cond-mat0603670/agi.tex
1: \documentclass[aps,prl,showpacs,twocolumn,letterpaper]{revtex4}
2: \usepackage{graphicx}
3: \usepackage{subfigure}
4: \usepackage{amsmath}
5: \usepackage{bm}
6: \begin{document}
7: 
8: \title{Dynamical Structure, Bonding, and Thermodynamics of the Superionic 
9: Sublattice in $\alpha$-AgI}
10: 
11: \author{Brandon C. Wood}
12: \author{Nicola Marzari}
13: \affiliation{Department of Materials Science and Engineering, Massachusetts 
14: Institute of Technology, Cambridge, MA 02139}
15: 
16: \begin{abstract}
17: Using extensive first-principles molecular dynamics calculations, we 
18: characterize the superionic phase transition and the lattice and electronic 
19: structures of the archetypal Type-I superionic conductor $\alpha$-AgI. We find 
20: that superionicity is signalled by a phase transition of the silver ions alone.
21: In the superionic phase, the first silver shell surrounding an iodine displays 
22: a distinct dynamical structure that would escape a time-averaged 
23: characterization, and we capture this structure in a set of ordering rules. The
24: electronic structure of the system demonstrates a unique chemical signature of 
25: the weakest-bound silver in the first shell, which in turn is most likely to 
26: diffuse. Upon melting, the silver diffusion decreases, pointing to an unusual 
27: entropic contribution to the stability of the superionic phase.
28: \end{abstract}
29: 
30: \pacs{71.15.Pd,66.30.-h,66.30.Dn}
31: 
32: \maketitle
33: 
34: Key advances in energy research have prompted a surge of interest in superionic
35: materials, as a crucial enabling technology for a variety of nanotechnological 
36: devices, including sensors, switches, batteries, and fuel cells. Of the 
37: superionics, AgI and related silver halides and sulfides have attracted 
38: particular attention because of the unusually high levels of ionic 
39: conductivity they exhibit, and as such are finding increased and varied 
40: technological implementation~\cite{terabe05,lee04,minami96}. At normal pressure,
41: AgI enters its superionic $\alpha$ phase above $T_{c}=420$~K, at which 
42: temperature a phase transition to a body-centered cubic structure is 
43: accompanied by an increase in the silver conductivity of nearly three orders of 
44: magnitude, to a value of 1.31~$\Omega^{-1}\thinspace\text{cm}^{-1}$~\cite{
45: kvist68}. Previous molecular dynamics studies using classical pair potentials 
46: have successfully reproduced experimental characteristics of the $\alpha$ and 
47: $\beta$ phases, as well as the $\alpha\rightarrow\beta$ transition~\cite{
48: vashishta78_parrinello83,tallon88,shimojo91,osullivan91,zimmer00,reviews}, but 
49: these are unable to describe the electronic structure in a dynamic environment, 
50: or to capture the phenomenology of the melting transition. In this regard, 
51: first-principles simulations provide unique and unbiased predictive power.
52: 
53: We perform Car-Parrinello molecular dynamics simulations in the canonical 
54: \textit{NVT} ensemble at temperatures ranging from 200~K to 1250~K for a total
55: of 800 ps.  All simulations were performed with a 54-atom unit cell
56: and $a_0=5.174$~\AA, except the Wannier function calculations, which were 
57: performed in a 32-atom unit cell~\footnote{Simulations were performed in a 
58: plane-wave basis set using the PBE-GGA XC functional, a $4d^{10}5s^{1}$ 
59: ultrasoft silver and $5s^{2}5p^{5}$ norm-conserving iodine pseudopotential, 
60: cutoffs of 22~Ry and 176~Ry for the wavefunctions and charge density, and a 
61: timestep of 20~au.}.
62: 
63: First, we find evidence of a phase transition of the silver ions near the 
64: experimental $T_c$ that is independent of the conformation and dynamics of the 
65: iodine sublattice and signals the transition into the superionic $\alpha$ 
66: phase. The silvers exhibit a sharp decrease in their diffusion behavior upon 
67: cooling below $T_c$, although cubic boundary conditions forbid the iodine 
68: structural transition to the hexagonal wurtzite $\beta$ phase. Results from a 
69: series of simulations in which we immobilized the iodines in a fixed bcc 
70: configuration provide a secondary, stronger indicator of the independence of 
71: the silver transition from any iodine dynamics. Fig.~\ref{fig:arrhenius} 
72: displays the associated silver ion diffusion coefficients $D_{Ag}$ for the 
73: fixed-iodine case and for a fully mobile lattice. Diffusion coefficients are 
74: derived from the mean-square displacement (MSD) via the Einstein relation, 
75: $D=\lim_{t\rightarrow\infty}\frac{1}{6t}\left\langle MSD(t)\right\rangle $. In
76: both cases, the slope of an Arrhenius plot of $D_{Ag}$ shows a characteristic 
77: discontinuity near the experimental transition temperature. Unexpectedly, this 
78: discontinuity is even more pronounced for the fixed-iodine case. Immobilizing 
79: the iodine sublattice does lead to an overall decrease in $D_{Ag}$, suggesting 
80: local lattice fluctuations beneficial for silver mobility are frozen out, but 
81: the system retains its superionic behavior. 
82: 
83: Also of note in Fig.~\ref{fig:arrhenius} is the unusual decrease of the silver 
84: ion diffusion coefficient upon melting at about 850~K, a tendency that has also
85: been observed experimentally~\cite{kvist68,araki99} but is not captured by 
86: classical potentials. The thermodynamic implication gives valuable insight 
87: into the stability of the superionic phase. The decreased mobility of the 
88: silver ions upon melting signals a decrease in the silver entropic 
89: contribution, meaning the iodines crystallize into the superionic state below 
90: $T_m$ to increase the entropy of the silvers and make them more ``liquid''. 
91: Accordingly, the superionic phase acts as an intermediate between a pure solid 
92: and a pure liquid, with a high-entropy liquid sublattice flowing through a 
93: low-energy solid matrix. An entropically driven stabilization of the $\alpha$
94: phase is also consistent with experimental results~\cite{biermann60}, which 
95: determine the entropy difference between superionic and liquid phases to be 
96: relatively tiny.
97: 
98: \begin{figure}
99: 	\centering
100: 	\includegraphics[angle=270]{arrhenius.ps}
101: 	\caption{Arrhenius plot of the diffusion coefficient for the fully 
102: 	mobile system (\textit{A}) and with the iodine sublattice fixed 
103: 	(\textit{B}). The dashed vertical lines indicate the experimental 
104: 	$T_c$ (420~K) and $T_m$ (830~K) at standard pressure.}
105: 	\label{fig:arrhenius}
106: \end{figure}
107: 
108: Although it is not possible to separate the total energy of the system into 
109: respective sublattice contributions, we can get a quantitative picture of the 
110: energetic fluctuations associated with each sublattice by instead integrating 
111: the forces on the ions, which are already decoupled. Since in the canonical 
112: ensemble, fluctuations $\sigma$ in the total energy are related to the 
113: specific heat capacity $C_{V}$ as $\sigma^2=\text{k}_BT^2C_V$, we can derive a 
114: heat capacity-like quantity $C_V^{(i)}$ that is sublattice resolved using the 
115: energetic fluctuations derived from the ionic forces. A plot of this quantity 
116: for the two sublattices, along with the total specific heat capacity for the 
117: entire system, is shown in Fig.~\ref{fig:cv_ion}. Over the full temperature 
118: range of the simulations, we observe the tail end of a divergence in the total 
119: heat capacity at 400~K and 850~K. These temperatures correspond closely to the 
120: superionic transition temperature $T_c$ and the melting temperature $T_m$, 
121: respectively.  Examination of the associated ion-resolved $C_V^{(i)}$ curves 
122: reveals a similar divergence for the silvers at $T_c$ but no detectable trend 
123: for the iodines. As in Fig.~\ref{fig:arrhenius}, we can thus link the 
124: superionic transition to the silver sublattice only. 
125: 
126: \begin{figure}
127: 	\centering
128: 	\includegraphics[angle=270]{c_v.ps}
129: 	\caption{Specific heat capacities for the silver sublattice, the iodine
130: 	sublattice, and the total system.}
131: 	\label{fig:cv_ion}
132: \end{figure}
133: 
134: We note that an order-disorder transition of the silver ions at $T_c$ has been 
135: reported previously~\cite{madden92,seok97_98}. Our results confirm separation 
136: of the occupied tetrahedral sites into the six inequivalent sublattices of 
137: Ref.~\cite{szabo86}, the occupancies of which are plotted as a function of 
138: temperature in Fig.~\ref{fig:tetr_occ}. The simulations reveal an ordering 
139: tendency for the silvers below $T_c$, characterized by a splitting into higher- 
140: and lower-occupancy sublattices. The independent ordering tendency of the silver 
141: sublattice upon cooling is observable despite the inhibition of the 
142: $\alpha\rightarrow\beta$ transition, in agreement with Ref.~\cite{
143: madden92}. However, at these low temperatures, even more extensive statistical
144: sampling is necessary to allow for comparison of the ordered phase with those 
145: proposed in the cited studies.
146: 
147: \begin{figure}
148: 	\centering
149: 	\includegraphics{tetr_occ.ps}
150: 	\caption{Average silver occupancies of the six inequivalent tetrahedral
151: 	bcc interstitial sublattices of Ref.~\cite{szabo86}, normalized
152: 	against the case for which all ions occupy random tetrahedral sites. 
153: 	The inset shows the network of tetrahedral sites surrounding a single 
154: 	iodine, and the dashed line indicates the overall fraction of silvers 
155: 	in tetrahedral sites.}
156: 	\label{fig:tetr_occ}
157: \end{figure}
158: 
159: We have also investigated the most frequented pathways for the silver ions by 
160: tracking their positions with respect to the iodines and averaging the resulting
161: trajectories over the course of the simulations. Fig.~\ref{fig:ag_contours} shows
162: isosurface and slice plots for the silver ion density in the first shell 
163: surrounding an iodine. Our results confirm that the highest-density regions lie
164: near the tetrahedral sites, with silver ion density smeared toward the 
165: octahedral sites. This agrees well with the experimental results of 
166: Ref.~\cite{wuensch77}.
167: 
168: \begin{figure}
169: 	\centering
170: 	\includegraphics[]{ag_contours.ps}
171: 	\caption{(a) The isosurface of silver trajectories at 750 K, enclosing 
172: 	a volume associated with the first silver shell surrounding an iodine. 
173: 	(b) A slice plot of (a), revealing higher density (darker) silver 
174: 	occupancy regions.}
175: 	\label{fig:ag_contours}
176: \end{figure}
177: 
178: 
179: From the ionic trajectories, we can identify a set of rules governing the 
180: instantaneous distribution of silvers. The silver-silver pair distribution 
181: function $g_{Ag-Ag}(r)$ (Fig.~\ref{fig:ag_dist}) illustrates a zero probability 
182: of finding silver ions closer together than $R_{Ag-Ag}=2.4$~\AA, which precludes 
183: simultaneous occupation of nearest-neighbor tetrahedral interstitial sites. 
184: Integration of $g_{Ag-I}(r)$ over the first peak reveals that on average, each 
185: iodine has four nearest neighbors in the first shell, consistent with geometric 
186: expectations for tetrahedral interstitial site occupancy. In addition, 
187: $g_{Ag-I}(r)$ indicates the highest probability distance for silvers surrounding 
188: an iodine is at $R_{Ag-I}=2.6$~\AA, which corresponds to the radial distance to 
189: a tetrahedral interstitial site. However, a time-resolved analysis suggests a 
190: more detailed decomposition of the first peak of $g_{Ag-I}(r)$. We find that 
191: most commonly, three of the four nearest-neighbor silver ions simultaneously 
192: occupy a shell that corresponds to the tetrahedral site distance (the average 
193: value varies from 2.7 to 3.0 depending on temperature, with lower temperatures 
194: favoring higher values). The fourth silver is seen to transition regularly 
195: between this shell and an equivalent shell for a neighboring iodine such that
196: on average, it fills the transition zone between the two, as defined by 
197: $2.8 \leq R_{Ag-I} \lesssim 4.2$~\AA. The transitioning rate is temperature 
198: dependent and disappears rapidly for $T<T_c$. The fourth silver also possesses 
199: an angular distribution distinct from its three inner counterparts, as indicated 
200: in the inset of Fig.~\ref{fig:ag_dist}. The Ag--I--Ag angles for any of the 
201: closest three silvers reveal preferences at $65^{\circ}$ and $105^{\circ}$, and 
202: higher angles are surprisingly uncommon. However, the angles introduced by the 
203: inclusion of the fourth nearest-neighbor shell are comparatively diffuse and 
204: have significant probabilities towards larger values. This suggests that 
205: whereas the closest three silvers are clustered and correlated in their 
206: positions, the fourth silver is relatively unconstrained in its angular 
207: configuration and is affected only marginally by the orientations of the 
208: remaining three. The rare permanent transitions of this unconstrained 
209: fourth silver between shells represent the driving factor for mass diffusion, 
210: an observation that would escape experimental investigations.
211: 
212: \begin{figure}
213: 	\centering
214: 	\includegraphics[angle=270]{ag_dist.ps}
215: 	\caption{Radial pair distribution functions for (\textit{A}) Ag--I, 
216: 	(\textit{B}) I--I, and (\textit{C}) Ag--Ag, along with the integrated 
217: 	Ag--I $n(r)$ curve (\textit{D}), at 750~K. Inset shows the angular 
218: 	distributions of the four silver ions in the first shell surrounding an
219: 	iodine, measured with respect to the remaining silvers in the shell and
220: 	resolved according to the three innermost silvers and the fourth outer
221: 	silver.}
222: 	\label{fig:ag_dist}
223: \end{figure}
224: 
225: The above analysis allows for the definition of a set of ordering rules that 
226: govern the instantaneous distribution of silver ions in the first shell 
227: surrounding an iodine: (1) four silver ions populate the first shell; (2) no 
228: two silver ions occupy neighboring tetrahedral interstitial sites; (3) on 
229: average, three silver ions surround an iodine at a radius of 
230: $R_{Ag-I}=2.6$~\AA, the tetrahedral interstitial distance; (4) a fourth 
231: silver transitions between that shell and a second shell at 
232: $R_{Ag-I}\gtrsim4.2$~\AA, associated with a neighboring iodine, and the 
233: transition rate between the two is temperature dependent and disappears below 
234: $T_c$; (5) the angular positions of the three inner silvers are correlated, 
235: with preferred Ag--I--Ag angles of $65^{\circ}$ and $105^{\circ}$, whereas the 
236: fourth (outer) silver is relatively unconstrained; and (6) silver ions tend to 
237: organize into high- and low-occupancy sublattices at temperatures below $T_c$.
238: 
239: We examined the maximally localized Wannier functions (MLWFs) to obtain a local 
240: picture of bonding in a dynamic environment~\cite{marzari97}. Plotting the 
241: time-averaged radial distribution of the four iodine Wannier function centers 
242: (WFCs) about their parent ion reveals an unexpected bimodal separation into 
243: short-distance, highly localized WFCs and longer-distance, partially delocalized 
244: WFCs (Fig.~\ref{fig:iiwfc_contour}). Further isolating Wannier functions 
245: associated with each of the two peaks and plotting the time- and 
246: statistical-averaged contours for the WFCs around the iodines yields the 
247: isosurfaces shown in the insets of Fig.~\ref{fig:iiwfc_contour}. The 
248: long-distance Wannier function centers (LWFCs) tend to align along the cubic 
249: axes toward the octahedral interstitial sites. Their orientations relate to the 
250: observed smearing of the silver occupancy from the electrostatically preferred 
251: tetrahedral sites towards the octahedral face centers and suggest that these 
252: orbitals correspond to directional interactions between silvers and iodines. On 
253: the other hand, the short-distance iodine Wannier function centers (SWFCs) 
254: exhibit a random angular distribution, as would be expected in a strictly 
255: Coulombic picture. Comparing the areas of the two peaks in 
256: Fig.~\ref{fig:iiwfc_contour} reveals that of the four WFCs surrounding an 
257: iodine, 30\% on average can be classified as SWFCs, a figure which agrees with 
258: the likelihood for a first-shell silver to be found in the mobile transition 
259: zone. This value also correlates well with the the experimentally determined 
260: likelihood of a silver to be located outside the tetrahedral interstitial 
261: site~\cite{wuensch77,nield93}, a quantity confirmed by our findings. 
262: 
263: \begin{figure}
264: 	\centering
265: 	\includegraphics[]{iiwfc_contour.ps}
266: 	\caption{Histogram of distances between iodine WFCs and iodine nuclei. 
267: 	The insets are contours of the iodine WFC distribution about their 
268: 	nuclei for WFCs on either side of the broken line.}
269: 	\label{fig:iiwfc_contour}
270: \end{figure}
271: 
272: A more detailed picture of the correlation between delocalization extent of the
273: iodine MLWFs and the positions of nearby silver atoms is offered in 
274: Fig.~\ref{fig:iwfc}. There is clear evidence of a chemical interaction between 
275: LWFCs and silver atoms within a threshold radius of 
276: $R_{Ag-I}\lesssim3.25$~\AA\ and a solid bond angle of 
277: $\lvert\theta\rvert\lesssim20^{\circ}$. This positional constraint on the 
278: silvers indicates the bonding with LWFCs is directional and not merely 
279: Coulombic. As a quantitative measure of these more complex chemical 
280: interactions, we have also calculated the Born effective charges for the ionic 
281: sublattices, using the electric-enthalpy method of Refs.~\cite{umari02_souza02}. 
282: We obtain values of $Z^*=\pm 1.22$, larger than the integral values expected 
283: for a purely ionic crystal, further supporting the conclusions of our Wannier 
284: function analysis. Fig.~\ref{fig:iwfc} shows no such directional interaction 
285: for SWFCs, indicating that corresponding interactions with silvers can be 
286: attributed to weaker electrostatics. Moreover, examination of the radial 
287: distribution for first-shell silvers closest to SWFCs places them within the 
288: transition zone. 
289: 
290: \begin{figure}
291: 	\centering
292: 	\includegraphics[angle=270]{iwfc.ps}
293: 	\caption{Histograms of interionic bond distance and bond angle for the 
294: 	silvers closest to iodine WFCs, resolved into (\textit{A}) 
295: 	long-distance and (\textit{B}) short-distance Wannier functions. 
296: 	The bond angle $\theta$ is defined as the angle between 
297: 	$\bm{R}_{Ag-I}$ and $\bm{R}_{IWFC-I}$.}
298: 	\label{fig:iwfc}
299: \end{figure}
300: 
301: We conclude that silvers contributing to $D_{Ag}$ are those which are not bound
302: to LWFCs, meaning they are minimally constrained both radially and angularly. 
303: Most commonly, the outermost of the four first-shell silvers fills these 
304: criteria, as a lack of stronger directional interactions with LWFCs increases 
305: the average Ag--I bond distance. This comparatively unconstrained, mobile fourth
306: silver fleetingly occupies the transition zone defined by 
307: $2.8 \leq R_{Ag-I} \lesssim 4.2$~\AA\ until it is captured by a neighboring 
308: iodine, leading to mass diffusion. At higher temperatures, thermal disordering 
309: breaks a greater number of bonds between silvers and LWFCs, promoting more
310: nearby silvers into the transition zone. As such, the overall fraction of 
311: occupied tetrahedral sites decreases (Fig.~\ref{fig:tetr_occ}) and diffusion 
312: is enhanced.
313: 
314: In conclusion, we have shown that the transition to the superionic $\alpha$
315: phase of AgI is signalled by an independent phase transition of the silver 
316: sublattice alone, characterized by a disordering of the silvers and a sharp 
317: increase in their diffusivity. Upon melting, the silver diffusion coefficient
318: decreases, pointing to an unusual entropic contribution to the stabilization of 
319: the superionic phase. We have 
320: also identified diffusion pathways for superionic silver ions above $T_c$, and 
321: a time-resolved analysis of ion trajectories has allowed us to define a set of 
322: ordering rules that govern the instantaneous distribution of silvers in the 
323: first shell surrounding an iodine. Finally, we have found that of the four 
324: first-shell silvers, the closest three are strongly correlated and restricted 
325: in their angular distribution, and that they are involved in anisotropic, 
326: directional bonding to an iodine. The fourth silver is bound only weakly and 
327: is relatively unconstrained, and we have isolated it as the dominant 
328: contributor to mass diffusion.
329: 
330: \begin{acknowledgments}
331: Funding for this work has been provided by the DOE CSGF fellowship and MURI 
332: Grant DAAD 19-03-1-0169. Calculations have been done with the Quantum-ESPRESSO
333: package~\cite{espresso} using computational facilities provided through NSF 
334: grant DMR-0414849. The authors also wish to thank Prof. Bernhard Wuensch 
335: for helpful discussions.
336: \end{acknowledgments}
337: 
338: \begin{thebibliography}{20}
339: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
340: \expandafter\ifx\csname bibnamefont\endcsname\relax
341:   \def\bibnamefont#1{#1}\fi
342: \expandafter\ifx\csname bibfnamefont\endcsname\relax
343:   \def\bibfnamefont#1{#1}\fi
344: \expandafter\ifx\csname citenamefont\endcsname\relax
345:   \def\citenamefont#1{#1}\fi
346: \expandafter\ifx\csname url\endcsname\relax
347:   \def\url#1{\texttt{#1}}\fi
348: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
349: \providecommand{\bibinfo}[2]{#2}
350: \providecommand{\eprint}[2][]{\url{#2}}
351: 
352: \bibitem[{\citenamefont{Terabe et~al.}(2005)\citenamefont{Terabe, Hasegawa,
353:   Nakayama, and Aono}}]{terabe05}
354: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Terabe}}
355:   \bibnamefont{et~al.}, 
356:   \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{433}}, \bibinfo{pages}{47}
357:   (\bibinfo{year}{2005}).
358: 
359: \bibitem[{\citenamefont{Lee et~al.}(2004)\citenamefont{Lee, O'Hayre, Prinz, and
360:   Gur}}]{lee04}
361: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Lee}} \bibnamefont{et~al.}, 
362:   \bibinfo{journal}{Appl. Phys. Lett.} \textbf{\bibinfo{volume}{85}},
363:   \bibinfo{pages}{3552} (\bibinfo{year}{2004}).
364: 
365: \bibitem[{\citenamefont{Minami et~al.}(1996)\citenamefont{Minami, Saito, and
366:   Tatsumisago}}]{minami96}
367: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Minami}} \bibnamefont{et~al.}, 
368:   \bibinfo{journal}{Solid State Ion.} \textbf{\bibinfo{volume}{86-8}},
369:   \bibinfo{pages}{415} (\bibinfo{year}{1996}).
370: 
371: \bibitem[{\citenamefont{Kvist and Josefson}(1968)}]{kvist68}
372: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Kvist}} \bibnamefont{and}
373:   \bibinfo{author}{\bibfnamefont{A.~M.} \bibnamefont{Josefson}},
374:   \bibinfo{journal}{Z. Naturforsch.} \textbf{\bibinfo{volume}{23a}},
375:   \bibinfo{pages}{625} (\bibinfo{year}{1968}).
376: 
377: \bibitem[{\citenamefont{Vashishta and Rahman}(1978)}]{vashishta78_parrinello83}
378: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Vashishta}} \bibnamefont{and}
379:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Rahman}},
380:   \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{40}},
381:   \bibinfo{pages}{1337} (\bibinfo{year}{1978});
382:   \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Parrinello}}
383:   \bibnamefont{et~al.}, 
384:   \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{50}},
385:   \bibinfo{pages}{1073} (\bibinfo{year}{1983}).
386: 
387: \bibitem[{\citenamefont{Shimojo and Kobayashi}(1991)}]{shimojo91}
388: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Shimojo}} \bibnamefont{and}
389:   \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Kobayashi}},
390:   \bibinfo{journal}{J. Phys. Soc. Jpn.} \textbf{\bibinfo{volume}{60}},
391:   \bibinfo{pages}{3725} (\bibinfo{year}{1991}).
392: 
393: \bibitem[{\citenamefont{O'Sullivan et~al.}(1991)\citenamefont{O'Sullivan,
394:   Chiarotti, and Madden}}]{osullivan91}
395: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{O'Sullivan}}
396:   \bibnamefont{et~al.}, \bibinfo{journal}{Phys. Rev. B}
397:   \textbf{\bibinfo{volume}{43}}, \bibinfo{pages}{13536} (\bibinfo{year}{1991}).
398: 
399: \bibitem[{\citenamefont{Zimmer et~al.}(2000)\citenamefont{Zimmer, Ballone,
400:   Maier, and Parrinello}}]{zimmer00}
401: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Zimmer}}
402:   \bibnamefont{et~al.}, 
403:   \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{112}},
404:   \bibinfo{pages}{6416} (\bibinfo{year}{2000}).
405: 
406: \bibitem[{\citenamefont{Tallon}(1988)}]{tallon88}
407: \bibinfo{author}{\bibfnamefont{J.~L.} \bibnamefont{Tallon}},
408:   \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{38}},
409:   \bibinfo{pages}{9069} (\bibinfo{year}{1988}).
410: 
411: \bibitem[{\citenamefont{Hull}(2004)}]{reviews}
412: For reviews on the subject, see
413: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Hull}}, \bibinfo{journal}{Rep.
414:   Prog. Phys.} \textbf{\bibinfo{volume}{67}}, \bibinfo{pages}{1233}
415:   (\bibinfo{year}{2004});
416:   \bibinfo{author}{\bibfnamefont{D.~A.} \bibnamefont{Keen}}, 
417:   \bibinfo{journal}{J. Phys. Cond. Matt.} \textbf{\bibinfo{volume}{14}}, 
418:   \bibinfo{pages}{R819} (\bibinfo{year}{2002}).
419: 
420: \bibitem[{\citenamefont{Araki et~al.}(1999)\citenamefont{Araki, Koishi, and
421:   Tamaki}}]{araki99}
422: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Araki}}
423:   \bibnamefont{et~al.}, 
424:   \bibinfo{journal}{J. Phys. Soc. Jpn.} \textbf{\bibinfo{volume}{68}},
425:   \bibinfo{pages}{134} (\bibinfo{year}{1999}).
426: 
427: \bibitem[{\citenamefont{Biermann et~al.}(1960)\citenamefont{Biermann and
428:   Jost}}]{biermann60}
429: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Biermann}} \bibnamefont{and}
430:   \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Jost}},
431:   \bibinfo{journal}{Z. Phys. Chem. (Frankfurt am Main)} \textbf{\bibinfo{volume}
432:   {17}}, \bibinfo{pages}{139} (\bibinfo{year}{1960}).
433: 
434: \bibitem[{\citenamefont{Madden et~al.}(1992)\citenamefont{Madden, O'Sullivan,
435:   and Chiarotti}}]{madden92}
436: \bibinfo{author}{\bibfnamefont{P.~A.} \bibnamefont{Madden}}
437:   \bibnamefont{et~al.}, 
438:   \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{45}},
439:   \bibinfo{pages}{10206} (\bibinfo{year}{1992}).
440: 
441: \bibitem[{\citenamefont{Seok and Oxtoby}(1998)}]{seok97_98}
442: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Seok}} \bibnamefont{and}
443:   \bibinfo{author}{\bibfnamefont{D.~W.} \bibnamefont{Oxtoby}},
444:   \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{56}},
445:   \bibinfo{pages}{11485} (\bibinfo{year}{1997});
446:   \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{58}},
447:   \bibinfo{pages}{5146} (\bibinfo{year}{1998}).
448: 
449: \bibitem[{\citenamefont{Szab\'{o}}(1986)}]{szabo86}
450: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Szab\'{o}}},
451:   \bibinfo{journal}{J. Phys. C} \textbf{\bibinfo{volume}{19}},
452:   \bibinfo{pages}{3775} (\bibinfo{year}{1986}).
453: 
454: \bibitem[{\citenamefont{Cava et~al.}(1977)\citenamefont{Cava, Reidinger, and
455:   Wuensch}}]{wuensch77}
456: \bibinfo{author}{\bibfnamefont{R.~J.} \bibnamefont{Cava}}
457:   \bibnamefont{et~al.}, 
458:   \bibinfo{journal}{Solid State Commun.}
459:   \textbf{\bibinfo{volume}{24}}, \bibinfo{pages}{411} (\bibinfo{year}{1977}).
460: 
461: \bibitem[{\citenamefont{Marzari and Vanderbilt}(1997)}]{marzari97}
462: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Marzari}} \bibnamefont{and}
463:   \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Vanderbilt}},
464:   \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{56}},
465:   \bibinfo{pages}{12847} (\bibinfo{year}{1997}).
466: 
467: \bibitem[{\citenamefont{Nield et~al.}(1993)\citenamefont{Nield, Keen, Hayes,
468:   and McGreevy}}]{nield93}
469: \bibinfo{author}{\bibfnamefont{V.~M.} \bibnamefont{Nield}}
470:   \bibnamefont{et~al.}, 
471:   \bibinfo{journal}{Solid State Ion.} \textbf{\bibinfo{volume}{66}},
472:   \bibinfo{pages}{247} (\bibinfo{year}{1993}).
473: 
474: \bibitem[{\citenamefont{Umari and Pasquarello}(2002)}]{umari02_souza02}
475: \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{Umari}} \bibnamefont{and}
476:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Pasquarello}},
477:   \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{89}},
478:   \bibinfo{pages}{157602} (\bibinfo{year}{2002});
479:   \bibinfo{author}{\bibfnamefont{I.}~\bibnamefont{Souza}}
480:   \bibnamefont{et~al.}, 
481:   \bibinfo{journal}{Ibid.}, \bibinfo{pages}{117602}.
482: 
483: \bibitem[{\citenamefont{Baroni et~al.}()\citenamefont{Baroni, Corso,
484:   de~Gironcoli, Giannozzi, Cavazzoni, Ballabio, Scandolo, Chiarotti, Focher,
485:   Pasquarello et~al.}}]{espresso}
486: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Baroni}}
487:   \bibnamefont{et~al.}, \urlprefix\url{http://www.quantum-espresso.org/}.
488: 
489: \end{thebibliography}
490: \end{document}
491: