattice constant
325: (derived from the volume of relaxed supercells) and nearest-neighbour
326: distances in pure materials and three mixed-composition supercells.
327: One can see that the (Zn,Be)Se lattice parameter quite accurately
328: follows the Vegard's law. At the same time, the Be--Se and Zn--Se bond
329: lengths, which differ by about 9\% in the parent compounds,
330: tend to remain diverse, and almost independent
331: on concentration. It means that the accommodation of short Be--Se and
332: long Zn--Se bonds in the lattice of a mixed crystal may lead to
333: local strains and frustrations. In fact, in the large supercell
334: one finds a distribution of bond lengths of one and another type,
335: shown as continuous curves in the bottom panel of Fig.~\ref{fig:bonds1}.%
336: \footnote{%
337: Obviously the number of nearest-neighbour bonds in the 64-atom supercell
338: is finite (=128); an artificial smearing of the bond distribution
339: function is introduced for better visibility.}
340: A further analysis of these different bond lengths shows that isolated
341: (non-chain) Be atoms can easily induce a local and ``isotropic''
342: lattice contraction, binding all their four Se neighbours at a distance
343: of 2.23 {\AA}, i.e. at nearly the nominal bond length of pure BeSe.
344: About the same, or even shorter (2.21 to 2.23 {\AA}), bond length
345: gets stabilized between Be atoms in the chains and their off-chain
346: Se neighbours. One finds finally that the Be--Se bonds
347: \emph{along the chains} are the most difficult to shorten,
348: and exhibit a variation of lengths 2.24 -- 2.28 {\AA}. This can be
349: understood because the chain is infinite, and its period is
350: essentially fixed by the mean lattice constant of a crystal with
351: a predominant Zn concentration. The only possibility to shorten the chain
352: links is at the price of increasing the Se--Be--Se angles, making
353: a zigzag chain more straight. This increases average distances
354: between in-chain Se atoms and their Zn neighbours beyond the comfortable
355: ZnSe bond length.
356:
357: \begin{figure}[t!] % Fig. 4
358: \begin{center}
359: \includegraphics[angle=0,width=11.5cm]{zbs-fig4.eps}
360: \end{center}
361: \caption {\small
362: Phonon densities of states for different categories of atoms
363: represented in the Be$_6$Zn$_{26}$Se$_{32}$ supercell.
364: Top panel: Be and Zn contributions, bottom panel: Se contributions.
365: Three acoustic modes at $\omega$=0 are removed, and the broadening of discrete
366: vibration lines is introduced.
367: \label{fig:PhDOS}
368: }
369: \end{figure}
370:
371: The discussed differences in interatomic distances, and their stretching
372: beyond, or below, the natural bond lengths has an effect on the calculated
373: force constants. A diagonalisation of the dynamical matrix yields frequencies
374: and eigenvectors of all 192 zone-center vibration modes of the supercell,
375: including three acoustic modes (the closeness of whose frequencies to zero
376: is a good test for the numerical accuracy of the calculated forces).
377: Fig.~\ref{fig:PhDOS} shows phonon densities of states, constructed as
378: %
379: %%%%%%%%%%%%%%%%%
380: \begin{eqnarray*}
381: %%%%%%%%%%%%%%%%%
382: %
383: n_{\{\alpha\}}(\omega) = \sum_{\{\alpha\}} \sum_i
384: \frac{\Delta}{\pi}\,
385: \frac{|\mathbf{A}_{\alpha}(\omega_i)|^2}{\Delta^2 + (\omega-\omega_i)^2}\,,
386: %
387: %%%%%%%%%%%%%%%%
388: \end{eqnarray*}
389: %%%%%%%%%%%%%%%%
390: where $\omega_i$ are (discrete) vibration frequencies in the supercell
391: (leaving aside the three zero-frequency acoustic modes),
392: $\mathbf{A}_{\alpha}$ are elements of the eigenvector of the corresponding
393: mode, and the summation is done over $\{\alpha\}$, a selection of atoms
394: in a certain structure category. A broadening parameter $\Delta$=10 cm$^{-1}$
395: was introduced for better visibility.
396:
397: The top panel of Fig.~\ref{fig:PhDOS} shows the contributions of Zn and Be
398: atoms, which are well separated in the frequency domain. The lower part
399: (up to $\sim$300 cm$^{-1}$) resembles the phonon density of states
400: in ZnSe \citep[see, e.g.,][]{PhL36A-376}. The high-frequency part
401: represents a Be--Se vibration; its fine structure reveals the differences
402: in the force constants which couple different Be--Se pairs.
403: The vibration modes involving the \emph{isolated} Be are densely
404: situated around 490 cm$^{-1}$. On the contrary, the vibration bands involving
405: Be atoms in the chains split into a softer and a harder components.
406: A comparison with the phonon DOS projected on Se atoms (Fig.~\ref{fig:PhDOS},
407: bottom panel) shows that the softer one affects almost exclusively
408: the Se atoms in the chains (i.e those with two Be and two Zn neighbours),
409: whereas the harder one involves also the peripheric Se atoms,
410: which build just one bond to Be in the chains. This separation immediately
411: follows from the distinction between extended Be--Se bonds along the chain,
412: with correspondingly weakened force constants, and ``natural''
413: Be--Se bond lengths to off-the-chain Se atoms, or around isolated Be
414: impurities. This observation clearly indicates that the splitting off
415: of a softer ``additional'' line from the main TO Be--Se peak in the
416: measured spectra (Fig.~\ref{fig:exp}) is due to the formation of
417: continuous (--Be--Se--) chains, where the Be--Se bond length is essentially
418: fixed by the predominantly ZnSe bulk, therefore extended over the
419: ``natural'' Be--Se distance, and results in weakened force constant.
420: The drift of the ``normal'' and ``additional'' lines with concentration,
421: seen in Fig.~\ref{fig:exp}, is not directly addressed by our simulation
422: on a single ``percolated'' supercell, but it can be easily understood
423: from the general trend of the average lattice constant. A gradual
424: shortening of Be--Se distances with the increase in Be concentration
425: hardens \emph{all} vibration modes.
426: The concept of percolation on the Be sublattice also makes clear why
427: the ``additional'' and not the ``normal'' mode transforms into the single
428: TO mode in the limit of high Be concentration (see Fig.~\ref{fig:exp}):
429: the ``additional'' mode was that involving chained (--Be--Se--) structures,
430: which at high Be concentration become omnipresent.
431:
432: \begin{figure}[b!]
433: \begin{center}
434: \includegraphics[angle=0,width=11.5cm]{zbs-fig5.eps}
435: \end{center}
436: \caption {\small
437: Snapshots of four selected vibration modes involving Be atoms
438: in the chains. The displacements of Se atoms are negligible
439: in the chosen scale. For each mode, a simplified schema of
440: its characteristic displacement is shown below.
441: \label{fig:modes}
442: }
443: \end{figure}
444:
445: In order to get a better idea of vibration patterns within ``normal'' and
446: ``additional'' lines, we shown in Fig.~\ref{fig:modes} the snapshots of
447: four vibration modes. Whereas the most of vibration modes involve many
448: atoms in the supercell in a complicated way, these four allow more or less
449: easy understanding of their underlying displacement. Such simplified
450: schemas are shown below each mode, neglecting the bending of the zigzag
451: chain. The most general observation is that the movement of Be atom
452: in the common plane with two Se neighbours occurs at lower frequency
453: than out-of-plane vibrations. This is however a quite simplifying
454: statement; one should consider that the off-the-chain Se atoms also
455: participate in these movements, up to different extent in harder
456: (``normal'') and sifter (``additional'') modes, as has been mentioned above.
457:
458: We leave for a subsequent more detailed discussion elsewhere
459: the analysis of vibrations in smaller supercells, which we used here only
460: as a reference to the variation of lattice parameters. Moreover, the analysis
461: of vibration modes with respect to their wave vector is important and
462: will be reported separately, as only $\mathbf{q}$=0 vibrations contribute
463: to the Raman spectra.
464:
465: %%%%%%%%%%%%%%%%%%%%%
466: \section*{5. SUMMARY}
467: %%%%%%%%%%%%%%%%%%%%%
468:
469: By performing a straightforward calculation of vibration frequencies
470: in a large supercell simulating (Zn,Be)Se mixed system,
471: we were able to explain the onset of ``anomalous'' vibration mode
472: in the frequency range close to the $\Gamma$-TO line of BeSe. This mode
473: appears due to a softening of corresponding force constants in expanded
474: Be--Se bonds, that occurs as continuous Be--Se chains start to traverse
475: the Be$_x$Zn$_{1-x}$Se crystal on reaching the Be percolation limit
476: of $x$=19 at.\%. We found remarkable variations of the Be--Se bond length in
477: a (Zn,Be)Se mixed crystal with the Zn predominance. The Be--Se
478: distances shorten most efficiently between Be ions in the percolating
479: chains and their neighbouring Se ions \emph{not} belonging to the
480: chains. On the contrary, the Be--Se distances within the chains
481: are essentially fixed by the ZnSe bulk, i.e. they are over-stretched,
482: as compared to relaxed BeSe crystal. The Be--Se bond length for
483: a Be ion being an isolated substitutional impurity in the ZnSe matrix
484: is intermediate between two previously mentioned cases. As a consequence,
485: one finds distinct Be--Se force constants for these three cases,
486: and hence the splittings of corresponding vibration modes.
487:
488: \begin{thebibliography}{21}
489: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
490: \expandafter\ifx\csname url\endcsname\relax
491: \def\url#1{{\tt #1}}\fi
492: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
493:
494: \bibitem[{Baroni {\em et~al.\/}(2001)Baroni, {de Gironcoli}, Dal~Corso and
495: Giannozzi}]{RMP73-515}
496: Baroni, S., S.~{de Gironcoli}, A.~Dal~Corso and P.~Giannozzi (2001).
497: \newblock Phonons and related crystal properties from density-functional
498: perturbation theory.
499: \newblock {\em Rev.~Mod.~Phys.\/} {\bf 73}(2), 515.
500:
501: \bibitem[{Blaha {\em et~al.\/}(1997)Blaha, Schwarz and Luitz}]{wien97}
502: Blaha, P., K.~Schwarz and J.~Luitz (1997).
503: \newblock {WIEN97}, {V}ienna {U}niversity of {T}echnology.
504: \newblock Improved and updated Unix version of the original copyrighted
505: WIEN-code, which was published by P.~Blaha, K.~Schwarz, P.~Sorantin, and
506: S.~B.~Trickey, in \emph{Comput.~Phys.~Commun.} {\bf 59}, 339 (1990).
507:
508: \bibitem[{Cohen(1985)}]{PRB32-7988}
509: Cohen, M.~L. (1985).
510: \newblock Calculation of bulk moduli of diamond and zinc-blende solids.
511: \newblock {\em Phys.~Rev.~B\/} {\bf 32}(12), 7988.
512:
513: \bibitem[{Gonze {\em et~al.\/}(1994)Gonze, Charlier, Allan and
514: Teter}]{PRB50-13035}
515: Gonze, X., J.-C. Charlier, D.~C. Allan and M.~P. Teter (1994).
516: \newblock Interatomic force constants from first principles: The case of
517: {$\alpha$}-quartz.
518: \newblock {\em Phys.~Rev.~B\/} {\bf 50}(17), 13035.
519:
520: \bibitem[{Hennion {\em et~al.\/}(1971)Hennion, Moussa, Pepy and
521: Kunc}]{PhL36A-376}
522: Hennion, B., F.~Moussa, G.~Pepy and K.~Kunc (1971).
523: \newblock Normal modes of vibrations in {ZnSe}.
524: \newblock {\em Phys.~Lett.\/} {\bf 36A}(5), 376.
525:
526: \bibitem[{Junquera {\em et~al.\/}(2001)Junquera, Paz, S{\'a}nchez-Portal and
527: Artacho}]{PRB64-235111}
528: Junquera, J., {\'O}.~Paz, D.~S{\'a}nchez-Portal and E.~Artacho (2001).
529: \newblock Numerical atomic orbitals for linear scaling.
530: \newblock {\em Phys.~Rev.~B\/} {\bf 64}(23), 235111.
531:
532: \bibitem[{Kohn(1999)}]{RMP71-1253}
533: Kohn, W. (1999).
534: \newblock Nobel lecture: Electronic structure of matter -- wave functions and
535: density functionals.
536: \newblock {\em Rev.~Mod.~Phys.\/} {\bf 71}(5), 1253.
537:
538: \bibitem[{Luo {\em et~al.\/}(1995)Luo, Ghandehari, Greene, Ruoff {\em
539: et~al.\/}}]{PRB52-7058}
540: Luo, H., K.~Ghandehari, R.~G. Greene, A.~L. Ruoff {\em et~al.\/} (1995).
541: \newblock Phase transformation of {BeSe} and {BeTe} to the {NiAs} structure at
542: high pressure.
543: \newblock {\em Phys.~Rev.~B\/} {\bf 52}(10), 7058.
544:
545: \bibitem[{Pag{\`e}s {\em et~al.\/}(2001)Pag{\`e}s, Ajjoun, Bormann, Chauvet
546: {\em et~al.\/}}]{PRB65-035213}
547: Pag{\`e}s, O., M.~Ajjoun, D.~Bormann, C.~Chauvet {\em et~al.\/} (2001).
548: \newblock Vibrational evidence for a percolative behavior in
549: {Zn$_{1-x}$Be$_x$Se}.
550: \newblock {\em Phys.~Rev.~B\/} {\bf 65}(3), 035213.
551:
552: \bibitem[{Pag{\`e}s {\em et~al.\/}(2002{\natexlab{a}})Pag{\`e}s, Ajjoun,
553: Bormann, Chauvet {\em et~al.\/}}]{JAP91-9187}
554: Pag{\`e}s, O., M.~Ajjoun, D.~Bormann, C.~Chauvet {\em et~al.\/}
555: (2002{\natexlab{a}}).
556: \newblock {R}aman study of {Zn$_{1-x}$Be$_x$Se/GaAs} systems with low {Be}
557: content ({$x\leq0.20$}).
558: \newblock {\em J.~Appl.~Phys.\/} {\bf 91}(11), 9187.
559:
560: \bibitem[{Pag\`es {\em et~al.\/}(2000)Pag\`es, Ajjoun, Laurenti, Bormann {\em
561: et~al.\/}}]{APL77-519}
562: Pag\`es, O., M.~Ajjoun, J.~P. Laurenti, D.~Bormann {\em et~al.\/} (2000).
563: \newblock {R}aman study of {Zn$_x$Be$_{1-x}$Se} alloy (100) epitaxial layers.
564: \newblock {\em Appl.~Phys.~Lett.\/} {\bf 77}(4), 519.
565:
566: \bibitem[{Pag{\`e}s {\em et~al.\/}(2002{\natexlab{b}})Pag{\`e}s, Ajjoun,
567: Laurenti, Bormann {\em et~al.\/}}]{PSSB229-25}
568: Pag{\`e}s, O., M.~Ajjoun, J.~P. Laurenti, D.~Bormann {\em et~al.\/}
569: (2002{\natexlab{b}}).
570: \newblock Vibrational evidence for percolative behavior in {ZnBeSe}.
571: \newblock {\em Phys.~Stat.~Sol. (b)\/} {\bf 229}(1), 25.
572:
573: \bibitem[{Parlinski and Kawazoe(1999)}]{PRB60-15511}
574: Parlinski, K. and Y.~Kawazoe (1999).
575: \newblock {\em Ab initio\/} study of phonons in hexagonal {GaN}.
576: \newblock {\em Phys.~Rev.~B\/} {\bf 60}(23), 15511.
577:
578: \bibitem[{Pavone {\em et~al.\/}(1993)Pavone, Karch, Sch{\"u}tt, Strauch {\em
579: et~al.\/}}]{PRB48-3156}
580: Pavone, P., K.~Karch, O.~Sch{\"u}tt, D.~Strauch {\em et~al.\/} (1993).
581: \newblock Ab initio lattice dynamics of diamond.
582: \newblock {\em Phys.~Rev.~B\/} {\bf 48}(5), 3156.
583:
584: \bibitem[{S{\'a}nchez-Portal {\em et~al.\/}(1996)S{\'a}nchez-Portal, Artacho
585: and Soler}]{JPCM8-3859}
586: S{\'a}nchez-Portal, D., E.~Artacho and J.~M. Soler (1996).
587: \newblock Analysis of atomic orbital basis sets from the projection of
588: plane-wave results.
589: \newblock {\em J.~Phys.:~Condens.~Matter\/} {\bf 8}(26), 3859.
590:
591: \bibitem[{Soler {\em et~al.\/}(2002)Soler, Artacho, Gale, Garc{\'{\i}}a {\em
592: et~al.\/}}]{JPCM14-2745}
593: Soler, J.~M., E.~Artacho, J.~D. Gale, A.~Garc{\'{\i}}a {\em et~al.\/} (2002).
594: \newblock The {SIESTA} method for \emph{ab initio} order-{$N$} materials
595: simulation.
596: \newblock {\em J.~Phys.:~Condens.~Matter\/} {\bf 14}(11), 2745.
597: \newblock {\sc Siesta} site: \url{http://www.uam.es/siesta}.
598:
599: \bibitem[{Sternik {\em et~al.\/}(1999)Sternik, Jochym and
600: Parlinski}]{CMS13-232}
601: Sternik, M., P.~T. Jochym and K.~Parlinski (1999).
602: \newblock Lattice dynamics of {Ga$_{1-x}$Al$_x$As} studied by ab initio
603: calculations.
604: \newblock {\em Comput. Mater. Sci.\/} {\bf 13}(4), 232.
605:
606: \bibitem[{Szybowicz {\em et~al.\/}(2003)Szybowicz, Kozielski, Firszt,
607: {\L}{\c{e}}gowski {\em et~al.\/}}]{CrRT38-359}
608: Szybowicz, M., M.~Kozielski, F.~Firszt, S.~{\L}{\c{e}}gowski {\em et~al.\/}
609: (2003).
610: \newblock {R}aman scattering study of {ZnBeSe} semiconducting mixed crystals.
611: \newblock {\em Cryst.~Res.~Technol.\/} {\bf 38}(3--5), 359.
612:
613: \bibitem[{Troullier and Martins(1991)}]{PRB43-1993}
614: Troullier, N. and J.~L. Martins (1991).
615: \newblock Efficient pseudopotentials for plane-wave calculations.
616: \newblock {\em Phys.~Rev.~B\/} {\bf 43}(3), 1993.
617:
618: \bibitem[{Wei and Chou(1992)}]{PRL69-2799}
619: Wei, S. and M.~Y. Chou (1992).
620: \newblock Ab initio calculation of force constants and full phonon dispersions.
621: \newblock {\em Phys.~Rev.~Lett.\/} {\bf 69}(19), 2799.
622:
623: \bibitem[{Wilmers {\em et~al.\/}(1999)Wilmers, Wethkamp, Esser, Cobet {\em
624: et~al.\/}}]{PRB59-10071}
625: Wilmers, K., T.~Wethkamp, N.~Esser, C.~Cobet {\em et~al.\/} (1999).
626: \newblock Ellipsometric studies of {Be$_x$Zn$_{1-x}$Se} between 3 eV and 25 eV.
627: \newblock {\em Phys.~Rev.~B\/} {\bf 59}(15), 10071.
628:
629: \end{thebibliography}
630:
631: %%%%%%%%%%%%%%
632: \end{document}
633: %%%%%%%%%%%%%%
634: