1: \documentclass[aps,prl,twocolumn,groupedaddress,amssymb,amsmath,showpacs]{revtex4}
2: \usepackage{dcolumn}
3: \usepackage{graphicx}
4:
5:
6: \bibliographystyle{apsrev}
7:
8: \begin{document}
9:
10: %\preprint{}
11:
12: %Title of paper
13: \title{Core-Level Photoelectron Spectroscopy Probing Local Strain at Silicon Surfaces and Interfaces}
14:
15: \author{Oleg V. Yazyev}
16: \email[Electronic address: ]{oleg.yazyev@epfl.ch}
17: \affiliation{Ecole Polytechnique F\'ed\'erale de Lausanne (EPFL),
18: Institute of Chemical Sciences and Engineering, CH-1015 Lausanne, Switzerland}
19: \author{Alfredo Pasquarello}
20: \affiliation{Ecole Polytechnique F\'ed\'erale de Lausanne (EPFL),
21: Institute of Theoretical Physics, CH-1015 Lausanne, Switzerland}
22: \affiliation{Institut Romand de Recherche Num\'erique en Physique
23: des Mat\'eriaux (IRRMA), CH-1015 Lausanne, Switzerland}
24: \date{\today}
25:
26: \begin{abstract}
27:
28: Using a first-principles approach, we investigate the origin of
29: the fine structure in Si 2$p$ photoelectron spectra at the
30: Si(100)2$\times$1 surface and at the Si(100)-SiO$_2$ interface.
31: Calculated and measured shifts show very good agreement for both
32: systems. By using maximally localized Wannier functions, we provide an
33: interpretation in which the effects due to the electronegativity of
34: second nearest neighbor atoms and due to the local strain field
35: are distinguished. Hence, in combination
36: with accurate modeling, photoelectron spectroscopy can provide a
37: direct measure of the strain field at the atomic scale.
38:
39: \end{abstract}
40:
41: \pacs{79.60.-i,73.20.-r,68.35.-p}
42:
43:
44: \maketitle
45:
46: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
47: %% MAINMATTER
48: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
49: The increasing availability of synchrotron radiation facilities
50: is bearing X-ray photoemission spectra of unprecedented resolution
51: characterizing surfaces and interfaces \cite{Himpsel90}.
52: The achieved sensitivity is
53: sufficient to distinguish inequivalent subsurface atoms with
54: identically composed first-neighbor shells. Hence, the
55: interpretation of such core-level spectra can no longer be
56: achieved with simple electronegativity arguments, but requires
57: the consideration of the interplay between local strain fields
58: and electronegativity effects of second-nearest neighbors.
59:
60: These difficulties are strikingly illustrated for the Si(100)2$\times$1 surface
61: and for the technologically relevant Si(100)-SiO$_2$ interface,
62: which have been the objects of numerous highly resolved X-ray
63: photoemission investigations.
64: While the shifts pertaining to the first-layer dimer atoms
65: of the Si(100)2$\times$1 surface have been identified,
66: the other lines appearing in highly resolved Si $2p$ spectra still
67: lack a consensual assignment \cite{Landemark92,DePadova98,Koh03}. For the Si(100)-SiO$_2$ interface,
68: highly resolved spectra show fine structure in
69: the nonoxidized Si line, with extra components at lower
70: (Si$^\alpha$) and higher binding energy (Si$^\beta$) with respect to the
71: Si bulk line \cite{Oh01,Dreiner05}.
72:
73: We investigate the origin of the fine structure in Si $2p$ photoemission
74: spectra at silicon surfaces and interfaces using density functional theory
75: calculations based on pseudopotentials (PPs) and plane waves \cite{Car85-CPMD}.
76: To interpret the Si 2$p$ photoelectron spectra at silicon
77: surfaces and interfaces, it is necessary to evaluate core-level
78: shifts with respect to the Si bulk line.
79: Since these shifts mainly result from the
80: relaxation of valence electron states, their accurate
81: determination is possible within a PP scheme, which does not
82: treat core electrons explicitly \cite{Pehlke93,Pasquarello95}.
83: We calculated Si 2$p$
84: shifts including the effect of core-hole relaxation by taking
85: total energy differences between two separate self-consistent calculations.
86: First, the ground-state energy is determined; then the PP of a
87: given Si atom is replaced by another PP which simulates the
88: presence of a screened $2p$ hole in its core \cite{Pehlke93}.
89:
90: \begin{table}
91: \caption{\label{tab2} Comparison between calculated and measured
92: Si $2p$ shifts at the Si(001)-$c$(4$\times$2) surface. The shifts
93: (in eV) are given with respect the Si bulk line. The meaning of labels
94: is explained in Fig.~\ref{fig1}.}
95: \begin{ruledtabular}
96: \begin{tabular}{ccccc}
97: & \multicolumn{1}{c}{Theory} & \multicolumn{3}{c}{Experiment} \\
98: & \multicolumn{1}{c}{Present} &
99: \multicolumn{1}{c}{Ref.\ \cite{Landemark92}} &
100: \multicolumn{1}{c}{Ref.\ \cite{DePadova98}} &
101: \multicolumn{1}{c}{Ref.\ \cite{Koh03}}\\
102: \hline
103: $1u$ & -0.49 & -0.49 & -0.49 & -0.50 \\
104: $1d$ & 0.02 & 0.06 & 0.06 & 0.06 \\
105: $2$ & -0.01 & & & \\
106: $3$ & -0.14 & -0.21 & -0.20 & -0.21 \\
107: $4$ & -0.26 & & & \\
108: $3'$ & 0.24 & 0.22 & 0.20 & 0.23 \\
109: $4'u$ & 0.21 & & & \\
110: $4'd$ & 0.10 & & & \\
111: \end{tabular}
112: \end{ruledtabular}
113: \end{table}
114:
115: \begin{figure}
116: \includegraphics[width=3.0cm]{scheme1.eps}
117: \caption{\label{fig1}
118: The adopted labeling of Si atoms at the Si(001)-$c$(4$\times$2) surface.
119: The $4'$ atoms can be further
120: separated into inequivalent up ($4'u$) and down ($4'd$) atoms.}
121: \end{figure}
122:
123: For the Si(100)-c(4$\times$2)
124: surface we find binding energies of $-$0.49 eV and 0.02 eV for
125: dimer-up and dimer-down atoms, in good agreement with experimental results \cite{Landemark92,DePadova98,Koh03}.
126: The second-layer
127: atoms are found to yield very small shifts ($-$0.01~eV) while
128: third- and fourth-layer atoms give shifts to
129: both lower ($3$, $4$) and higher binding energies ($3'$, $4'$) with
130: respect to the Si bulk line (Table~\ref{tab2}). Shifts of deeper layers are
131: negligible.
132:
133: For the Si(100)-SiO$_2$ interface, we adopt a model structure which
134: incorporates a realistic transition region \cite{Bongiorno03}.
135: Calculated binding energies are
136: given in Fig.\ \ref{fig2} along the direction orthogonal to the
137: interface plane. The binding energies of oxidized Si atoms
138: increase almost linearly with oxidation state
139: \cite{Pasquarello95}, and show quantitative agreement
140: with experimental values (Table \ref{tab3}). The focus of the
141: present investigation is on nonoxidized Si atoms, which give
142: shifts with a significant spread near the interface. The
143: simulated spectrum associated to these Si atoms could be
144: decomposed into three components (Fig.\ \ref{fig2}, inset),
145: yielding shifts for the Si$^\alpha$ and Si$^\beta$ lines of $-$0.21 eV and 0.32 eV,
146: which also agree with experimental data (Table~\ref{tab3}).
147:
148: \begin{figure}
149: \includegraphics[width=8.5cm]{figure2.eps}
150: \caption{\label{fig2}
151: Calculated Si 2$p$ shifts at the Si(100)-SiO$_2$
152: interface along an orthogonal direction to the interface.
153: The Si$^{+n}$ correspond to oxidized Si atoms with $n$
154: O neighbors ($n$=1 - $\blacktriangledown$, 2 - $\blacktriangle$,
155: 3 - $\blacktriangleright$, 4 - $\blacktriangleleft$) and are
156: color-coded in the atomistic model.
157: Nonoxidized Si atoms (Si$^0$) with ($\circ$) and without ($\bullet$)
158: second-neighbor O atoms are distinguished.
159: Inset: simulated spectrum for Si$^0$
160: (Gaussian broadened: $\sigma$=0.08~eV) and
161: its decomposition in three components.}
162: \end{figure}
163:
164: \begin{table}
165: \caption{\label{tab3} Comparison between calculated and measured
166: Si $2p$ shifts at the Si(100)-SiO$_2$ interface. The Si bulk line
167: is taken as reference.}
168: \begin{ruledtabular}
169: \begin{tabular}{ccccc}
170: & \multicolumn{1}{c}{Theory} & \multicolumn{2}{c}{Experiment} \\
171: & \multicolumn{1}{c}{Present} &
172: \multicolumn{1}{c}{Ref.\ \cite{Oh01}} &
173: \multicolumn{1}{c}{Ref.\ \cite{Dreiner05}}\\
174: \hline
175: $\alpha$ & -0.21 & -0.25 & -0.22 \\
176: $\beta$ & 0.32 & 0.20 & 0.34 \\
177: +1 & 0.78 & 1.00 & 0.95 \\
178: +2 & 1.40 & 1.82 & 1.78 \\
179: +3 & 2.37 & 2.62 & 2.60 \\
180: +4 & 3.64 & 3.67 & 3.72 \\
181: \end{tabular}
182: \end{ruledtabular}
183: \end{table}
184:
185: Binding energy shifts of Si atoms with identically composed first-neighbor shells
186: are caused by electron density displacements due to either the local strain fields
187: or electronegativity effects.
188: We carried out an analysis in terms of maximally localized Wannier functions (MLWFs)
189: \cite{Marzari97}
190: in order to distinguish between these two effects \cite{Yazyev06}.
191: For the Si(100)-c(4$\times$2) surface, we find that, apart from the shifts associated
192: with the dimer atoms, the other shifts mainly result from the local strain induced
193: by the surface reconstruction.
194: For the interface, the Si atoms with second-neighbor O atoms contribute to the
195: Si$^\alpha$ line, while the Si$^\beta$ line originates from bond elongations
196: of Si atoms without second-neighbor O atoms. From the experimental shift of about
197: 0.3~eV, we infer the occurrence of Si atoms with an average bond length elongation
198: of $\sim$~0.05~\AA.
199:
200: In conclusion, we revealed the physical mechanisms underlying
201: the fine structure in Si 2$p$ photoelectron spectra
202: at silicon surfaces and interfaces.
203: Maximally localized Wannier functions offer a powerful tool
204: to identify core-level shifts originating from the
205: electronegativity of farther neighbors. A key result of our work
206: is then that the remaining lines provide an atomic-scale probe
207: of the strain in the structure. This confers to photoelectron
208: spectroscopy a new functionality in addition to the detection
209: of chemical composition.
210:
211: \begin{thebibliography}{12}
212:
213: \bibitem{Himpsel90}
214: F. J. Himpsel {\it et al.},
215: in {\it Proceedings of the International School of Physics Enrico Fermi},
216: edited by M. Campagna and R. Rosei (Elsevier, Amsterdam, 1990), p.\ 203--236.
217:
218: \bibitem{Landemark92}
219: E. Landemark {\it et al.},
220: {Phys.\ Rev.\ Lett.}\ {\bf 69}, 1588 (1992).
221:
222: \bibitem{DePadova98}
223: P. De~Padova {\it et al.},
224: {Phys.\ Rev.\ Lett.}\ {\bf 81}, 2320 (1998).
225:
226: \bibitem{Koh03}
227: H. Koh {\it et al.},
228: {Phys.\ Rev.\ B} {\bf 67}, 073306 (2003).
229:
230: \bibitem{Oh01}
231: J.~H. Oh {\it et al.},
232: {Phys.\ Rev.\ B} {\bf 63}, 205310 (2001).
233:
234: \bibitem{Dreiner05}
235: S. Dreiner, M. Sch\"urmann, and C. Westphal,
236: {J.\ Electron Spectrosc.\ Relat.\ Phenom.}\ {\bf 144-147}, 405 (2005).
237:
238: \bibitem{Car85-CPMD}
239: R. Car and M. Parrinello,
240: {Phys.\ Rev.\ Lett.}\ {\bf 55}, 2471 (1985);
241: CPMD version 3.9.1, Copyright IBM Corp 1990-2004,
242: Copyright MPI f\"ur Festk\"orperforschung
243: Stuttgart 1997-2001, \texttt{http://www.cpmd.org}.
244:
245: \bibitem{Pehlke93}
246: E. Pehlke and M. Scheffler,
247: { Phys.\ Rev.\ Lett.}\ {\bf 71}, 2338 (1993).
248:
249: \bibitem{Pasquarello95}
250: A. Pasquarello, M.~S. Hybertsen, and R. Car,
251: {Phys.\ Rev.\ Lett.}\ {\bf 64}, 1024 (1995);
252: { Phys.\ Rev.\ B} {\bf 53}, 10942 (1996).
253:
254: \bibitem{Bongiorno03}
255: A. Bongiorno, A. Pasquarello, M.~S. Hybertsen, and L. C. Feldman,
256: { Phys.\ Rev.\ Lett.}\ {\bf 90}, 186101 (2003).
257:
258: \bibitem{Marzari97}
259: N. Marzari and D. Vanderbilt,
260: { Phys.\ Rev.\ B} {\bf 56}, 12847 (1997).
261:
262: \bibitem{Yazyev06}
263: O.~V.~Yazyev and A.~Pasquarello,
264: { Phys.\ Rev.\ Lett.} {\bf 96}, 157601 (2006).
265:
266: \end{thebibliography}
267:
268:
269: \end{document}
270:
271: \endinput
272: %%
273: %% End of file `template-8d.tex'.
274:
275: