tarted w first two combos

[lectures/latex.git] / posic / thesis / defects.tex

\chapter{Point defects in silicon}

Given the conversion mechnism of SiC in crystalline silicon introduced in \ref{section:assumed_prec} the understanding of carbon and silicon interstitial point defects in c-Si is of great interest.
Both types of defects are examined in the following both by classical potential as well as density functional theory calculations.

In case of the classical potential calculations a simulation volume of nine silicon lattice constants in each direction is used.
Calculations are performed in an isothermal-isobaric NPT ensemble.
Coupling to the heat bath is achieved by the Berendsen thermostat with a time constant of 100 fs.
The temperature is set to zero Kelvin.
Pressure is controlled by a Berendsen barostat again using a time constant of 100 fs and a bulk modulus of 100 GPa for silicon.
To exclude surface effects periodic boundary conditions are applied.

Due to the restrictions in computer time three silicon lattice constants in each direction are considered sufficiently large enough for DFT calculations.
The ions are relaxed by a conjugate gradient method.
The cell volume and shape is allowed to change using the pressure control algorithm of Parinello and Rahman \cite{}.
Periodic boundary conditions in each direction are applied.
All point defects are calculated for the neutral charge state.

\begin{figure}[h]
\begin{center}
\includegraphics[width=9cm]{unit_cell_e.eps}
\end{center}
\caption[Insertion positions for the tetrahedral ({\color{red}$\bullet$}), hexagonal  ({\color{green}$\bullet$}), \hkl<1 0 0> dumbbell ({\color{yellow}$\bullet$}), \hkl<1 1 0> dumbbell ({\color{magenta}$\bullet$}) and bond-centered ({\color{cyan}$\bullet$}) interstitial configuration.]{Insertion positions for the tetrahedral ({\color{red}$\bullet$}), hexagonal  ({\color{green}$\bullet$}), \hkl<1 0 0> dumbbell ({\color{yellow}$\bullet$}), \hkl<1 1 0> dumbbell ({\color{magenta}$\bullet$}) and bond-centered ({\color{cyan}$\bullet$}) interstitial configuration. The black dots ({\color{black}$\bullet$}) correspond to the silicon atoms and the blue lines ({\color{blue}-}) indicate the covalent bonds of the perfect c-Si structure.}
\label{fig:defects:ins_pos}
\end{figure}

The interstitial atom positions are displayed in figure \ref{fig:defects:ins_pos}.
In seperated simulation runs the silicon or carbon atom is inserted at the
\begin{itemize}
 \item tetrahedral, $\vec{r}=(0,0,0)$, ({\color{red}$\bullet$})
 \item hexagonal, $\vec{r}=(-1/8,-1/8,1/8)$, ({\color{green}$\bullet$})
 \item nearly \hkl<1 0 0> dumbbell, $\vec{r}=(-1/4,-1/4,-1/8)$, ({\color{yellow}$\bullet$})
 \item nearly \hkl<1 1 0> dumbbell, $\vec{r}=(-1/8,-1/8,-1/4)$, ({\color{magenta}$\bullet$})
 \item bond-centered, $\vec{r}=(-1/8,-1/8,-3/8)$, ({\color{cyan}$\bullet$})
\end{itemize}
interstitial position.
For the dumbbell configurations the nearest silicon atom is displaced by $(0,0,-1/8)$ and $(-1/8,-1/8,0)$ respectively of the unit cell length to avoid too high forces.
A vacancy or a substitutional atom is realized by removing one silicon atom and switching the type of one silicon atom respectively.

From an energetic point of view the free energy of formation $E_{\text{f}}$ is suitable for the characterization of defect structures.
For defect configurations consisting of a single atom species the formation energy is defined as
\begin{equation}
E_{\text{f}}=\left(E_{\text{coh}}^{\text{defect}}
                  -E_{\text{coh}}^{\text{defect-free}}\right)N
\label{eq:defects:ef1}
\end{equation}
where $N$ and $E_{\text{coh}}^{\text{defect}}$ are the number of atoms and the cohesive energy per atom in the defect configuration and $E_{\text{coh}}^{\text{defect-free}}$ is the cohesive energy per atom of the defect-free structure.
The formation energy of defects consisting of two or more atom species is defined as
\begin{equation}
E_{\text{f}}=E-\sum_i N_i\mu_i
\label{eq:defects:ef2}
\end{equation}
where $E$ is the free energy of the interstitial system and $N_i$ and $\mu_i$ are the amount of atoms and the chemical potential of species $i$.
The chemical potential is determined by the cohesive energy of the structure of the specific type in equilibrium at zero Kelvin.
For a defect configuration of a single atom species equation \eqref{eq:defects:ef2} is equivalent to equation \eqref{eq:defects:ef1}.

\section{Silicon self-interstitials}

Point defects in silicon have been extensively studied, both experimentally and theoretically \cite{fahey89,leung99}.
Quantum-mechanical total-energy calculations are an invalueable tool to investigate the energetic and structural properties of point defects since they are experimentally difficult to assess.

The formation energies of some of the silicon self-interstitial configurations are listed in table \ref{tab:defects:si_self} for both methods used in this work as well as results obtained by former studies \cite{leung99}.
\begin{table}[h]
\begin{center}
\begin{tabular}{l c c c c c}
\hline
\hline
 & T & H & \hkl<1 0 0> DB & \hkl<1 1 0> DB & V \\
\hline
 Erhard/Albe MD & 3.40 & 4.48$^*$ & 5.42 & 4.39 & 3.13 \\
 VASP & 3.77 & 3.42 & 4.41 & 3.39 & 3.63 \\
 LDA \cite{leung99} & 3.43 & 3.31 & - & 3.31 & - \\
 GGA \cite{leung99} & 4.07 & 3.80 & - & 3.84 & - \\
\hline
\hline
\end{tabular}
\end{center}
\caption[Formation energies of silicon self-interstitials in crystalline silicon determined by classical potential molecular dynamics and density functional calculations.]{Formation energies of silicon self-interstitials in crystalline silicon determined by classical potential molecular dynamics and density functional calculations. The formation energies are given in eV. T denotes the tetrahedral, H the hexagonal, B the bond-centered and V the vacancy interstitial configuration. The dumbbell configurations are abbreviated by DB. Formation energies for unstable configurations are marked by an asterisk and determined by using the low kinetic energy configuration shortly before the relaxation into the more favorable configuration starts.}
\label{tab:defects:si_self}
\end{table}
The final configurations obtained after relaxation are presented in figure \ref{fig:defects:conf}.
\begin{figure}[h]
\begin{center}
%\hrule
%\vspace*{0.2cm}
%\begin{flushleft}
%\begin{minipage}{5cm}
%\underline{\hkl<1 1 0> dumbbell}\\
%$E_{\text{f}}=3.39\text{ eV}$\\
%\includegraphics[width=3.0cm]{si_pd_vasp/110_2333.eps}
%\end{minipage}
%\begin{minipage}{5cm}
%\underline{Hexagonal}\\
%$E_{\text{f}}=3.42\text{ eV}$\\
%\includegraphics[width=3.0cm]{si_pd_vasp/hex_2333.eps}
%\end{minipage}
%\begin{minipage}{5cm}
%\underline{Tetrahedral}\\
%$E_{\text{f}}=3.77\text{ eV}$\\
%\includegraphics[width=3.0cm]{si_pd_vasp/tet_2333.eps}
%\end{minipage}\\[0.2cm]
%\begin{minipage}{5cm}
%\underline{\hkl<1 0 0> dumbbell}\\
%$E_{\text{f}}=4.41\text{ eV}$\\
%\includegraphics[width=3.0cm]{si_pd_vasp/100_2333.eps}
%\end{minipage}
%\begin{minipage}{5cm}
%\underline{Vacancy}\\
%$E_{\text{f}}=3.63\text{ eV}$\\
%\includegraphics[width=3.0cm]{si_pd_vasp/vac_2333.eps}
%\end{minipage}
%\begin{minipage}{5cm}
%\begin{center}
%VASP\\
%calculations\\
%\end{center}
%\end{minipage}
%\end{flushleft}
%\vspace*{0.2cm}
%\hrule
\begin{flushleft}
\begin{minipage}{5cm}
\underline{Tetrahedral}\\
$E_{\text{f}}=3.40\text{ eV}$\\
\includegraphics[width=4.0cm]{si_pd_albe/tet.eps}
\end{minipage}
\begin{minipage}{10cm}
\underline{Hexagonal}\\[0.1cm]
\begin{minipage}{4cm}
$E_{\text{f}}^*=4.48\text{ eV}$\\
\includegraphics[width=4.0cm]{si_pd_albe/hex_a.eps}
\end{minipage}
\begin{minipage}{0.8cm}
\begin{center}
$\Rightarrow$
\end{center}
\end{minipage}
\begin{minipage}{4cm}
$E_{\text{f}}=3.96\text{ eV}$\\
\includegraphics[width=4.0cm]{si_pd_albe/hex.eps}
\end{minipage}
\end{minipage}\\[0.2cm]
\begin{minipage}{5cm}
\underline{\hkl<1 0 0> dumbbell}\\
$E_{\text{f}}=5.42\text{ eV}$\\
\includegraphics[width=4.0cm]{si_pd_albe/100.eps}
\end{minipage}
\begin{minipage}{5cm}
\underline{\hkl<1 1 0> dumbbell}\\
$E_{\text{f}}=4.39\text{ eV}$\\
\includegraphics[width=4.0cm]{si_pd_albe/110.eps}
\end{minipage}
\begin{minipage}{5cm}
\underline{Vacancy}\\
$E_{\text{f}}=3.13\text{ eV}$\\
\includegraphics[width=4.0cm]{si_pd_albe/vac.eps}
\end{minipage}
\end{flushleft}
%\hrule
\end{center}
\caption[Relaxed silicon self-interstitial defect configurations obtained by classical potential calculations.]{Relaxed silicon self-interstitial defect configurations obtained by classical potential calculations. The silicon atoms and the bonds (only for the interstitial atom) are illustrated by yellow spheres and blue lines.}
\label{fig:defects:conf}
\end{figure}

There are differences between the various results of the quantum-mechanical calculations but the consesus view is that the \hkl<1 1 0> dumbbell followed by the hexagonal and tetrahedral defect is the lowest in energy.
This is nicely reproduced by the DFT calculations performed in this work.

It has turned out to be very difficult to capture the results of quantum-mechanical calculations in analytical potential models.
Among the established analytical potentials only the EDIP \cite{} and Stillinger-Weber \cite{} potential reproduce the correct order in energy of the defects.
However, these potenitals show shortcomings concerning the description of other physical properties and are unable to describe the C-C and C-Si interaction.
In fact the Erhard/Albe potential calculations favor the tetrahedral defect configuration.
The hexagonal configuration is not stable opposed to results of the authors of the potential \cite{albe_sic_pot}.
In the first two pico seconds while kinetic energy is decoupled from the system the Si interstitial seems to condense at the hexagonal site.
The formation energy of 4.48 eV is determined by this low kinetic energy configuration shortly before the relaxation process starts.
The Si interstitial atom then begins to slowly move towards an energetically more favorable position very close to the tetrahedral one but slightly displaced along the three coordinate axes.
The formation energy of 3.96 eV for this type of interstitial is equal to the result for the hexagonal one in the original work \cite{albe_sic_pot}.
Obviously the authors did not carefully check the relaxed results assuming a hexagonal configuration.
In figure \ref{fig:defects:kin_si_hex} the relaxation process is shown on the basis of the kinetic energy plot.
\begin{figure}[h]
\begin{center}
\includegraphics[width=10cm]{e_kin_si_hex.ps}
\end{center}
\caption{Kinetic energy plot of the relaxation process of the hexagonal silicon self-interstitial defect simulation using the Erhard/Albe classical potential.}
\label{fig:defects:kin_si_hex}
\end{figure}
To exclude failures in the implementation of the potential or the MD code itself the hexagonal defect structure was double-checked with the PARCAS MD code \cite{}.
The same type of interstitial arises using random insertions.
In addition, variations exist in which the displacement is only along two \hkl<1 0 0> axes ($E_\text{f}=3.8\text{ eV}$) or along a single \hkl<1 0 0> axes ($E_\text{f}=3.6\text{ eV}$) successively approximating the tetdrahedral configuration and formation energy.
The existence of these local minima located near the tetrahedral configuration seems to be an artifact of the analytical potential without physical authenticity revealing basic problems of analytical potential models for describing defect structures.
However, the energy barrier is small.
Todo: Check!
Hence these artifacts should have a negligent influence in finite temperature simulations.

The bond-centered configuration is unstable and the \hkl<1 0 0> dumbbell interstitial is the most unfavorable configuration for both, the Erhard/Albe and VASP calculations.

In the case of the classical potential simulations bonds between atoms are displayed if there is an interaction according to the potential model, that is if the distance of two atoms is within the cutoff region $S_{ij}$ introduced in equation \eqref{eq:basics:fc}.
For the tetrahedral and the slightly displaced configurations four bonds to the atoms located in the center of the planes of the unit cell exist in addition to the four tetrahedral bonds.
The length of these bonds are, however, close to the cutoff range and thus are weak interactions not constituting actual chemical bonds.
The same applies to the bonds between the interstitial and the upper two atoms in the \hkl<1 1 0> dumbbell configuration.

A more detailed description of the chemical bonding is achieved by quantum-mechanical calculations by investigating the accumulation of negative charge between the nuclei.
Todo: Plot the electron density for these types of defect to derive conclusions of existing bonds?

\section{Carbon related point defects}

Carbon is a common and technologically important impurity in silicon.
Concentrations as high as $10^{18}\text{ cm}^{-3}$ occur in Czochralski-grown silicon samples.
It is well established that carbon and other isovalent impurities prefer to dissolve substitutionally in silicon.
However, radiation damage can generate carbon interstitials \cite{watkins76} which have enough mobility at room temeprature to migrate and form defect complexes.

Formation energies of the most common carbon point defects in crystalline silicon are summarized in table \ref{tab:defects:c_ints} and the relaxed configurations obtained by classical potential calculations visualized in figure \ref{fig:defects:c_conf}.
The type of reservoir of the carbon impurity to determine the formation energy of the defect was chosen to be SiC.
This is consistent with the methods used in the articles \cite{tersoff90,dal_pino93}, which the results are compared to in the following.
Hence, the chemical potential of silicon and carbon is determined by the cohesive energy of silicon and silicon carbide.
\begin{table}[h]
\begin{center}
\begin{tabular}{l c c c c c c}
\hline
\hline
 & T & H & \hkl<1 0 0> DB & \hkl<1 1 0> DB & S & B \\
\hline
 Erhard/Albe MD & 6.09 & 9.05$^*$ & 3.88 & 5.18 & 0.75 & 5.59$^*$ \\
 %VASP & unstable & unstable & 3.15 & 3.60 & 1.39 & 4.10 \\
 VASP & unstable & unstable & 3.72 & 4.16 & 1.95 & 4.66 \\
 Tersoff \cite{tersoff90} & 3.8 & 6.7 & 4.6 & 5.9 & 1.6 & 5.3 \\
 ab initio & - & - & x & - & 1.89 \cite{dal_pino93} & x+2.1 \cite{capaz94} \\
\hline
\hline
\end{tabular}
\end{center}
\caption[Formation energies of carbon point defects in crystalline silicon determined by classical potential molecular dynamics and density functional calculations.]{Formation energies of carbon point defects in crystalline silicon determined by classical potential molecular dynamics and density functional calculations. The formation energies are given in eV. T denotes the tetrahedral, H the hexagonal, B the bond-centered and S the substitutional interstitial configuration. The dumbbell configurations are abbreviated by DB.  Formation energies for unstable configurations are marked by an asterisk and determined by using the low kinetic energy configuration shortly before the relaxation into the more favorable configuration starts.}
\label{tab:defects:c_ints}
\end{table}
\begin{figure}[h]
\begin{center}
\begin{flushleft}
\begin{minipage}{4cm}
\underline{Hexagonal}\\
$E_{\text{f}}^*=9.05\text{ eV}$\\
\includegraphics[width=4.0cm]{c_pd_albe/hex.eps}
\end{minipage}
\begin{minipage}{0.8cm}
\begin{center}
$\Rightarrow$
\end{center}
\end{minipage}
\begin{minipage}{4cm}
\underline{\hkl<1 0 0>}\\
$E_{\text{f}}=3.88\text{ eV}$\\
\includegraphics[width=4.0cm]{c_pd_albe/100.eps}
\end{minipage}
\begin{minipage}{0.5cm}
\hfill
\end{minipage}
\begin{minipage}{5cm}
\underline{Tetrahedral}\\
$E_{\text{f}}=6.09\text{ eV}$\\
\includegraphics[width=4.0cm]{c_pd_albe/tet.eps}
\end{minipage}\\[0.2cm]
\begin{minipage}{4cm}
\underline{Bond-centered}\\
$E_{\text{f}}^*=5.59\text{ eV}$\\
\includegraphics[width=4.0cm]{c_pd_albe/bc.eps}
\end{minipage}
\begin{minipage}{0.8cm}
\begin{center}
$\Rightarrow$
\end{center}
\end{minipage}
\begin{minipage}{4cm}
\underline{\hkl<1 1 0> dumbbell}\\
$E_{\text{f}}=5.18\text{ eV}$\\
\includegraphics[width=4.0cm]{c_pd_albe/110.eps}
\end{minipage}
\begin{minipage}{0.5cm}
\hfill
\end{minipage}
\begin{minipage}{5cm}
\underline{Substitutional}\\
$E_{\text{f}}=0.75\text{ eV}$\\
\includegraphics[width=4.0cm]{c_pd_albe/sub.eps}
\end{minipage}
\end{flushleft}
\end{center}
\caption[Relaxed carbon point defect configurations obtained by classical potential calculations.]{Relaxed carbon point defect configurations obtained by classical potential calculations. The silicon/carbon atoms and the bonds (only for the interstitial atom) are illustrated by yellow/grey spheres and blue lines.}
\label{fig:defects:c_conf}
\end{figure}

Substitutional carbon in silicon is found to be the lowest configuration in energy for all potential models.
An experiemntal value of the formation energy of substitutional carbon was determined by a fit to solubility data yielding a concentration of $3.5 \times 10^{24} \exp{(-2.3\text{ eV}/k_{\text{B}}T)} \text{ cm}^{-3}$ \cite{bean71}.
However, there is no particular reason for treating the prefactor as a free parameter in the fit to the experimental data.
It is simply given by the atomic density of pure silicon, which is $5\times 10^{22}\text{ cm}^{-3}$.
Tersoff \cite{tersoff90} and Dal Pino et al. \cite{dal_pino93} pointed out that by combining this prefactor with the calculated values for the energy of formation ranging from 1.6 to 1.89 eV an excellent agreement with the experimental solubility data within the entire temeprature range of the experiment is obtained.
This reinterpretation of the solubility data, first proposed by Tersoff and later on reinforced by Dal Pino et al. is in good agreement with the results of the quantum-mechanical calculations performed in this work.
Unfortunately the Erhard/Albe potential undervalues the formation energy roughly by a factor of two.

Except for Tersoff's tedrahedral configuration results the \hkl<1 0 0> dumbbell is the energetically most favorable interstital configuration.
The low energy of formation for the tetrahedral interstitial in the case of the Tersoff potential is believed to be an artifact of the abrupt cutoff set to 2.5 \AA{} (see ref. 11 and 13 in \cite{tersoff90}) and the real formation energy is, thus, supposed to be located between 3 and 10 eV.
Keeping these considerations in mind, the \hkl<1 0 0> dumbbell is the most favorable interstitial configuration for all interaction models.
In addition to the theoretical results compared to in table \ref{tab:defects:c_ints} there is experimental evidence of the existence of this configuration \cite{watkins76}.
It is frequently generated in the classical potential simulation runs in which carbon is inserted at random positions in the c-Si matrix.
In quantum-mechanical simulations the unstable tetrahedral and hexagonal configurations undergo a relaxation into the \hkl<1 0 0> dumbbell configuration.
Thus, this configuration is of great importance and discussed in more detail in section \ref{subsection:100db}.

The highest energy is observed for the hexagonal interstitial configuration using classical potentials.
Quantum-mechanical calculations reveal this configuration to be unstable, which is also reproduced by the Erhard/Albe potential.
In both cases a relaxation towards the \hkl<1 0 0> dumbbell configuration is observed.
In fact the stability of the hexagonal interstitial could not be reproduced in simulations performed in this work using the unmodifed Tersoff potential parameters.
Unfortunately, apart from the modified parameters, no more conditions specifying the relaxation process are given in Tersoff's study on carbon point defects in silicon \cite{tersoff90}.

The tetrahedral is the second most unfavorable interstitial configuration using classical potentials and keeping in mind the abrupt cutoff effect in the case of the Tersoff potential as discussed earlier.
Again, quantum-mechanical results reveal this configuration unstable.
The fact that the tetrahedral and hexagonal configurations are the two most unstable configurations in classical potential calculations and, thus, are less likely to arise in MD simulations acts in concert with the fact that these configurations are found to be unstable in the more accurate quantum-mechanical calculations.

Just as for the Si self-interstitial a carbon \hkl<1 1 0> dumbbell configuration exists.
For the Erhard/Albe potential the formation energy is situated in the same order as found by quantum-mechanical results.
Similar structures arise in both types of simulations with the silicon and carbon atom sharing a silicon lattice site aligned along \hkl<1 1 0> where the carbon atom is localized slightly closer to the next nearest silicon atom located in the opposite direction to the site-sharing silicon atom even forming a bond to the next but one silicon atom in this direction.

The bond-centered configuration is unstable for the Erhard/Albe potential.
The system moves into the \hkl<1 1 0> interstitial configuration.
This, like in the hexagonal case, is also true for the unmodified Tersoff potential and the given relaxation conditions.
Quantum-mechanical results of this configuration are discussed in more detail in section \ref{subsection:bc}.
In another ab inito study Capaz et al. \cite{capaz94} determined this configuration as an intermediate saddle point structure of a possible migration path, which is 2.1 eV higher than the \hkl<1 0 0> dumbbell configuration.
In calculations performed in this work the bond-centered configuration in fact is a real local minimum and an energy barrier is needed to reach this configuration starting from the \hkl<1 0 0> dumbbell configuration, which is discussed in section \ref{subsection:100mig}.

\subsection[\hkl<1 0 0> dumbbell interstitial configuration]{\boldmath\hkl<1 0 0> dumbbell interstitial configuration}
\label{subsection:100db}

As the \hkl<1 0 0> dumbbell interstitial is the lowest configuration in energy it is the most probable hence important interstitial configuration of carbon in silicon.
It was first identified by infra-red (IR) spectroscopy \cite{bean70} and later on by electron paramagnetic resonance (EPR) spectroscopy \cite{watkins76}.

Figure \ref{fig:defects:100db_cmp} schematically shows the \hkl<1 0 0> dumbbell structure and table \ref{tab:defects:100db_cmp} lists the details of the atomic displacements, distances and bond angles obtained by analytical potential and quantum-mechanical calculations.
For comparison, the obtained structures for both methods visualized out of the atomic position data are presented in figure \ref{fig:defects:100db_vis_cmp}.
\begin{figure}[h]
\begin{center}
\includegraphics[width=12cm]{100-c-si-db_cmp.eps}
\end{center}
\caption[Sketch of the \hkl<1 0 0> dumbbell structure.]{Sketch of the \hkl<1 0 0> dumbbell structure. Atomic displacements, distances and bond angles are listed in table \ref{tab:defects:100db_cmp}.}
\label{fig:defects:100db_cmp}
\end{figure}
%
\begin{table}[h]
\begin{center}
Displacements\\
\begin{tabular}{l c c c c c c c c c}
\hline
\hline
 & & & & \multicolumn{3}{c}{Atom 2} & \multicolumn{3}{c}{Atom 3} \\
 & $a$ & $b$ & $|a|+|b|$ & $\Delta x$ & $\Delta y$ & $\Delta z$ & $\Delta x$ & $\Delta y$ & $\Delta z$ \\
\hline
Erhard/Albe & 0.084 & -0.091 & 0.175 & -0.015 & -0.015 & -0.031 & -0.014 & 0.014 & 0.020 \\
VASP & 0.109 & -0.065 & 0.174 & -0.011 & -0.011 & -0.024 & -0.014 & 0.014 & 0.025 \\
\hline
\hline
\end{tabular}\\[0.5cm]
\end{center}
\begin{center}
Distances\\
\begin{tabular}{l c c c c c c c c r}
\hline
\hline
 & $r(1C)$ & $r(2C)$ & $r(3C)$ & $r(12)$ & $r(13)$ & $r(34)$ & $r(23)$ & $r(25)$ & $a_{\text{Si}}^{\text{equi}}$\\
\hline
Erhard/Albe & 0.175 & 0.329 & 0.186 & 0.226 & 0.300 & 0.343 & 0.423 & 0.425 & 0.543 \\
VASP & 0.174 & 0.341 & 0.182 & 0.229 & 0.286 & 0.347 & 0.422 & 0.417 & 0.548 \\
\hline
\hline
\end{tabular}\\[0.5cm]
\end{center}
\begin{center}
Angles\\
\begin{tabular}{l c c c c }
\hline
\hline
 & $\theta_1$ & $\theta_2$ & $\theta_3$ & $\theta_4$ \\
\hline
Erhard/Albe & 140.2 & 109.9 & 134.4 & 112.8 \\
VASP & 130.7 & 114.4 & 146.0 & 107.0 \\
\hline
\hline
\end{tabular}\\[0.5cm]
\end{center}
\caption[Atomic displacements, distances and bond angles of the \hkl<1 0 0> dumbbell structure obtained by the Erhard/Albe potential and VASP calculations.]{Atomic displacements, distances and bond angles of the \hkl<1 0 0> dumbbell structure obtained by the Erhard/Albe potential and VASP calculations. The displacements and distances are given in nm and the angles are given in degrees. Displacements, distances and angles are schematically displayed in figure \ref{fig:defects:100db_cmp}. In addition, the equilibrium lattice constant for crystalline silicon is listed.}
\label{tab:defects:100db_cmp}
\end{table}
\begin{figure}[h]
\begin{center}
\begin{minipage}{6cm}
\begin{center}
\underline{Erhard/Albe}
\includegraphics[width=5cm]{c_pd_albe/100_cmp.eps}
\end{center}
\end{minipage}
\begin{minipage}{6cm}
\begin{center}
\underline{VASP}
\includegraphics[width=5cm]{c_pd_vasp/100_cmp.eps}
\end{center}
\end{minipage}
\end{center}
\caption{Comparison of the visualized \hkl<1 0 0> dumbbel structures obtained by Erhard/Albe potential and VASP calculations.}
\label{fig:defects:100db_vis_cmp}
\end{figure}
\begin{figure}[h]
\begin{center}
\includegraphics[height=10cm]{c_pd_vasp/eden.eps}
\includegraphics[height=12cm]{c_pd_vasp/100_2333_ksl.ps}
\end{center}
\caption[Charge density isosurface and Kohn-Sham levels of the C \hkl<1 0 0> dumbbell structure obtained by VASP calculations.]{Charge density isosurface and Kohn-Sham levels of the C \hkl<1 0 0> dumbbell structure obtained by VASP calculations. Yellow and grey spheres correspond to silicon and carbon atoms. The blue surface is the charge density isosurface. In the energy level diagram red and green lines and dots mark occupied and unoccupied states.}
\label{img:defects:charge_den_and_ksl}
\end{figure}
The silicon atom numbered '1' and the C atom compose the dumbbell structure.
They share the lattice site which is indicated by the dashed red circle and which they are displaced from by length $a$ and $b$ respectively.
The atoms no longer have four tetrahedral bonds to the silicon atoms located on the alternating opposite edges of the cube.
Instead, each of the dumbbell atoms forms threefold coordinated bonds, which are located in a plane.
One bond is formed to the other dumbbell atom.
The other two bonds are bonds to the two silicon edge atoms located in the opposite direction of the dumbbell atom.
The distance of the two dumbbell atoms is almost the same for both types of calculations.
However, in the case of the VASP calculation, the dumbbell structure is pushed upwards compared to the Erhard/Albe results.
This is easily identified by comparing the values for $a$ and $b$ and the two structures in figure \ref{fig:defects:100db_vis_cmp}.
Thus, the angles of bonds of the silicon dumbbell atom ($\theta_1$ and $\theta_2$) are closer to $120^{\circ}$ signifying the predominance of $sp^2$ hybridization.
On the other hand, the carbon atom forms an almost collinear bond ($\theta_3$) with the two silicon edge atoms implying the predominance of $sp$ bonding.
This is supported by the image of the charge density isosurface in figure \ref{img:defects:charge_den_and_ksl}.
The two lower Si atoms are $sp^3$ hybridised and form $\sigma$ bonds to the silicon dumbbell atom.
The same is true for the upper two silicon atoms and the C dumbbell atom.
In addition the dumbbell atoms form $\pi$ bonds.
However, due to the increased electronegativity of the carbon atom the electron density is attracted by and thus localized around the carbon atom.
In the same figure the Kohn-Sham levels are shown.
There is no magnetization density.
An acceptor level arises at approximately $E_v+0.35\text{ eV}$ while a band gap of about 0.75 eV can be estimated from the Kohn-Sham level diagram for plain silicon.

\subsection{Bond-centered interstitial configuration}
\label{subsection:bc}

\begin{figure}[h]
\begin{center}
\begin{minipage}{8cm}
\includegraphics[width=8cm]{c_pd_vasp/bc_2333.eps}\\
\hrule
\vspace*{0.2cm}
\includegraphics[width=8cm]{c_100_mig_vasp/im_spin_diff.eps}
\end{minipage}
\begin{minipage}{7cm}
\includegraphics[width=7cm]{c_pd_vasp/bc_2333_ksl.ps}
\end{minipage}
\end{center}
\caption[Structure, charge density isosurface and Kohn-Sham level diagram of the bond-centered interstitial configuration.]{Structure, charge density isosurface and Kohn-Sham level diagram of the bond-centered interstitial configuration. Gray, green and blue surfaces mark the charge density of spin up, spin down and the resulting spin up electrons in the charge density isosurface, in which the carbon atom is represented by a red sphere. In the energy level diagram red and green lines mark occupied and unoccupied states.}
\label{img:defects:bc_conf}
\end{figure}
In the bond-centerd insterstitial configuration the interstitial atom is located inbetween two next neighboured silicon atoms forming linear bonds.
In former studies this configuration is found to be an intermediate saddle point configuration determining the migration barrier of one possibe migration path of a \hkl<1 0 0> dumbbel configuration into an equivalent one \cite{capaz94}.
This is in agreement with results of the Erhard/Albe potential simulations which reveal this configuration to be unstable relaxing into the \hkl<1 1 0> configuration.
However, this fact could not be reproduced by spin polarized VASP calculations performed in this work.
Present results suggest this configuration to be a real local minimum.
In fact, an additional barrier has to be passed to reach this configuration starting from the \hkl<1 0 0> interstitital configuration, which is investigated in section \ref{subsection:100mig}.
After slightly displacing the carbon atom along the \hkl<1 0 0> (equivalent to a displacement along \hkl<0 1 0>), \hkl<0 0 1>, \hkl<0 0 -1> and \hkl<1 -1 0> direction the resulting structures relax back into the bond-centered configuration.
As we will see in later migration simulations the same would happen to structures where the carbon atom is displaced along the migration direction, which approximately is the \hkl<1 1 0> direction.
These relaxations indicate that the bond-cenetered configuration is a real local minimum instead of an assumed saddle point configuration.
Figure \ref{img:defects:bc_conf} shows the structure, the charge density isosurface and the Kohn-Sham levels of the bond-centered configuration.
The linear bonds of the carbon atom to the two silicon atoms indicate the $sp$ hybridization of the carbon atom.
Two electrons participate to the linear $\sigma$ bonds with the silicon neighbours.
The other two electrons constitute the $2p^2$ orbitals resulting in a net magnetization.
This is supported by the charge density isosurface and the Kohn-Sham levels in figure \ref{img:defects:bc_conf}.
The blue torus, reinforcing the assumption of the p orbital, illustrates the resulting spin up electron density.
In addition, the energy level diagram shows a net amount of two spin up electrons.

\section[Migration of the carbon \hkl<1 0 0> interstitial]{\boldmath Migration of the carbon \hkl<1 0 0> interstitial}
\label{subsection:100mig}

In the following the problem of interstitial carbon migration in silicon is considered.
Since the carbon \hkl<1 0 0> dumbbell interstitial is the most probable hence most important configuration the migration simulations focus on this defect.

\begin{figure}[h]
\begin{center}
\begin{minipage}{15cm}
\underline{\hkl<0 0 -1> $\rightarrow$ \hkl<0 0 1>}\\
\begin{minipage}{4.5cm}
\includegraphics[width=4.5cm]{c_pd_vasp/100_2333.eps}
\end{minipage}
\begin{minipage}{0.5cm}
$\rightarrow$
\end{minipage}
\begin{minipage}{4.5cm}
\includegraphics[width=4.5cm]{c_pd_vasp/bc_2333.eps}
\end{minipage}
\begin{minipage}{0.5cm}
$\rightarrow$
\end{minipage}
\begin{minipage}{4.5cm}
\includegraphics[width=4.5cm]{c_pd_vasp/100_next_2333.eps}
\end{minipage}
\end{minipage}\\
\begin{minipage}{15cm}
\underline{\hkl<0 0 -1> $\rightarrow$ \hkl<0 -1 0>}\\
\begin{minipage}{4.5cm}
\includegraphics[width=4.5cm]{c_pd_vasp/100_2333.eps}
\end{minipage}
\begin{minipage}{0.5cm}
$\rightarrow$
\end{minipage}
\begin{minipage}{4.5cm}
\includegraphics[width=4.5cm]{c_pd_vasp/00-1-0-10_2333.eps}
\end{minipage}
\begin{minipage}{0.5cm}
$\rightarrow$
\end{minipage}
\begin{minipage}{4.5cm}
\includegraphics[width=4.5cm]{c_pd_vasp/0-10_2333.eps}
\end{minipage}
\end{minipage}\\
\begin{minipage}{15cm}
\underline{\hkl<0 0 -1> $\rightarrow$ \hkl<0 -1 0> (in place)}\\
\begin{minipage}{4.5cm}
\includegraphics[width=4.5cm]{c_pd_vasp/100_2333.eps}
\end{minipage}
\begin{minipage}{0.5cm}
$\rightarrow$
\end{minipage}
\begin{minipage}{4.5cm}
\includegraphics[width=4.5cm]{c_pd_vasp/00-1_ip0-10_2333.eps}
\end{minipage}
\begin{minipage}{0.5cm}
$\rightarrow$
\end{minipage}
\begin{minipage}{4.5cm}
\includegraphics[width=4.5cm]{c_pd_vasp/0-10_ip_2333.eps}
\end{minipage}
\end{minipage}
\end{center}
\caption{Migration pathways of the carbon \hkl<1 0 0> interstitial dumbbell in silicon.}
\label{img:defects:c_mig_path}
\end{figure}
Three different migration paths are accounted in this work, which are shown in figure \ref{img:defects:c_mig_path}.
The first migration investigated is a transition of a \hkl<0 0 -1> into a \hkl<0 0 1> dumbbell interstitial configuration.
During this migration the carbon atom is changing its silicon dumbbell partner.
The new partner is the one located at $\frac{a}{4}\hkl<1 1 -1>$ relative to the initial one.
Two of the three bonds to the next neighboured silicon atoms are preserved while the breaking of the third bond and the accompanying formation of a new bond is observed.
The carbon atom resides in the \hkl(1 1 0) plane.
This transition involves an intermediate bond-centerd configuration.
Results discussed in \ref{subsection:bc} indicate, that the bond-ceneterd configuration is a real local minimum.
Thus, the \hkl<0 0 -1> to \hkl<0 0 1> migration can be thought of a two-step mechanism in which the intermediate bond-cenetered configuration constitutes a metastable configuration.
Due to symmetry it is enough to consider the transition from the bond-centered to the \hkl<1 0 0> configuration or vice versa.
In the second path, the carbon atom is changing its silicon partner atom as in path one.
However, the trajectory of the carbon atom is no longer proceeding in the \hkl(1 1 0) plane.
The orientation of the new dumbbell configuration is transformed from \hkl<0 0 -1> to \hkl<0 -1 0>.
Again one bond is broken while another one is formed.
As a last migration path, the defect is only changing its orientation.
Thus, it is not responsible for long-range migration.
The silicon dumbbell partner remains the same.
The bond to the face-centered silicon atom at the bottom of the unit cell breaks and a new one is formed to the face-centered atom at the forefront of the unit cell.
Todo: \hkl<1 1 0> to \hkl<1 0 0> and bond-centerd configuration (in progress).
Todo: \hkl<1 1 0> to \hkl<0 -1 0> (rotation of the DB, in progress).
Todo: Comparison with classical potential simulations or explanation to only focus on ab initio calculations.

Since the starting and final structure, which are both local minima of the potential energy surface, are known, the aim is to find the minimum energy path from one local minimum to the other one.
One method to find a minimum energy path is to move the diffusing atom stepwise from the starting to the final position and only allow relaxation in the plane perpendicular to the direction of the vector connecting its starting and final position.
No constraints are applied to the remaining atoms in order to allow relaxation of the surrounding lattice.
To prevent the remaining lattice to migrate according to the displacement of the defect an atom far away from the defect region is fixed in all three coordinate directions.
However, it turned out, that this method tremendously failed applying it to the present migration pathways and structures.
Abrupt changes in structure and free energy occured among relaxed structures of two successive displacement steps.
For some structures even the expected final configurations were never obtained.
Thus, the method mentioned above was adjusted adding further constraints in order to obtain smooth transitions, either in energy as well as structure is concerned.
In this new method all atoms are stepwise displaced towards their final positions.
Relaxation of each individual atom is only allowed in the plane perpendicular to the last individual displacement vector.
The modifications used to add this feature to the VASP code and a short instruction on how to use it can be found in appendix \ref{app:patch_vasp}.
Due to these constraints obtained activation energies can effectively be higher.
Todo: To refine the migration barrier one has to find the saddle point structure and recalculate the free energy of this configuration with a reduced set of constraints.

\begin{figure}[h]
\begin{center}
\includegraphics[width=13cm]{im_00-1_nosym_sp_fullct_thesis.ps}\\[1.5cm]
\begin{picture}(0,0)(150,0)
\includegraphics[width=2.5cm]{vasp_mig/00-1.eps}
\end{picture}
\begin{picture}(0,0)(-10,0)
\includegraphics[width=2.5cm]{vasp_mig/bc_00-1_sp.eps}
\end{picture}
\begin{picture}(0,0)(-120,0)
\includegraphics[width=2.5cm]{vasp_mig/bc.eps}
\end{picture}
\begin{picture}(0,0)(25,20)
\includegraphics[width=2.5cm]{110_arrow.eps}
\end{picture}
\begin{picture}(0,0)(200,0)
\includegraphics[height=2.2cm]{001_arrow.eps}
\end{picture}
\end{center}
\caption[Migration barrier and structures of the \hkl<0 0 -1> dumbbell (left) to bond-centered (right) transition.]{Migration barrier and structures of the \hkl<0 0 -1> dumbbell (left) to bond-centered (right) transition. Bonds of the carbon atoms are illustrated by blue lines.}
\label{fig:defects:00-1_001_mig}
\end{figure}
In figure \ref{fig:defects:00-1_001_mig} results of the \hkl<0 0 -1> to \hkl<0 0 1> migration fully described by the migration of the \hkl<0 0 -1> dumbbell to the bond-ceneterd configuration is displayed.
To reach the bond-centered configuration, which is 0.94 eV higher in energy than the \hkl<0 0 -1> dumbbell configuration, an energy barrier of approximately 1.2 eV, given by the saddle point structure at a displacement of 60 \%, has to be passed.
This amount of energy is needed to break the bond of the carbon atom to the silicon atom at the bottom left.
In a second process 0.25 eV of energy are needed for the system to revert into a \hkl<1 0 0> configuration.

\begin{figure}[h]
\begin{center}
\includegraphics[width=13cm]{vasp_mig/00-1_0-10_nosym_sp_fullct.ps}\\[1.6cm]
\begin{picture}(0,0)(140,0)
\includegraphics[width=2.5cm]{vasp_mig/00-1_a.eps}
\end{picture}
\begin{picture}(0,0)(20,0)
\includegraphics[width=2.5cm]{vasp_mig/00-1_0-10_sp.eps}
\end{picture}
\begin{picture}(0,0)(-120,0)
\includegraphics[width=2.5cm]{vasp_mig/0-10.eps}
\end{picture}
\begin{picture}(0,0)(25,20)
\includegraphics[width=2.5cm]{100_arrow.eps}
\end{picture}
\begin{picture}(0,0)(200,0)
\includegraphics[height=2.2cm]{001_arrow.eps}
\end{picture}
\end{center}
\caption[Migration barrier and structures of the \hkl<0 0 -1> dumbbell (left) to the \hkl<0 -1 0> dumbbell (right) transition.]{Migration barrier and structures of the \hkl<0 0 -1> dumbbell (left) to the \hkl<0 -1 0> dumbbell (right) transition. Bonds of the carbon atoms are illustrated by blue lines.}
\label{fig:defects:00-1_0-10_mig}
\end{figure}
Figure \ref{fig:defects:00-1_0-10_mig} shows the migration barrier and structures of the \hkl<0 0 -1> to \hkl<0 -1 0> dumbbell transition.
The resulting migration barrier of approximately 0.9 eV is very close to the experimentally obtained values of 0.73 \cite{song90} and 0.87 eV \cite{tipping87}.

\begin{figure}[h]
\begin{center}
\includegraphics[width=13cm]{vasp_mig/00-1_ip0-10_nosym_sp_fullct.ps}\\[1.8cm]
\begin{picture}(0,0)(140,0)
\includegraphics[width=2.2cm]{vasp_mig/00-1_b.eps}
\end{picture}
\begin{picture}(0,0)(20,0)
\includegraphics[width=2.2cm]{vasp_mig/00-1_ip0-10_sp.eps}
\end{picture}
\begin{picture}(0,0)(-120,0)
\includegraphics[width=2.2cm]{vasp_mig/0-10_b.eps}
\end{picture}
\begin{picture}(0,0)(25,20)
\includegraphics[width=2.5cm]{100_arrow.eps}
\end{picture}
\begin{picture}(0,0)(200,0)
\includegraphics[height=2.2cm]{001_arrow.eps}
\end{picture}
\end{center}
\caption[Migration barrier and structures of the \hkl<0 0 -1> dumbbell (left) to the \hkl<0 -1 0> dumbbell (right) transition in place.]{Migration barrier and structures of the \hkl<0 0 -1> dumbbell (left) to the \hkl<0 -1 0> dumbbell (right) transition in place. Bonds of the carbon atoms are illustrated by blue lines.}
\label{fig:defects:00-1_0-10_ip_mig}
\end{figure}
The third migration path in which the dumbbell is changing its orientation is shown in figure \ref{fig:defects:00-1_0-10_ip_mig}.
An energy barrier of roughly 1.2 eV is observed.
Experimentally measured activation energies for reorientation range from 0.77 eV to 0.88 eV \cite{watkins76,song90}.
Thus, this pathway is more likely to be composed of two consecutive steps of the second path.

Since the activation energy of the first and last migration path is much greater than the experimental value, the second path is identified to be responsible as a migration path for the most likely carbon interstitial in silicon explaining both, annealing and reorientation experiments.
The activation energy of roughly 0.9 eV nicely compares to experimental values.
The theoretical description performed in this work is improved compared to a former study \cite{capaz94}, which underestimates the experimental value by 35 \%.
In addition the bond-ceneterd configuration, for which spin polarized calculations are necessary, is found to be a real local minimum instead of a saddle point configuration.

\section{Combination of point defects}

The structural and energetic properties of combinations of point defects are investigated in the following.
The focus is on combinations of the \hkl<0 0 -1> dumbbell interstitial with a second defect.
The second defect is either another \hkl<1 0 0>-type interstitial occupying different orientations, a vacany or a substitutional carbon atom.
Several distances of the two defects are examined.
Investigations are restricted to quantum-mechanical calculations.
\begin{figure}[h]
\begin{center}
\begin{minipage}{7.5cm}
\includegraphics[width=7cm]{comb_pos.eps}
\end{minipage}
\begin{minipage}{6.0cm}
\underline{Positions given in $a_{\text{Si}}$}\\[0.3cm]
Initial interstitial I: $\frac{1}{4}\hkl<1 1 1>$\\
Relative silicon neighbour positions:
\begin{enumerate}
 \item $\frac{1}{4}\hkl<1 1 -1>$, $\frac{1}{4}\hkl<-1 -1 -1>$
 \item $\frac{1}{2}\hkl<1 0 1>$, $\frac{1}{2}\hkl<0 1 -1>$,\\[0.2cm]
       $\frac{1}{2}\hkl<0 -1 -1>$, $\frac{1}{2}\hkl<-1 0 -1>$
 \item $\frac{1}{4}\hkl<1 -1 1>$, $\frac{1}{4}\hkl<-1 1 1>$
 \item $\frac{1}{4}\hkl<-1 1 -3>$, $\frac{1}{4}\hkl<1 -1 -3>$
 \item $\frac{1}{2}\hkl<-1 -1 0>$, $\frac{1}{2}\hkl<1 1 0>$
\end{enumerate}
\end{minipage}\\
\begin{picture}(0,0)(190,20)
\includegraphics[width=2.3cm]{100_arrow.eps}
\end{picture}
\begin{picture}(0,0)(220,0)
\includegraphics[height=2.2cm]{001_arrow.eps}
\end{picture}
\end{center}
\caption[\hkl<0 0 -1> dumbbell interstitial defect and positions of next neighboured silicon atoms used for the second defect.]{\hkl<0 0 -1> dumbbell interstitial defect and positions of next neighboured silicon atoms used for the second defect. Two possibilities exist for red numbered atoms and four possibilities exist for blue numbered atoms.}
\label{fig:defects:pos_of_comb}
\end{figure}
\begin{table}[h]
\begin{center}
\begin{tabular}{l c c c c c}
\hline
\hline
 & 1 & 2 & 3 & 4 & 5 \\
 \hline
 \hkl<0 0 -1> & {\color{red}-0.08} & -1.15 & {\color{red}-0.08} & 0.04 & -1.66\\
 \hkl<0 0 1> & 0.34 & 0.004 & -2.05 & 0.26 & -1.53\\
 \hkl<0 -1 0> & {\color{orange}-2.39} & -2.16 & {\color{green}-0.10} & {\color{blue}-0.27} & {\color{magenta}-1.88}\\
 \hkl<0 1 0> & {\color{cyan}-2.25} & -0.36 & {\color{cyan}-2.25} & {\color{purple}-0.12} & {\color{violet}-1.38}\\
 \hkl<-1 0 0> & {\color{orange}-2.39} & -1.90 & {\color{cyan}-2.25} & {\color{purple}-0.12} & {\color{magenta}-1.88}\\
 \hkl<1 0 0> & {\color{cyan}-2.25} & -0.17 & {\color{green}-0.10} & {\color{blue}-0.27} & {\color{violet}-1.38} \\
 \hline
 C substitutional (C$_{\text{S}}$) & 0.26 & -0.51 & -0.93 & -0.15 & 0.49 \\
 Vacancy & -5.39 ($\rightarrow$ C$_{\text{S}}$) & -0.59 & -3.14 & -0.54 & -0.50 \\
\hline
\hline
\end{tabular}
\end{center}
\caption[Energetic results of defect combinations.]{Energetic results of defect combinations. The given energies in eV are defined by equation \eqref{eq:defects:e_of_comb}. Equivalent configurations are marked by identical colors. The first column lists the types of the second defect combined with the initial \hkl<0 0 -1> dumbbell interstitial. The position index of the second defect is given in the first row according to figure \ref{fig:defects:pos_of_comb}.}
\label{tab:defects:e_of_comb}
\end{table}
Figure \ref{fig:defects:pos_of_comb} shows the initial \hkl<0 0 -1> dumbbell interstitial defect and the positions of next neighboured silicon atoms used for the second defect.
Table \ref{tab:defects:e_of_comb} summarizes energetic results obtained after relaxation of the defect combinations.
The energy of interest $E_{\text{b}}$ is defined to be
\begin{equation}
E_{\text{b}}=
E_{\text{f}}^{\text{defect combination}}-
E_{\text{f}}^{\text{C \hkl<0 0 -1> dumbbell}}-
E_{\text{f}}^{\text{2nd defect}}
\label{eq:defects:e_of_comb}
\end{equation}
with $E_{\text{f}}^{\text{defect combination}}$ being the formation energy of the defect combination, $E_{\text{f}}^{\text{C \hkl<0 0 -1> dumbbell}}$ being the formation energy of the C \hkl<0 0 -1> dumbbell interstitial defect and $E_{\text{f}}^{\text{2nd defect}}$ being the formation energy of the second defect.
For defects far away from each other the formation energy of the defect combination should approximately become the sum of the formation energies of the individual defects without an interaction resulting in $E_{\text{b}}=0$.
Thus, $E_{\text{b}}$ can be best thought of a binding energy, which is required to bring the defects to infinite separation.
In fact, further \hkl<0 0 -1> dumbbell interstitials created at position $\frac{a_{\text{Si}}}{2}\hkl<3 2 3>$ ($\approx 10.2$ \AA) and $\frac{a_{\text{Si}}}{2}\hkl<2 3 2>$ ($\approx 12.8$ \AA) relative to the initial one result in energies as low as -0.19 eV and -0.12 eV.
There is still a low interaction which is due to the equal orientation of the defects.
By changing the orientation of the second dumbbell interstitial to the \hkl<0 -1 0>-type the interaction is even mor reduced, which results in an energy of $E_{\text{b}}=...\text{ eV}$ for a distance of $\frac{a_{\text{Si}}}{2}\hkl<2 3 2>$, which is the maximum that can be reached due to periodic boundary conditions.
Configurations wih energies greater than zero are energetically unfavorable and expose a repulsive interaction.
These configurations are unlikely to arise or to persist for non-zero temperatures.
Energies below zero indicate configurations favored compared to configurations in which these point defects are separated far away from each other.

Investigating the first part of table \ref{tab:defects:e_of_comb}, namely the combinations with another \hkl<1 0 0>-type interstitial, most of the combinations result in energies below zero.
Surprisingly the most favorable configurations are the ones with the second defect created at the very next silicon neighbour and a change in orientation compared to the initial one.
This leads to the conclusion that an agglomeration of C-Si dumbbell interstitials as proposed by the precipitation model introduced in section \ref{section:assumed_prec} is indeed an energetically favored configuration of the system.
The reason for nearby interstitials being favored compared to isolated ones is most probably the reduction of strain energy enabled by combination in contrast to the strain energy created by two individual defects.
\begin{figure}[h]
\begin{center}
\begin{minipage}[t]{7cm}
a) \underline{$E=-2.25\text{ eV}$}
\begin{center}
\includegraphics[width=6cm]{00-1dc/2-25.eps}
\end{center}
\end{minipage}
\begin{minipage}[t]{7cm}
b) \underline{$E=-2.39\text{ eV}$}
\begin{center}
\includegraphics[width=6cm]{00-1dc/2-39.eps}
\end{center}
\end{minipage}
\end{center}
\caption{Relaxed structure of defect complexes consisting of two \hkl<1 0 0>-type dumbbell interstitial defects.}
\label{fig:defects:comb_db_01}
\end{figure}
Figure \ref{fig:defects:comb_db_01} shows the structure of these two configurations.
Structure b) is the energetically most favorable configuration.
The two carbon atoms form a bond with a length of 1.38 \AA close to the nex neighbour distance in diamond or graphite, which is approximately 1.54 \AA.
This suggests prefered C clustering as a competing mechnism to the C-Si dumbbell agglomeration inevitable for the SiC precipitation.
Todo: Activation energy to obtain a configuration of separated C atoms again?
However, for the second most favorable configuration, presented in figure a), the amount of possibilities for this configuration is twice as high.
C-Si and C-C PC ...


001 at pos 2 looks as if there is no interaction.
There is an interaction but in the same time strain is reduced due to the opposing orientations of the defects, which leads to this low energy value.

Explanation of results of defects created along <110>.

-1.90 ...

-2.16 ...

the more far-off ones:
-0.27 and -0.12 ...
but better ...
-1.88 and -1.38 ...

Minimum E (reorientation) per distance

Todo: Si int and C sub ...
Todo: Model of kick-out and kick-in mechnism?

Todo: Jahn-Teller distortion (vacancy) $\rightarrow$ actually three possibilities! :(