1: %\documentclass[aps,pra,showpacs]{revtex4}
2: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: %\documentclass[showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
4: \usepackage{graphicx}% Include figure files
5: \usepackage{dcolumn}% Align table columns on decimal point
6: \usepackage{bm}% bold math
7: \setlength{\parindent}{0pt}
8: \setlength{\parskip}{1ex plus 0.5ex minus 0.2ex}
9:
10:
11: \begin{document}
12:
13: \title{Dynamical response function of a compressed lithium monolayer.}
14:
15: \author{A. Rodriguez-Prieto$^{1,2}$}
16: %\email{alvaro@wm.lc.ehu.es}
17: \author{V. M. Silkin$^1$}
18: \author{A. Bergara$^{1,2}$}
19: \author{P. M. Echenique$^{1,3}$}
20:
21: \affiliation{1 Donostia International Physics Center (DIPC), Paseo de Manuel Lardizabal, 20018, Donostia, Basque Country, Spain\\ 2 Materia Kondentsatuaren Fisika Saila, Zientzia eta Teknologia Fakultatea, Euskal Herriko
22: Unibertsitatea, 644 Postakutxatila, 48080 Bilbo, Basque Country, Spain\\ 3 Materialen Fisika Saila, Kimika Fakultatea,\\ Euskal Herriko Unibertsitatea, 20080, Donostia, Basque Country, Spain}
23:
24: \begin{abstract}
25: Since recent both theoretical and experimental results have proved that the simple behaviour light alkaline metals
26: present at equilibrium breaks when high pressures are applied, they have become an important object of study in
27: Condensed Matter Physics. On the other hand, development of new techniques in the atomic
28: manipulation allows the growth of atomic monolayers (ML's), therefore rising the interest to analyze low dimensional
29: systems under different conditions. In particular, new \textit{ab initio} calculations performed for a lithium ML
30: show that its electronic properties experience important modifications under pressure, which could lead to
31: significant modifications in its dynamical response function.
32: In this article we perform \textit{ab initio} calculations of the dynamical response function of a
33: lithium ML analyzing its evolution with increasing applied
34: pressure. We show that besides the well known intraband and interband plasmons,
35: rising electronic density induces characteristic features of acoustic plasmons
36: related to the presence of two types of carriers at the Fermi level.
37: \end{abstract}
38:
39: \maketitle
40:
41: \section{Introduction}
42: Light alkali metals have lately become an important object of study due to the
43: complexity induced by pressure in their behaviour, in contrast to the simple picture they present under normal
44: conditions of pressure and
45: temperature. At equilibrium, alkalies are frequently considered to be simple metals due to
46: their crystallization in high symmetric
47: structures, monovalency and high conductivity. At the same time the interaction between valence electrons and ionic cores,
48: or pseudopotential, has been argued to be weak in such systems.
49: As a consequence, the nearly free electron (NFE) model accurately describes their electronic properties in absence
50: of applied
51: pressure \cite{wigner} as it can be easily checked by their almost spherical Fermi surfaces. However, when high pressures are
52: applied, the magnitude of the pseudopotential rises, the core electrons substantially overlap and the otherwise valid NFE model breaks.
53: Neaton \textit{et al} \cite{NA} theoretically
54: analyzed compressed bulk lithium, predicting that high pressures could induce phase transitions to less
55: symmetric, lower coordinated structures, associated to electronic localizations. These theoretical predictions were
56: experimentally confirmed by Hanfland \textit{et al} \cite{hanfland}, who found that lithium undergoes several phase transitions from a
57: simple, high symmetric, $bcc$ structure at equilibrium to a complex $cI16$ with 16 atoms per unit cell at around 40 GPa.
58: It is important to mention
59: that pressure induced transitions from simple to more complex structures are not singular to lithium but have also
60: been observed in heavier alkalines \cite{neaton}. Another important feature which shows up the complexity induced by pressure in the
61: behaviour of light alkalies is that despite the superconducting transition for lithium at equilibrium has not been
62: found yet, $T_{c}<100\,\mu$K \cite{liambient}, when compressed to around 30 GPa, $T_{c}$ rises up to 15 K \cite{shimizu,deemyad},
63: % he quitado str, struzkhin et al
64: becoming the highest
65: transition temperature between simple elements \cite{ashcroft_nature}. Therefore, the characterization and understanding of the physical properties of compressed light elements becomes a priority. On the other hand, the development of new techniques for the atomic manipulation allows the growth of atomic monolayers
66: (ML's) on
67: inert substratum, semiconductors or noble gases. These new possibilities rise the interest to analyze physical properties
68: of low-dimensional systems under different conditions, as it could be the case of a lithium ML under pressure. In addition,
69: extending our conclusions to the bulk will also give another perspective to understand the physical origin under the
70: experimentally observed features in compressed lithium. Previous \textit{ab initio} analysis of the structural and electronic properties of a lithium ML \cite{aitor,prb} reveal important
71: modifications in both its band structure and Fermi \textit{line}, which will also lead to significant modifications on its
72: dynamical response function, up to now just studied at equilibrium \cite{aitor_plasmon}. In this article we perform
73: calculations of the dynamical response function of a lithium ML at different pressures, analyzing its evolution with increasing
74: electronic density. Besides the common intraband and interband plasmons, at a certain value of the lateral
75: pressure applied to the ML, we observe characteristic features of acoustic plasmons related to the presence, at this
76: pressure range, of two types of carriers at the Fermi level. In Section \ref{sec:T} we describe the theoretical and
77: computational background of this work. Results and Discussion are presented at Section \ref{sec:R}. Unless otherwise
78: stated we use atomic units throughout, i.e., $e^{2}=\hbar=m_{e}=1$.
79:
80: \section{Theoretical and computational background.}
81: \label{sec:T}
82: We consider a perturbing positive charge located far from the ML, $z_{0}>>d$ ($d$ being the thickness of the ML). Then, the
83: differential cross section for a process in which the electron is scattered with energy $\omega$ and momentum transfer
84: $|\bf{q}|$ is proportional to Im $g(\bf{q},\omega)$, where $g(\bf{q};\omega)$ is defined as the density response
85: function of the ML \cite{persson,tsuei},
86: %quito a liebsch
87: \begin{eqnarray}
88: \label{eq:g}
89: &g({\bf{q}};\omega)=-\frac{2\pi}{|{\bf{q}}|}\int\,dz\int\,dz'\times&
90: \nonumber\\
91: &\times \chi_{{\bf{G}}=0,{\bf{G}'}=0}({\bf{q}},z,z';\omega)e^{|{\bf{q}}|(z+z')},&
92: \end{eqnarray}
93: where $\bf{q}$ belongs to the two-dimensional Brillouin zone (2DBZ) and ${\bf{G}}$ and ${\bf{G}'}$ are reciprocal two-dimensional lattice vectors. In order to obtain $g({\bf{q}};\omega)$ we strictly follow the method presented in Refs. \cite{aitor_plasmon,slava_plasmon}. Thus, our calculations are performed within the framework of time dependent density functional theory (TDDFT), where the random phase approximation (RPA) \cite{pines} has been applied. We use a local density
94: approximation (LDA) in order to calculate the total energy and one-particle eigenvalues and eigenvectors
95: of the ground state, which were evaluated with the use of Troullier-Martins Li pseudopotentials \cite{tm}.
96: The description of the ML is implemented by a supercell which contains 20 layers of
97: vacuum between ML's in order to minimize interaction between them. Effects of compression were simulated by reducing the
98: lattice parameter. For the evaluation of $\chi^{0}_{{\bf{G}},\bf{G}'}$, previous step in the calculation of $\chi_{{\bf{G}},\bf{G}'}$, we implement up to $\sim$80 bands. In addition,
99: we use $\sim7700$ $\bf{k}$-points for the sampling of the 2DBZ.
100:
101: \section{Results and Discussion}
102: \label{sec:R}
103: According to previous \textit{ab initio} calculations, the lithium ML adopts an hexagonal ($hex$) structure at
104: equilibrium, which corresponds to $r_{s}=3.02$\footnote{$r_{s}$ is the two-dimensional linear density parameter, defined by the relation $A/N=\pi(r_{s}a_{0})^{2}$ where $N$ is the number of nuclei in a ML with area $A$ and $a_{0}$ is the Bohr radius.}. However, with applying pressure, a phase transition to the more opened
105: square ($sq$) structure is produced at $r_{s}=2.25$. In Fig. \ref{fig:bandas_monoat} we plot the band structure and density of states (DOS) of a lithium ML corresponding to
106: the $hex$ structure at equilibrium and the $sq$ structure with $r_{s}=2.15$, close to the structural phase transition.
107: \begin{figure}[h!]
108: %\includegraphics[height=\linewidth,bb=85 33 573 714,angle=-90]{LiMLHEX_eq.eps}\\
109: %\vspace*{1cm}
110: %\includegraphics[height=\linewidth,bb=85 33 573 714,angle=-90]{LiMLSQ_rs2.1.eps}
111: \includegraphics[width=\linewidth,clip=true]{Band_hex_eq.eps}\\
112: %\vspace*{1cm}
113: \includegraphics[width=\linewidth,clip=true]{Band_sq_rs2.1.eps}
114: \caption{Band structure and density of states (DOS) for (a) a hexagonal lithium ML at equilibrium ($r_s=3.02$ a.u.) and
115: (b) a square structure at $r_{s}=2.15$ a.u. (close to the structural transition). The Fermi energy is represented by the
116: dotted line. The bandwidth of the first band decreases with pressure and induces the occupation of the second band.}
117: \label{fig:bandas_monoat}
118: \end{figure}
119: We can observe a quasi free electron-like behaviour at equilibrium, where only one band with a parabolic-type dispersion
120: of $s$-character is occupied. In addition, the DOS for occupied states is almost
121: constant, reflecting the quasi two-dimensional character of the ML. However, we can notice important modifications at
122: higher pressures: the first band flattens associated to the increasing of the band gap at the zone boundary, indicating a
123: clear electronic localization. At the same time, as a consequence of the decrease in the energy difference between the
124: first two bands at $\Gamma$, the $p_{z}$-band, which is antisymmetric in the direction perpendicular to the plane of the
125: ML, starts to be occupied. Therefore, we have two types of carriers at the Fermi energy, a fact which
126: plays a fundamental role in the pressure induced properties of the electronic collective excitations, as will be explained
127: bellow. It is noteworthy that the Fermi \textit{line} of the ML also suffers significant deviations with increasing electronic density \cite{prb}, as it is shown in Fig. \ref{fig:fermi}.
128: \begin{figure}[h!]
129: \hspace*{0.5cm}\includegraphics[scale=0.28]{ECOSS_rs3.02.eps}\hspace*{0.75cm}
130: \includegraphics[scale=0.28]{ECOSS_rs2.15.eps}
131: \caption{Fermi \textit{lines} of a hexagonal lithium ML at equilibrium, $r_{s}=3.02$ (left), and a square lithium ML at $r_{s}=2.15$ (right). Solid line corresponds to the Fermi \textit{line} of the $s$-band. For values of the density $r_{s}\leq2.4$, the $p_{z}$-band starts to be occupied, so that this band is represented by a dashed line. The zone boundary is plotted with a dotted line.}
132: \label{fig:fermi}
133: \end{figure}
134:
135: These important pressure induced modifications in both the band structure and the Fermi \textit{line} of the ML are expected
136: to have a profound effect in the dynamical response function. In what follows, we will analyze the loss function,
137: proportional to the imaginary part of $g({\bf{q}};\omega)$, at two different selected densities, $r_{s}=3.02$ (equilibrium) and
138: $r_{s}=2.4$. At the latter density the phase transition has not yet been produced whereas the $p_{z}$-band is already
139: occupied. In such a way, we can conclude that the physical origin of the interesting phenomena which will be described bellow is not related to the $hex\rightarrow sq$ phase transition, but is a direct consequence of the presence of two different conductors at the Fermi energy.
140: \begin{figure}[h!]
141: \includegraphics[width=\linewidth,clip=true]{ss_equilibrium_plasmon.eps}
142: \caption{Loss function Im $g({\bf{q}};\omega)$ of the hexagonal lithium ML at equilibrium for two different values of the wavefunction in the $\Gamma K$ direction. One can see one-particle electron-hole excitations and two peaks associated to the
143: intraband and interband plasmons. }
144: \label{fig:eq_plasmon}
145: \end{figure}
146: In Fig. \ref{fig:eq_plasmon} we present the loss function of a lithium ML at equilibrium for two selected wavevectors in
147: the $\Gamma K$ direction. As we are mainly interested in collective excitations, small momentum transfers are considered.
148: Our spectra consists in one-particle electron-hole excitations with features corresponding to intraband excitations
149: at small energies and interband excitations at higher energies, together with two plasmon peaks. The first
150: peak is situated at an energy close to $\sim$3 eV and linked to the intraband collective excitations
151: inside the $s$-band, whereas the second one with energy of $\sim$5 eV is related to the interband plasmon, which physically
152: corresponds to collective motion of electrons in the direction perpendicular to the ML.
153: \begin{figure}[h!]
154: \vspace*{0.5cm}
155: \hspace*{1.0cm}\includegraphics[scale=0.23]{eq_dispersion.eps}
156: \caption{Dispersion curves of the \textit{ab initio} collective intraband (squares) and interband (circles) excitations in the $\Gamma K$ direction. The non-zero minimum energy in the limit ${\bf{q}}\rightarrow 0$ indicates that electrons located at neighbouring layers feel the interaction.}
157: \label{fig:eq_dispersion}
158: \end{figure}
159: Dispersion curves for both intraband and interband collective excitations are plotted in Fig. \ref{fig:eq_dispersion}.
160: As our selected wavefunctions are much smaller than the characteristic Fermi momentum of the ML,
161: we would expect a $\sqrt{\bf{q}}$ dispersion \cite{stern} for the two-dimensional intraband plasmon, instead of the non-zero minimum energy for 3D plasmons \cite{pines}. However, the Fourier transform of the electronic potential of
162: electrons located at neighboring layers decays exponentially with $|{\bf{q}|}L$, so that when dealing
163: with vectors whose size is
164: comparable to the inverse of the intermonolayer distance, $q\simeq 1/L=0.05$ a.u, electrons feel the interaction with
165: other layers. Therefore, the plasmons show a three dimensional character at very small momentum, which is reflected
166: in the gap at $|{\bf{q}}|\rightarrow 0$ that can be seen in Fig. \ref{fig:eq_dispersion}.
167:
168: Loss function of the ML with
169: $r_{s}=2.4$ at selected momenta in the $\Gamma K$ direction is presented in
170: Fig. \ref{fig:rs2.4_plasmon}.
171: \begin{figure}[h!]
172: %\vspace*{0.5cm}
173: \includegraphics[width=\linewidth,clip=true]{ss_rs2.4_plasmon.eps}
174: \caption{Loss function Im $g({\bf{q}};\omega)$ of the hexagonal lithium ML with $r_{s}=2.4$ for two different values of the wavefunction in the $\Gamma K$ direction. Besides the common intraband plasmon, close to the same energy as at equilibrium, and interband one, located at a higher energy, one can also notice a remarkable peak at small $\omega$, the characteristic feature of an acoustic plasmon.}
175: \label{fig:rs2.4_plasmon}
176: \end{figure}
177: The $s$-bandwidth at this density is very similar to that at equilibrium. As a consequence, the collective excitations
178: inside the $s$-band require a very similar energy at both densities, equilibrium and $r_{s}=2.4$. Thus,
179: the intraband
180: plasmon peak is situated at $\sim$3.5 eV, close to the location at equilibrium. However, the energy gap between the first
181: two bands, $s$ and $p_{z}$, increases under pressure, therefore requiring more energy the collective motions of electrons
182: between these two bands to be excitated, and shifting the corresponding interband plasmon peak to around $\sim$6.5 eV.
183: As well as the two plasmon peaks appearing also in absence of applied pressure, here
184: one can also observe peaks associated with two other interband collective excitations with energies $\sim$7.7 eV and
185: $\sim$9.5 eV, due to the presence of two partly occupied energy bands. In addition, a remarkable peak is induced by
186: pressure at energies close to zero. Besides the common intraband and interband collective excitations, the presence of
187: two type of carriers at the Fermi level with different velocities allows the existence of another kind of collective
188: modes, known as acoustic plasmons because of their linear dispersion in $|\bf{q}|$ \cite{slava_acoustic}, as it can be checked in Fig.
189: \ref{fig:rs2.4_dispersion}. The same acoustic plasmon was also predicted for a berilium ML \cite{aitor_plasmon}. Contrary to the case of two dimensional-like plasmons, which are well established experimentally \cite{nagao1,nagao2}, the acoustic
190: counterparts have not yet been unambiguously identified. It is known that, in this case, the light electrons screen the
191: Coulomb repulsion between the electrons with a lower velocity, so that an additional intraband acoustic collective
192: excitation is produced.
193:
194: \begin{figure}[h!]
195: \vspace*{0.5cm}
196: \hspace*{0.8cm}\includegraphics[scale=0.5]{acustico.eps}
197: \caption{Dispersion curve of the \textit{ab initio} acoustic plasmon induced by the applied pressure in a hexagonal lithium ML with $r_{s}=2.4$. A clear lineal dependence with $|\bf{q}|$ can be observed. Dots are calculations whereas dashed line is a fit.
198: The slope of the fit is close to the velocity of the slow electrons. }
199: \label{fig:rs2.4_dispersion}
200: \end{figure}
201: \begin{thebibliography}{55}
202: %\expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
203: %\expandafter\ifx\csname bibnamefont\endcsname\relax
204: % \def\bibnamefont#1{#1}\fi
205: %\expandafter\ifx\csname bibfnamefont\endcsname\relax
206: % \def\bibfnamefont#1{#1}\fi
207: %\expandafter\ifx\csname citenamefont\endcsname\relax
208: % \def\citenamefont#1{#1}\fi
209: %\expandafter\ifx\csname url\endcsname\relax
210: % \def\url#1{\texttt{#1}}\fi
211: %\expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
212: %\providecommand{\bibinfo}[2]{#2}
213: %\providecommand{\eprint}[2][]{\url{#2}}
214:
215: \bibitem{wigner}
216: E. Wigner and F. Seitz, {Phys. Rev.} 43 (1933) 804.
217:
218: \bibitem{NA}
219: J. B. Neaton and N. W. Ashcroft, {Nature} 400 (1999) 141.
220:
221: \bibitem{hanfland}
222: M. Hanfland, K. Syassen, N. E. Christensen and D. L. Novikov, {Nature} 408 (2000) 174.
223:
224: \bibitem{neaton}
225: J. B. Neaton and N. W. Ashcroft, {Phys. Rev. Lett.} 86 (2001) 2830.
226:
227: \bibitem{liambient}
228: K. I. Juntunen and J. T. Tuoriniemi, {Phys. Rev. Lett.} 93 (2004) 157201.
229:
230: \bibitem{shimizu}
231: K. Shimizu, H. Ishikawa, D. Takao, T. Yagi and K. Amaya, {Nature} 597 (2002) 597.
232:
233: %\bibitem{str}
234: %V. V. Struzhkin, M. I. Eremets, W. Gan, H. K. Mao and R. J. Hemley, {Science} 298 (2002) 1213.
235:
236: \bibitem{deemyad}
237: S. Deemyad and J. S. Shilling, {Phys. Rev. Lett.} 91 (2003) 167001.
238:
239: \bibitem{ashcroft_nature}
240: N. W. Ashcroft, {Nature} 419 (2002) 569.
241:
242: \bibitem{aitor}
243: A. Bergara, J. B. Neaton and N. W. Ashcroft, {Phys. Rev. B} 62 (2000) 8494.
244:
245: \bibitem{prb}
246: A. Rodriguez-Prieto and A. Bergara, {Phys. Rev. B} 72 (2005) 125406.
247:
248: \bibitem{aitor_plasmon}
249: A. Bergara, V. M. Silkin, E. V. Chulkov and P. M. Echenique, {Phys. Rev. B} 67 (2003) 245402.
250:
251: \bibitem{persson}
252: B. N. J. Persson and E. Zaremba, {Phys. Rev. B} 31 (1985) 1863.
253:
254: \bibitem{tsuei}
255: K. D. Tsuei, E. W. Plummer, A. Liebsch, E. Pehlke, K. Kempa and P. Bakshi, {Surf. Sci.} 302 (1991) 247.
256:
257: %\bibitem{liebsch}
258: %A. Liebsch, {Electronic Excitations at Metal Surfaces}, Plenum, New York, 1997.
259:
260: \bibitem{slava_plasmon}
261: V. M. Silkin, E. V. Chulkov and P. M. Echenique, {Phys. Rev. Lett.} 93 (2004) 176801.
262:
263: \bibitem{pines}
264: See, e.g., D. Pines, {Elementary Excitations in Solids}, Addison-Wesley, New York, 1963.
265:
266: \bibitem{tm}
267: N. Troullier and J. L. Martins, {Phys. Rev. B} 43 (1991) 1993.
268:
269: %\bibitem{tm}
270: %F. Aryasetiawan and O. Gunnarsson, {Phys. Rev. B} 49, 16214 (1994)
271:
272: %\bibitem{rs}
273: %$r_{s}$ is the two-dimensional linear density parameter, defined by the relation $A/N=\pi(r_{s}a_{0})^{2}$ where $N$ is the number of nuclei in a ML with area $A$ and $a_{0}$ is the Bohr radius.
274:
275: \bibitem{stern}
276: F. Stern, {Phys. Rev. Lett.} 18 (1967) 546.
277:
278: \bibitem{slava_acoustic}
279: V. M. Silkin, A. Garc\'ia-Lekue, J. M. Pitarke, E. V. Chulkov, E. Zaremba and P. M. Echenique, Europhys. Lett. 66 (2004) 260.
280:
281: \bibitem{nagao1}
282: T. Nagao, T. Hildebrant, M. Henzler and S. Hasegawa, {Surf. Sci.} 493 (2001) 680.
283:
284: \bibitem{nagao2}
285: T. Nagao, T. Hildebrant, M. Henzler and S. Hasegawa, {Phys. Rev. Lett.} 86 (2001) 5747.
286:
287: \end{thebibliography}
288:
289: \end{document}
290:
291:
292: We acknowledge partial support by the University of the Basque
293: Country, the Basque Hezkuntza, Unibertsitate eta Ikerketa Saila, MCyT
294: (Grant No. MAT 2001-0946) and the European FP6 Network of Excellence
295: (FP6-NoE NANOQUANTA (500198-2)).
296: