Structure determination of the indium-induced Si(111)-(4x1) reconstruction by surface x-ray diffraction
Extracted source text ยท arXiv cond-mat/9909286
\documentstyle[multicol,epsfig,prb,aps]{revtex}
\begin{document}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% title and abstract %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\title{Structure determination of the indium-induced Si(111)-(4$\times$1)
reconstruction by surface x-ray diffraction}
\author{O. Bunk, \cite{author_email}
G. Falkenberg, J.H. Zeysing, L. Lottermoser, and R.L. Johnson}
\address{II. Institut f\"ur
Experimentalphysik, Universit\"at Hamburg, Luruper Chaussee 149,\\
D-22761 Hamburg, Germany}
\author{M. Nielsen, F. Berg-Rasmussen, J. Baker, and R. Feidenhans'l}
\address{Condensed Matter Physics and Chemistry Department,\\
Ris\o\ National Laboratory, DK-4000 Roskilde, Denmark}
\date{21 October 1998}
\maketitle
\begin{abstract}
A detailed structural model for the indium-induced
Si(111)-(4$\times$1) surface reconstruction has been
determined by analyzing an extensive set of x-ray-diffraction data recorded
with monochromatic ($\hbar\omega$ = 9.1~keV) synchrotron radiation.
The reconstruction is quasi-one-dimensional. The main features
in the structure are chains of silicon atoms alternating with
zigzag chains of indium atoms on top of an essentially
unperturbed silicon lattice.
The indium coverage corresponds to one monolayer.
The structural model consistently explains all previously published
experimental data.
\end{abstract}
\pacs{
PACS: 68.35.Bs % Surface structure and topography
}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% main text %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{multicols}{2}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%\section{Introduction}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Considerable interest has focused recently on ad\-sor\-bate-induced
modification of semiconductor surfaces as a technique to create
nanoscale quantum structures of high perfection.
In this paper we report the formation of quasi-one-dimensional (1d) chains
on the (4$\times$1)-reconstructed Si(111) surface induced by the adsorption
of indium and present the atomic structure as determined by
surface x-ray diffraction (SXRD).
Several surface reconstructions are induced by indium
on Si(111) (see, e.g., Kraft {\it et al.},
Ref.\ \onlinecite{kraft_si111in_review}) depending on the coverage;
the Si(111)-(4$\times$1)-In
reconstruction marks the borderline between the semiconducting, low indium
coverage, and metallic, high-coverage, phases. \cite{kraft_si111in_review}
Despite the fact that the indium-induced Si(111)-(4$\times$1)
reconstruction was reported for the first time by Lander and Morrison
in 1965\cite{landermorrison_si111alin} and has been investigated with
a variety of techniques subsequently, no definitive structural model could be
established to date and some of the experimental results are apparently
inconsistent.
In direct \cite{abukawa_si1114x1in_arpes} and inverse
\cite{hillmclean_si1114x1in_ipes_prb56} photoemission investigations on
single-domain samples and recently in inverse photoemission
investigations on a three-domain sample \cite{hillmclean_si1114x1in_3dom_ipes}
the Si(111)-(4$\times$1)-In reconstruction showed a quasi-1d metallic
behavior.
Additionally to this interesting electronic property the surface exhibits
an image state that also showed quasi-1d behavior in its strongly
anisotropic dispersion: Along a certain direction the dispersion is very well
described by a free electron parabola whereas in the perpendicular direction
the dispersion falls below the free electron parabola.
\cite{hillmclean_ipes_imagestate}
Scanning tunneling microscopy (STM) investigations of the
Si(111)-(4$\times$1)-In reconstruction
\cite{kraft_si111in_review,nogami_si111in_prb36,stevens_si1114x1in_prb47,%
saranin_si1114x1in_prb56,saranin_si1114x1in_prb55,%
owmanmartensson_si111s3xs3in_stm}
resolved zig-zag chains running in the $\langle\overline{1}10\rangle$
directions in the filled-state images and linear chains in the
empty-state images. Tunneling data was acquired at bias
voltages down to 0.04~V consistent with metallic behavior in agreement
with scanning tunneling spectroscopy results. \cite{kraft_si111in_review}
Adsorbed hydrogen was found to displace the indium atoms.
\cite{saranin_si1114x1in_prb56,saranin_si1114x1in_prb55,%
owmanmartensson_si111s3xs3in_stm}
Filled-state STM images of the hydrogenated substrate
show a (4$\times$1) reconstruction with straight chains
instead of the broad zigzag chains on the indium-terminated surface.
\cite{saranin_si1114x1in_prb56,saranin_si1114x1in_prb55,%
owmanmartensson_si111s3xs3in_stm}
From these observations and results found in the literature
Saranin {\it et al.} \cite{saranin_si1114x1in_prb56}
proposed a structural model.
This model as well as the model derived by Collazo-Davila {\it et al.}
\cite{ldm_si1114x1in_surfrevlett4} by applying
direct methods to transmission electron diffraction (TED) data are at
variance with the x-ray photoelectron spectroscopy (XPS) results of
Abukawa {\it et al.} \cite{abukawa_si1114x1in_xps}
Since the core-level spectra did not show peaks corresponding to surface
silicon atoms a complex substrate reconstruction can be ruled out.
All the inconsistencies between the earlier experimental
results are eliminated with our structural model.
The determination of the geometric structure lays the foundation for a
theoretical investigation of the interesting electronic
structure of this system.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%\section{Experimental}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
We used the well established experimental technique SXRD to determine
the structure of the Si(111)-(4$\times$1)-In reconstruction.
To minimize the uncertainties induced by not well controlled sample
preparation conditions we employed the unique combination of an
ultra high vacuum (UHV) system equipped with standard techniques like
reflection high energy electron diffraction and
low energy electron diffraction (LEED) facilities close to the wiggler
beamline BW2 at HASYLAB (Hamburger Synchrotronstrahlungslabor)
and a portable UHV chamber for the SXRD measurements.
We used p-type (B doped) Si(111) wafers (Wacker Chemie) with a resistivity of
7~$\Omega$cm and nominally no miscut.
STM measurements on the clean Si(111)-(7$\times$7) surface showed domain
sizes of more than 10000~\AA{} so the miscut was less than 0.02$^\circ$.
The carefully outgased sample was cleaned repeatedly by `flashing' to
$\sim$1150$^\circ$C for 15-20~s and slow cooling from 900$^\circ$C to
room temperature.
Indium was deposited from a Knudsen cell at a rate of
$\sim$0.4~ML/min on the sample at $\sim$500$^\circ$C until the
($\sqrt[]{3}\times\sqrt[]{3}$)R30$^\circ$ reconstruction was observed;
at this stage the stacking fault of the Si(111)-(7$\times$7) reconstruction
was removed.
Further deposition at a lower temperature of $\sim$430$^{\circ}$C
(with a lower probability for indium desorption) yielded first the
($\sqrt[]{31}\times\sqrt[]{31}$) and finally the ($4\times$1) reconstruction.
The sample was transferred in a portable UHV chamber to the wiggler
beamline BW2 at HASYLAB for the x-ray diffraction measurements.
The incident monochromized x-rays
with an energy of 9.1~keV impinged on the sample at a grazing angle
of 0.5$^\circ$. An in-plane data-set of 296 reflections with
$l=0.07$ was recorded by rotating the
crystal about the surface normal. The background subtracted integrated
intensities were corrected for the Lorentz factor, polarization factor,
active sample area and the rod interception appropriate for the
z-axis geometry. \cite{vlieg_correction_factors}
In SXRD measurements the fractional-order reflections which belong to one
domain do not overlap with the reflections which belong to the other two
rotational domains.
By comparing equivalent fractional-order reflections the areas of the
three rotational domains were found to correspond to
25\%, 37\% and 38\% of the total surface area. The equivalent
rescaled fractional-order reflections were averaged under the
assumption of mirror-lines running along $\langle 11\overline{2}\rangle$
and a systematic error of $\epsilon=13.7$\% was determined.
The rods were scaled according to
the corresponding in-plane intensity and therefore only one overall
scale factor was necessary for the data analysis.
In total the data-set consists of 550 symmetry inequivalent reflections,
61 along two crystal truncation rods (ctrs), 337 along 11 fractional-order
rods and 152 in-plane reflections.
In the following we use LEED coordinates with
${\bf a} = 1/2 [10\overline{1}]_{\mbox{cubic}}$,
${\bf b} = 1/2 [\overline{1}10]_{\mbox{cubic}}$ and
${\bf c} = 1/3 [111]_{\mbox{cubic}}$. The cubic coordinates are
in units of the silicon lattice constant (5.43~\AA) and therefore
$|a| = |b| = 3.84$~\AA{} and $|c| = 3.14$~\AA. The absolute values
of the reciprocal coordinates including a factor of $2\pi$ are
$|a^\ast| = |b^\ast| = 1.89$~\AA$^{-1}$ and
$|c^\ast| = 2.00$~\AA$^{-1}$.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%\section{Results and Discussion}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Usually the first step in the analysis of surface diffraction data is
to plot the Patterson function, i.e.\ the pair correlation
function of the electron density. The Patterson function
projected in the surface plane can be calculated from the
in-plane reflections which have a small momentum transfer perpendicular
to the surface. A contour plot of this function is shown in
Fig.~\ref{fig:pat_proj} and is qualitatively in good agreement with that
previously reported by Finney {\it et al.} \cite{finney_si1114x1in_sxrd}
Each peak in the Patterson function corresponds to an important interatomic
vector in the surface reconstruction. Since indium ($Z^2=49^2$) is a much
stronger scatterer than silicon ($Z^2=14^2$) the peaks correspond
to interatomic vectors between indium atoms.
The three interatomic distance vectors which can be seen in the
Patterson function in Fig.~\ref{fig:pat_proj} indicate that at least three
different indium sites must be involved in the reconstruction.
It is highly unlikely that the indium coverage can be
0.5~ML (monolayer) since that would only correspond to two indium atoms
per unit cell.
It has been argued that a reconstruction of
the substrate might lead to further peaks in the Patterson function
\cite{ldm_si1114x1in_surfrevlett4} but the XPS data
\cite{abukawa_si1114x1in_xps} make this unlikely.
During the course of the data analysis we tested
several possible initial indium atom configurations and determined
the substrate structure in the subsequent refinement.
The model found is shown in Fig.~\ref{fig:model}.
The correctness of the model is proven both by the good agreement of the
measured data with the intensities calculated from the model structure
shown in Fig.~\ref{fig:data} and
by the overall reduced $\chi^2$ value of 1.5. We will now describe the
building blocks of the model and resolve some of the apparent
inconsistencies between the previously reported experimental results.
The model shown in Fig.~\ref{fig:model} consists
of a zigzag chain of silicon atoms as found in the $\pi$-bonded chain model
for the (2$\times$1) reconstruction of the Si(111) surface
\cite{pandey_si1112x1}
on top of an essentially unreconstructed substrate. This model is consistent
with the XPS investigations which showed that
no strong silicon surface component was present in the Si $2p$ spectra.
\cite{abukawa_si1114x1in_xps}
Even in the absence of indium atoms the silicon chains on the substrate
possess a (4$\times$1) periodicity and it is highly likely that the chains
observed after hydrogen adsorption
\cite{saranin_si1114x1in_prb56,owmanmartensson_si111s3xs3in_stm}
are made up of silicon atoms.
The Si(111)-(4$\times$1)-In reconstruction is completed by adding two
zigzag rows of indium atoms in the space between
the silicon chains. The arrangement of the indium atoms is similar to the
arrangement proposed on the basis of $\mu$-probe Auger electron
diffraction investigations, \cite{nakamura_si1114x1in_aed_surfsci256}
and also used in a previous SXRD study.
\cite{finney_si1114x1in_sxrd}
There are two inequivalent types of indium atoms in
agreement with the XPS results. \cite{abukawa_si1114x1in_xps}
The indium atoms next to
the silicon chains are probably covalently bonded to the silicon chain atoms.
For the inner indium atoms the bonding configuration is not so obvious.
In Fig.~\ref{fig:model} a bonding configuration of covalent $p_{x,y,z}$
bonds to the neighboring indium atoms and down to the silicon substrate
under an angle of approximately 90$^\circ$ are
shown. The nearest neighbor distance of the indium atoms is within
the range of 2.98~\AA{} to 3.14~\AA{}.
Electron counting can not be strictly applied to determine the bonding
configuration because the surface has 1d metallic character.
If we assume that the bonding configuration shown in Fig.~\ref{fig:model}
is a first approximation to the more complex real configuration then
it is evident that the surface free energy of this reconstruction is
lower than for other models since there are no silicon dangling bonds and all
the group III indium atoms are trivalently bonded.
The present SXRD data do not permit an accurate determination of the
bond charge densities; the detailed bonding configuration will have to be the
subject of a future theoretical investigation.
An important question is how the model shown in Fig.~\ref{fig:model}
with four indium atoms per unit cell (1~ML) can be used to explain the
results of the TED \cite{ldm_si1114x1in_surfrevlett4} and
impact collision ion scattering spectrometry (ICISS) investigations
\cite{stevens_si1114x1in_prb47}
which resulted in models with only two indium atoms per unit cell.
In the TED study direct methods were used to analyze the data and
electron density maps with indium atom configurations compatible with the
TED data were presented including
plausible configurations with four indium atoms
(e.g.\ Fig.~2 (f) in Ref.\ \onlinecite{ldm_si1114x1in_surfrevlett4}).
The 1/2~ML model which was compatible with the TED
data is at variance with the SXRD data as shown in Fig.~\ref{fig:data}~(c)
where the dashed lines calculated using the
1/2~ML model do not reproduce the measured data adequately.
A preliminary data analysis of the SXRD data using direct methods resulted
in the indium atom configuration shown in Fig.~\ref{fig:model} thereby
indicating the correctness of our model. \cite{ldmarks_privcom} The
TED data were not sufficient to include dynamical diffraction effects
and the resolution was probably not good enough to rule out the wrong
indium atom configuration. \cite{ldmarks_privcom}
The reduced $\chi^2$ value for the model proposed by Saranin {\it et al.}
\cite{saranin_si1114x1in_prb56} is 7.3 and therefore this model
can also be definitely eliminated.
The side view of the Si(111)-(4$\times$1)-In
reconstruction shown in Fig.~\ref{fig:model}(b) and the atomic coordinates
given in Table \ref{tab:pos} show that the atoms in the topmost layer
have three different heights.
The highest atoms are the indium atoms
bonded to the silicon chain, the inner indium atoms of the indium ``stripe''
are lower and the silicon chain atoms are lowest. The existence of different
heights is in agreement with the results of a previously performed
STM investigation. \cite{owmanmartensson_si111s3xs3in_stm}
Low energy ion scattering is very sensitive to the topmost surface layer.
The configuration of the topmost indium atoms bonded to the silicon chains
strongly resembles the 1/2~ML model with indium atoms on
$H_3$ and $T_4$ sites that has been found to reproduce the ICISS data
collected using 2~keV Li$^+$ ions. \cite{stevens_si1114x1in_prb47}
Another 1/2~ML model
with a zigzag chain of indium atoms on $T_4$ sites also gave reasonable
agreement with the ICISS data \cite{stevens_si1114x1in_prb47}
since such chains are present in our model as shown in Fig.~\ref{fig:model}.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%\section{Summary and Outlook}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In summary, the structure of the Si(111)-(4$\times$1)-In reconstruction
has been determined using surface x-ray diffraction.
The quasi-one-dimensional
character of this surface reconstruction is given by zigzag chains of
silicon atoms on top of an unreconstructed silicon substrate and four indium
atoms per unit cell (1~ML) arranged in two zigzag chains in the gap
between the silicon chains.
The indium atom arrangement may also be regarded as being
quasi hexagonal. We have shown that most of the previously published
experimental data are consistent with the new structural model.
We hope that the atomic coordinates given in Table \ref{tab:pos} will pave
the way for detailed theoretical investigations of the interesting electronic
structure of this system.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%\section{Acknowledgments}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
We thank the staff of HASYLAB for their technical assistance.
Financial support from the Danish Research Council through Dansync,
the Bundesministerium f\"ur Bildung, Wissenschaft,
Forschung und Technologie (BMBF) under project no. 05622GUA1 and the
Volks\-wagen Stiftung is gratefully acknowledged.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% references %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{references}
\bibitem[*]{author_email} e-mail address: Oliver.Bunk@desy.de
\bibitem{kraft_si111in_review}
J. Kraft, M.G. Ramsey, and F.P. Netzer,
Phys.\ Rev.\ B {\bf 55}, 5384 (1997).
\bibitem{landermorrison_si111alin}
J.J. Lander and J. Morrison, J. Appl.\ Phys.\ {\bf 36}, 1706 (1965).
\bibitem{abukawa_si1114x1in_arpes}
T. Abukawa, M. Sasaki, F. Hisamatsu, T. Goto, T. Kinoshita,
A. Kakizaki, and S. Kono,
Surf.\ Sci.\ {\bf 325}, 33 (1995).
\bibitem{hillmclean_si1114x1in_ipes_prb56}
I.G. Hill and A.B. McLean, Phys.\ Rev.\ B {\bf 56}, 15725 (1997).
\bibitem{hillmclean_si1114x1in_3dom_ipes}
I.G. Hill and A.B. McLean, Phys.\ Rev.\ B. {\bf 59}, 9791 (1999).
\bibitem{hillmclean_ipes_imagestate}
I.G. Hill and A.B. McLean, Phys.\ Rev.\ Lett.\ {\bf 82}, 2155 (1999).
\bibitem{nogami_si111in_prb36}
J. Nogami, S. Park, and C.F. Quate, Phys.\ Rev.\ B {\bf 36} 6221 (1987).
\bibitem{stevens_si1114x1in_prb47}
J.L. Stevens, M.S. Worthington, and I.S.T. Tsong,
Phys.\ Rev.\ B {\bf 47}, 1453 (1993).
\bibitem{saranin_si1114x1in_prb56}
A.A. Saranin, A.V. Zotov, K.V. Ignatovich, V.G. Lifshits, T. Numata,
O. Kubo, H. Tani, M. Katayama, and K. Oura,
Phys.\ Rev.\ B {\bf 56}, 1017 (1997).
\bibitem{saranin_si1114x1in_prb55}
A.A. Saranin, E.A. Khramtsova, K.V. Ignatovich, V.G. Lifshitz,
T. Numata, O. Kubo, M. Katayama, I. Katayama, and K. Oura,
Phys.\ Rev.\ B {\bf 55}, 5353 (1997).
\bibitem{owmanmartensson_si111s3xs3in_stm}
F. Owman and P. M{\aa}rtensson, Surf.\ Sci.\ {\bf 359}, 122 (1996).
\bibitem{ldm_si1114x1in_surfrevlett4}
C. Collazo-Davila, L.D. Marks, K. Nishii, and Y. Tanishiro,
Surf.\ Rev.\ Lett.\ {\bf 4}, 65 (1997).
\bibitem{abukawa_si1114x1in_xps}
T. Abukawa, M. Sasaki, F. Hisamatsu, N. Nakamura, T. Kinoshita,
A. Kakizaki, T. Goto, and S. Kono,
J. Electron.\ Spectrosc.\ \& Relat.\ Phenom.\ {\bf 80}, 233 (1996).
\bibitem{vlieg_correction_factors}
E. Vlieg, J. Appl.\ Cryst.\ {\bf 30} 532 (1997).
\bibitem{finney_si1114x1in_sxrd}
M.S. Finney, C. Norris, P.B. Howes, M.A. James, J.E. MacDonald,
A.D. Johnson, and E. Vlieg, Physica B {\bf 198}, 246 (1994).
\bibitem{pandey_si1112x1}
K.C. Pandey, Phys.\ Rev.\ Lett.\ {\bf 47}, 1913 (1981).
\bibitem{nakamura_si1114x1in_aed_surfsci256}
N. Nakamura, K. Anno, and S. Kono, Surf.\ Sci.\ {\bf 256}, 129 (1991).
\bibitem{ldmarks_privcom} L.D. Marks (private communication).
\end{references}
\newpage
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% table: atom positions %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{table}[h]
\begin{center}
\begin{tabular}[h]{|c|c|c|d|}
& position [LEED-coord.] & deviation $\bf d$ [\AA] %
&$|{\bf d}|$ [\AA] \\\hline\hline
In & (0.11,0.06, 0.86) & & \\\hline
In & (0.86,0.93, 0.85) & & \\\hline
In & (1.53,0.77, 0.99) & & \\\hline
In & (3.43,0.22, 0.99) & & \\\hline
Si & (2.28,0.14, 0.73) & & \\\hline
Si & (2.71,0.86, 0.76) & & \\\hline\hline
Si & (0.31,0.65,-0.26) & (-0.11,-0.05,-0.03) & 0.10\\\hline
Si & (3.96,0.98,-0.00) & (-0.15,-0.08,-0.01) & 0.13\\\hline
Si & (1.30,0.65,-0.25) & (-0.12,-0.06, 0.01) & 0.10\\\hline
Si & (0.96,0.98, 0.04) & (-0.16,-0.08, 0.13) & 0.19\\\hline
Si & (2.29,0.64,-0.33) & (-0.17,-0.09,-0.24) & 0.28\\\hline
Si & (2.01,0.00,-0.03) & ( 0.03, 0.02,-0.10) & 0.10\\\hline
Si & (3.29,0.65,-0.23) & (-0.16,-0.08, 0.05) & 0.15\\\hline
Si & (2.95,0.97,-0.01) & (-0.20,-0.10,-0.03) & 0.18\\\hline\hline
Si & (0.66,0.33,-1.24) & (-0.04,-0.02, 0.04) & 0.07\\\hline
Si & (0.33,0.67,-1.01) & (-0.01,-0.01,-0.02) & 0.02\\\hline
Si & (1.65,0.32,-1.26) & (-0.08,-0.04,-0.03) & 0.07\\\hline
Si & (1.32,0.66,-0.99) & (-0.06,-0.03, 0.03) & 0.06\\\hline
Si & (2.67,0.33,-1.27) & ( 0.00, 0.00,-0.05) & 0.05\\\hline
Si & (2.32,0.66,-1.04) & (-0.05,-0.03,-0.13) & 0.14\\\hline
Si & (3.66,0.33,-1.25) & (-0.04,-0.02, 0.00) & 0.03\\\hline
Si & (3.31,0.65,-0.98) & (-0.09,-0.05, 0.04) & 0.09\\\hline
\end{tabular}
\end{center}
\caption{The atom positions in the Si(111)-(4$\times$1)-In reconstruction
derived from the analysis of the SXRD data. The positions of the atoms
are given
in LEED coordinates, the deviations from the bulk-like positions and
the absolute values of these deviations are given in \AA{}. For the
silicon atoms an isotropic Debye-Waller factor of 0.5~\AA$^2$ and for the
indium atoms of 4.5~\AA$^2$ was used in the data analysis.}
\label{tab:pos}
\end{table}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% figure captions %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{figure}[h]
\caption{Patterson function of the electron density projected
in the surface plane calculated from the fractional-order in-plane
reflections.
The axes are scaled in LEED coordinates (1.0 corresponds to 3.84~\AA).
The distance vectors shown are 1. (0.69,0.85), 2. (1.42,0.71) and
3. (2.00,0.50). The dashed line from (0,0) to (2,1) indicates the
mirror line (along $\langle 11\overline{2}\rangle$ in bulk coordinates).}
\label{fig:pat_proj}
\end{figure}
\begin{figure}[h]
\caption{Ball and stick model of the Si(111)-(4$\times$1)-In
reconstruction in top (a) and side (b) views. Indium atoms are
drawn dark grey, silicon atoms are drawn light grey. The contributions
to the peaks of the Patterson function are shown by arrows. The
standard LEED (4$\times$1) unit cell is indicated by a dashed line.
The dashed-dotted line along $[11\overline{2}]$ indicates a mirror line.}
\label{fig:model}
\end{figure}
\begin{figure}[h]
\caption{Measured and calculated SXRD intensities:
(a) in-plane with $l=0.07$. The radii of the filled (empty) semi-circles
are proportional to the measured (calculated) intensities.
Hatched circles are scaled with a factor of 0.5.
(b) Fractional-order rod-scans. The solid line is calculated using
the model shown in Fig.~\ref{fig:model} with the coordinates
given in Table \ref{tab:pos}.
(c) Integer order rods. The dashed lines correspond to the
intensities calculated using the best fit to the model proposed in
Ref.\ \protect\onlinecite{ldm_si1114x1in_surfrevlett4}.}
\label{fig:data}
\end{figure}
\newpage
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% figures %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% Fig. 1: Patterson function %%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
~\vspace{2cm}
\begin{center}
\epsfxsize=3.5cm
\epsffile{pat_proj.eps}
\vfill
{\LARGE Fig.~\ref{fig:pat_proj}}
\end{center}
\vfill
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% Fig. 2: ball and stick model %%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{center}
~\vspace{2cm}
\epsfxsize=5cm
\epsffile{model_top.eps}
\bigskip
\epsfxsize=5cm
\epsffile{model_side.eps}
\vfill
{\LARGE Fig.~\ref{fig:model}}
\end{center}
\vfill
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% Fig. 3: data %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{center}
~\vspace{2cm}
\epsfxsize=8.4cm
\epsffile{inplane_scaled.eps}
\medskip
\epsfxsize=8.4cm
\epsffile{rods_frac.eps}
\medskip
\epsfxsize=8.4cm
\epsffile{ctrs.eps}
\vfill
{\LARGE Fig.~\ref{fig:data}}
\end{center}
\end{multicols}
\end{document}
%% end of file si111-4x1-in.tex