Intrinsic point defects and volume swelling in ZrSiO4 under irradiation
Extracted source text ยท arXiv cond-mat/0404433
%% ****** Start of file defects.aps ****** %
%%
%%
%% This file is part of the APS files in the REVTeX 4 distribution.
%% Version 4.0 of REVTeX, August 2001
%%
%%
%% Copyright (c) 2001 The American Physical Society.
%%
%% See the REVTeX 4 README file for restrictions and more information.
%%
%
% Group addresses by affiliation; use superscriptaddress for long
% author lists, or if there are many overlapping affiliations.
% For Phys. Rev. appearance, change preprint to twocolumn.
% Choose pra, prb, prc, prd, pre, prl, prstab, or rmp for journal
% Add 'draft' option to mark overfull boxes with black boxes
% Add 'showpacs' option to make PACS codes appear
% Add 'showkeys' option to make keywords appear
%\documentclass[showpacs,preprint,aps,prb,twocolumnn,superscriptaddress,showpacs]{revtex4}
%\documentclass[showpacs,aps,prb,superscriptaddress,showpacs]{revtex4}
\documentclass[twocolumn,showpacs,aps,prb,showpacs,altaffillsymbol,groupaddress]{revtex4}
%\documentclass[preprint,aps,prb,twocolumnn,superscriptaddress,showpacs]{revtex4}
%\documentclass[aps,prl,preprint,superscriptaddress]{revtex4}
%\documentclass[aps,prb,twocolumn,groupedaddress]{revtex4}
\usepackage{graphicx}
\usepackage{dcolumn}
% You should use BibTeX and apsrev.bst for references
% Choosing a journal automatically selects the correct APS
% BibTeX style file (bst file), so only uncomment the line
% below if necessary.
%\bibliographystyle{apsrev}
\begin{document}
% Use the \preprint command to place your local institutional report
% number in the upper righthand corner of the title page in preprint mode.
% Multiple \preprint commands are allowed.
% Use the 'preprintnumbers' class option to override journal defaults
% to display numbers if necessary
%\preprint{}
%Title of paper
\title{Intrinsic point defects and volume swelling in ZrSiO$_4$ under irradiation}
\author{J.M. Pruneda}
\email[Corresponding Author:]{mpru02@esc.cam.ac.uk}
\author{T. D. Archer}
\altaffiliation{Present address: Physics Department, Trinity College, Dublin, Ireland.}
\author{Emilio Artacho}
\affiliation{Department of Earth Sciences, University of Cambridge
Downing Street, Cambridge, CB2 3EQ, United Kingdom}
%\affiliation{Department of Earth Sciences, University of Cambridge
%Downing Street, Cambridge, CB2 3EQ, United Kingdom}
% repeat the \author .. \affiliation etc. as needed
% \email, \thanks, \homepage, \altaffiliation all apply to the current
% author. Explanatory text should go in the []'s, actual e-mail
% address or url should go in the {}'s for \email and \homepage.
% Please use the appropriate macro foreach each type of information
% \affiliation command applies to all authors since the last
% \affiliation command. The \affiliation command should follow the
% other information
% \affiliation can be followed by \email, \homepage, \thanks as well.
%\author{}
%\email[]{Your e-mail address}
%\homepage[]{Your web page}
%\thanks{}
%\altaffiliation{}
%Collaboration name if desired (requires use of superscriptaddress
%option in \documentclass). \noaffiliation is required (may also be
%used with the \author command).
%\collaboration can be followed by \email, \homepage, \thanks as well.
%\collaboration{}
%\noaffiliation
\date{\today}
\begin{abstract}
The effects of high concentration of point defects in crystalline ZrSiO$_4$
as originated by exposure to radiation, have been simulated using first principles
density functional calculations. Structural relaxation and vibrational
studies were performed for a catalogue of intrinsic point defects, with
different charge states and concentrations. The experimental evidence
of a large anisotropic volume swelling in natural and artificially
irradiated samples is used to select the subset of defects
that give similar lattice swelling for the concentrations studied,
namely interstitials of O and Si, and the anti-site Zr$_{\text{Si}}$,
Calculated vibrational spectra for the interstitials show additional
evidence for the presence of high concentrations of some of these defects in
irradiated zircon.
\end{abstract}
% insert suggested PACS numbers in braces on next line
\pacs{61.72.-y,61.80.-x,78.30.-j}
%\pacs{71.15.-m,71.15.Nc,79.20.Ap,61.80.-x,34.20.Cf}
%\pacs{{\it Keywords:} Density Functional Theory; }
%
%\keywords{}
%\maketitle must follow title, authors, abstract, \pacs, and \keywords
\maketitle
% body of paper here - Use proper section commands
\section{Introduction}
Large concentration of defects in materials produce strains that
affect the crystalline lattice structure, which can undergo
amorphization, change into a different crystalline phase, or remain in
the same crystal phase with changes in the lattice parameters. In
addition to these changes, the electronic properties are affected in a
way that can produce substantial differences in the physical and chemical
properties of the original material.
Radiation effects in ceramic materials have been subject of a
considerable number of studies due to their use in nuclear
fusion applications, and in the stabilization and immobilization
of nuclear high-level waste\cite{Weber98,Grimes}. Defect accumulation and its
consequences for the degradation of the mechanical properties
have to be understood in order to control possible leaching, creep and
fatigue of these materials.
%Materials can be destabilized by such
%defect accumulation producing amorphization.
A number of issues have to be investigated,
such as the stabilities of possible defects, their structure,
migration energies for interstitials and vacancies,
displacement threshold energies, defect interactions and clustering,
and their electronic structure.
Zircon (ZrSiO$_4$) has attracted considerable attention due to several
physical properties of interest, such as high permitivity (zirconium
silicates are candidates as a gate dielectric in future silicon
field effect transistors\cite{Wilk}), low thermal conductivity and
high chemical stability. It has also been proposed as a ceramic host
for the immobilization of nuclear waste due to its high ability to
accommodate actinides by substitution of Zr, and its high resistance
to corrosion\cite{Weber98,Ewing95}. Natural zircon may contain
uranium in concentrations of up to 20,000 ppm, and samples with up to
$~$4 billion years have been dated. Thus, even if possibly not the
best final material for nuclear waste encapsulation, it is an
ideal system to study the effects of radiation damage.
After a radioactive impurity undergoes an $\alpha$-decay transition,
strong damage is produced in the crystalline structure. The heavy
recoiling atom (tens to hundreds of keV) propagates through the material
losing energy mainly through atomic collisions, thus producing large
cascade collisions and in consequence, amorphized regions. The highly
energetic $\alpha$ particle ($\sim$5 MeV) that propagates in the opposite
direction to the recoil atom, dissipates most if its energy
by ionization processes, and a relatively smaller number of atomic
displacements mainly at the end of its path.
When the dose of $\alpha$-decays is not too large there is a coexistence
between the amorphized (metamict) region and the crystalline phase,
the latter containing point defects. Barriers to recombination cause the
accumulation of large concentration of these defects, resulting
in changes of the unit cell volume and shape. This swelling is measured from
the shifts in the reflection angle in X-ray diffraction (XRD) and it is
clearly distinct from the one produced by the amorphous region, observed
by changes in the total density.
Experiments based on natural
samples, Pu-doped zircon and ion-beam irradiation, show a unit cell
volume expansion of up to 5\%. This volume swelling is anisotropic,
giving an increase of $\sim$1.5\% along the $ab$-axis, and $\sim$2\%
along the $c$-axis\cite{H&G}. Over geological periods the recovery of
a large fraction of point defects is observed in natural zircons
(as opposed to Pu-doped samples\cite{H&G,Weber93}), with preferential
self-annealing of the $ab$ plane\cite{Rios}.
Characterization of structural changes in the crystalline
phase\cite{Rios} show that the dodecadeltahedron ZrO$_8$ is
the less stable unit in zircon, and the defects are expected
to be more closely related to it, as the Si atoms are well
confined in the SiO$_4$ tetrahedra. Theoretical studies of
neutral point defects show that the oxygen interstitial is the
most stable one\cite{Crocombette}. A non negligible concentration
is expected at thermal equilibrium.
%Vacancy migration simulations\cite{migration} further support the
%experimental results, with three-dimensional migration networks fully
%connected for Zr and O, but with underestimated energy barriers
%($\sim$ 1.2eV for Zr, and slightly smaller for O).
The energies required to form native defects in zircon were
investigated by molecular dynamics simulations of primary-knock-on-atom
collisions with empirical potentials\cite{PKA}.
It was shown that the effective energy required to displace O from
its crystal position is $\sim$17 eV, whereas it is larger for Si
and Zr ($\sim$48, and $\sim$60 respectively).
The previous theoretical studies of point defects in zircon were focused
on the energetics of neutral defects. However, due to ionization processes
produced by the $\alpha$-particle, and the fast that the resulting defect
landscape affects the electronic chemical potential, the defects can be charged.
The accumulation of charged defects produce electric fields
that would change the kinetics of accumulation and diffusivity.
In the present work, we use first principles electronic structure
calculations to study the influence of very high concentration of neutral and
charged point defects on the crystalline structure of ZrSiO$_4$.
We perform structural relaxations, studying electronic properties,
as well as some spectroscopic calculations of a wide variety of defects, forming
a quite complete catalogue that is compared with experimental data.
The effects of order/disorder of these defects have not been considered in the
present study, where periodic supercells were used to describe them.
%The paper is organized as follows. In the next section we will describe the
%theoretical methods used. Next, we will introduce the physical properties
%of crystalline zircon. We then continue with the studies of a variety
%of point defects (interstitials, vacancies and others), summarizing the
%stable ionic configurations in different charge states, and comparing with
%available data. We conclude with a discussion of the results in terms of
%the effect on the volume swelling induced by the defects, and the proposal of
%a small set of reasonable candidates.
\section{Methodology}
Calculations of the electronic properties of ZrSiO$_4$ are performed with
the self-consistent {\it ab initio} {\sc siesta} method\cite{siesta}, using
Density Functional Theory (DFT)\cite{DFT} within the Local Density
Approximation (LDA)\cite{ca}.
The core electrons are replaced by norm-conserving
pseudopotentials\cite{tm2} in the Kleinman-Bylander form\cite{kb},
generated with Zr(4s$^2$,4p$^6$,4d$^2$,5s$^2$), Si(3s$^2$,3p$^2$),
and O(2s$^2$,2p$^4$) atomic valence configurations.
The 4p electrons in Zr were explicitly included in the calculations as
semicore states, due to the large overlap with the valence states.
The wave functions are described with linear
combinations of strictly localized numerical atomic orbitals\cite{basis}.
We used a single-$\zeta$ basis set for the semicore states of Zr,
and double-$\zeta$ plus polarization for the valence states\cite{optZr}. The
charge density is projected on a real space uniform grid with an
equivalent plane-wave cutoff of 227 Ry, to calculate the
exchange-correlation and Hartree matrix elements.
To simulate the defect structures, the host crystal is represented
by a supercell generated by repetition of the conventional tetragonal
unit cell (4 formula units). The point defect is then introduced
inside this supercell (adding atoms for interstitial, removing
atoms for vacancies, etc), that have to be of the right size to
describe the sought large defect concentrations.
We use host supercells with 24, 48, and 96 atoms. The supercell
with 48 atoms is a repetition of the tetragonal cell along the $c$ axis,
and the one with 96, is a repetition along the $a$ and $b$ axis.
In this way, different concentrations of defects can be simulated by
changing the number of repetitions in the cell. The supercells with
24 and 48 atoms were already used by Crocombette in his study of points
defects, and he estimated that unreal physical interaction between
periodic images of the defects was of the order of 0.5 eV. For the high
concentrations we are interested in, these interactions between defects
are expected and only their periodic arrangement might be unrealistic.
The use of more than one defect per supercell can alleviate this
approximation, and will be discussed later.
Simulating charged defects with periodic boundary conditions,
requires handling with care the long-range coulombic interactions.
The unphysical divergence in the energy coming from the long range
Coulomb interactions of a periodic array of charges
is compensated by a uniform electron-charge neutralizing
background\cite{payne}.
The minimum-energy structure for each defect was obtained by relaxing
with a conjugate-gradient minimization for the forces and stresses.
The lattice vectors and the atomic positions were allowed to relax
until atomic forces and stresses from the electronic structure calculations
were smaller than 30 meV/\AA\ and 6 meV/\AA$^3$, respectively.
The formation energies of the defects in the different charge states
are function both of the electron chemical potential $\mu_e$, and of the
atomic chemical potentials for the species involved in the defect, and
will be discussed elsewhere\cite{Edefects}.
The local vibrational modes (LVM) associated to the defects are
obtained from the dynamical matrix
%, defined as second derivatives
%of the energy with respect to atomic displacements. This matrix is
%obtained
in the relaxed structure, by finite differences, i.e. computing
the energies with each atom displaced by $\sim\pm$0.016\AA\ along
each of the cartesian coordinates. The eigenvalues and eigenvectors
of the dynamical matrix give the vibrational
frequencies and eigenmodes in the harmonic approximation. Previous
studies show a deviation of $\sim$5\% for vibrational frequencies of
point defects in semiconductors.
The Born-effective charge tensors, $Z_{ij,\alpha}^*$,
are used to determine the infrared intensities, and the splitting
between longitudinal (LO) and transverse (TO) optic modes\cite{LOTO}.
They are computed by finite differences as the change in the $i$
component of the polarization (computed with the Berry phase
formalism\cite{Vanderbilt}) when the atom $\alpha$ is displaced
along the $j$ direction.
%\begin{enumerate}
% \item k point sampling???
%\end{enumerate}
\section{Crystal $ZrSiO_4$}
Zircon (ZrSiO$_4$) has tetragonal $I4_1/amd$ space group.
The structure consists of alternating SiO$_4$ tetrahedra
and ZrO$_8$ triangular dodecadeltahedra sharing edges and
forming chains parallel to the crystallographic $c$ axis
(Fig. \ref{polyhedra}). A body-centred unit cell can be chosen,
containing four formula units. The structure is fully described with 4
parameters (see table \ref{lattice}): The lattice constants $a$
and $c$, and the internal parameters $y$ and $z$. Zirconium and
silicon atoms are located in the $4a$ and $4b$ Wyckoff positions,
$(0,\frac{3}{4},\frac{1}{8})$ and $(0,\frac{1}{4},\frac{3}{8})$,
respectively, and the oxygen atoms in the $16h$ sites (0,y,z).
Each O is shared between one Si and two Zr atoms, at distances 1.62,
2.12 and 2.25\AA\ respectively. These 4 atoms are in the O-site's
mirror plane that can be perpendicular to the $a$ or $b$ axis.
The relaxed parameters calculated are in good agreement with
experimental values.
\begin{figure}[t]
\includegraphics[scale=0.75]{polyhedra2b.eps}
\caption{({\it Color Online})
(a) Edge-sharing Si-tetrahedra and Zr-dodecadeltahedra
along the $c$-axis, corner sharing in the $ab$-plane. The
dodecadeltahedra also have edge-sharing with each other along
the $ab$-plane. (b) and (c) show the projection of the unit
cell in the $ab$ plane, and in a plane perpendicular to the $b$ axis
(the second layer of polyhedra has been removed in (c) for a better
visualization). The dashed circles in (b) show the empty spaces in which we
place the interstitials.}
\label{polyhedra}
\end{figure}
\begin{table}[b]
\caption{Calculated structural parameters for crystalline ZrSiO$_4$
compared to experimental values. }
\begin{tabular*}{8cm}{c@{\extracolsep{\fill}}ccc}
& This work & Ref. [\onlinecite{Gonze}]
& Experiment\cite{latt} \\
\hline
Volume & 129 (\AA$^3$)& 127 & 131 \\
a & 6.59 (\AA) & 6.54 & 6.61 \\
c & 5.96 (\AA) & 5.92 & 5.98 \\
y & 0.068 & 0.0645 & 0.0661 \\
z & 0.184 & 0.1945 & 0.1953 \\
\\
d(Si-O) & 1.62 & 1.61 & 1.62 \\
d(Zr-O$^{\alpha}$) & 2.25 & 2.24 & 2.27 \\
d(Zr-O$^{\beta}$) & 2.12 & 2.10 & 2.13 \\
$\widehat{O\textendash Si\textendash O}$ & 96$^\circ$ & 97$^\circ$ & 97$^\circ$\\
& 116$^\circ$ & 116$^\circ$ & 116$^\circ$\\
\hline
\end{tabular*}
\label{lattice}
\end{table}
%The electronic density of states is shown in figure \ref{dos}, where four clear groups of bands can be distinguished.
The Born effective-charge tensor is shown in table \ref{BC} and is
in excellent agreement with the results obtained with linear response
calculations\cite{Gonze}. Because of the local symmetry in the Zr
and Si sites,
the Born charge tensors for these atoms are diagonal, with only two
independent components, Z$^*_\parallel$ and Z$^*_\perp$ parallel
and perpendicular to the crystallographic $c$-axis, respectively.
The large value of Z$^*_{Zr}$ and Z$^*_{Si}$ with respect to the
nominal ionic electronic charge is interpreted in terms of mixed
covalent-ionic bonding due to hybridization with oxygen\cite{Gonze}.
For O, the Born charge tensor is anisotropic, with the charge in the
mirror plane being larger than the nominal ionic charge (Z=-2), and
the tensor component perpendicular to the mirror plane being smaller.
\begin{table}[b]
\caption{Born effective charge tensors for Zr, Si, and O atoms.
The Born charge for O, correspond to the atom located at $(0,u,v)$, with
the mirror plane perpendicular to $x$.
For Zr and Si, only the diagonal elements are shown. For O, the whole
charge tensor is presented, as well as the corresponding eigenvalues.
See also Ref. [\onlinecite{Gonze}]}
\begin{tabular*}{8cm}{c@{\extracolsep{\fill}}cc}
\hline
\hline
\\
Z$^*_{Zr}$ & $\begin{pmatrix}{5.33 & 5.33 & 4.66} \end{pmatrix}$ & \\
Z$^*_{Si}$ & $\begin{pmatrix}{3.21 & 3.21 & 4.38} \end{pmatrix}$ & \\
Z$^*_{O}$ & $\left(\begin{matrix}{-1.19 & 0 & 0 \\
0 & -3.11 & -0.15 \\
0 & -0.34 & -2.26 } \end{matrix}\right)$ &
$\left[\begin{matrix}{ -1.19 \\
-3.16 \\
-2.20 } \end{matrix}\right]$ \\
\\
\hline
\end{tabular*}
\label{BC}
\end{table}
Seven infrared active modes (3A$_{2u}$+4E$_u$) are predicted by
group theory in ZrSiO$_4$ at $\Gamma$. The induced polarization
for E$_u$ modes is perpendicular to the $c$ axis, while A$_{2u}$
modes are observed for electric field vectors parallel to c.
The spectrum between 650 and 1400 cm$^{-1}$ is characterized
by Si-O stretching vibrations (E$_u$(4) and A$_{2u}$(3) modes
respectively \cite{Zhang2001}) and the O-Si-O bending
modes E$_u$(3) and A$_{2u}$(2).
Our computed frequencies for these modes are 851, 969, 414, and
602 cm$^{-1}$ respectively, with a maximum deviation of 3.8\%
with experiments. The modes below 400 cm$^{-1}$ are
related to SiO$_4$ group motions against Zr atoms, and motions
of Zr atoms.
\begin{table}[b]
\caption{Frequencies for IR active modes in ZrSiO$_4$ (in cm$^{-1}$).}
\begin{tabular*}{8cm}{c@{\extracolsep{\fill}}ccc}
\hline
Mode & This work & Ref. [\onlinecite{Gonze}] & Expt.\cite{Zhang2001} \\
\hline
\hline
A$_{2u}$(1) & 371 & 348 & 338 \\
A$_{2u}$(2) & 602 & 601 & 608 \\
A$_{2u}$(3) & 969 & 979 & 989 \\
E$_{u}$(1) & 295 & 285 & 287 \\
E$_{u}$(2) & 402 & 383 & 389 \\
E$_{u}$(3) & 414 & 422 & 430 \\
E$_{u}$(4) & 851 & 867 & 885 \\
\hline
\label{IR-modes}
\end{tabular*}
\end{table}
\section{Point defects}
We have considered a catalogue of neutral and charged intrinsic
defects, that includes interstitials (X$_i$) and vacancies (V$_X$) of the three
elements ($X$) present in zircon, Zr and Si anti-site defects (Zr$_{\text Si}$,
and Si$_{\text Zr}$), Frenkel pairs ($X_{FP}$), and other combinations of
interstitials and vacancies.
\subsection{Volume swelling}
\begin{figure*}[t]
\includegraphics[scale=0.70]{candidates.ps}
\caption{({\it Color online}) Lattice parameter swelling (in \%) for
the set of defects
considered in this work. An average in the basal (ab) plane is done
to compare with the relaxed crystalline structure. X$_i$, V$_X$,X$_{FP}$
denote interstitials, vacancies and Frenkel pairs of X, respectively.
The different charge states are denoted by $\pm q$
and the different concentrations of defects by $\times n$, with
$n$ denoting the number of repetitions of the tetragonal unit cell
(cells with 24, 48 and 96 atoms, for $n=1,2,4$ respectively). Unless
stated, the neutral defect with $n=2$ cell is considered.}
\label{swelling}
\end{figure*}
The effect of point defects in the lattice structure after relaxation
is summarized in Fig. \ref{swelling}, where we show the change in the
lattice parameters relative to the neutral structure for the defects studied.
Some defects produce a considerable distortion of the crystalline lattice,
whereas others such as oxygen interstitial, or the vacancies have a much
smaller effect. The magnitude of the distortion depends on the
concentration of defects (supercell size in our study). For
the values of interest in this work, a clear trend
is observed, with an almost linear dependence between the distortion
and the concentration (see Fig. \ref{concentration}).
For the highest concentration studied (one defect per unit cell), we
explored the effect of defect disorder by introducing two point defects
in the supercell made from two unit cells. Fig. \ref{concentration} shows
a dispersion in the swelling due to the different configurations
studied in this way. Even if the effect is sizeable it does not affect
our arguments or conclusions, though further work in this direction
will be needed.
\begin{figure}[t]
\includegraphics[scale=0.45]{concentration.ps}
\caption{({\it Color online}) Change in lattice constants (\%) as a function of the
concentration of defects, for interstitials and vacancies of the three
constituent elements. Results are shown for 1 defect for each
$N$-repetitions of the tetragonal unit cell, with $N=1,2,4$ and 8.
The small arrows show the anisotropic swelling in the $ab$ plane
for Zr$_i$. The different points for 1 defect/cell refer to different
disorder realizations.}
\label{concentration}
\end{figure}
In general, there is a considerable anisotropy in the relaxation, with
different changes in the $c$-axis and in the $ab$-plane.
Experimentally\cite{H&G}, an increase in $c$ of $\sim$2\% is
observed, while $a$ increases by a smaller amount ($\sim$1.5\%).
Based in this evidence, we can focus
our study on those defects that give a behaviour in the lattice
swelling similar to the one observed, eliminating those that
produce a decrease in the cell parameters, or a larger change
in $a$ than in $c$. The defects with the stronger distortions
are the interstitial of silicon, Si$_i$, and the anti-site of
zirconium in the position of silicon, Zr$_{\text Si}$.
The calculations for charged systems were done with the supercell
of 2 tetragonal unit cells, repeated along the $c$-axis.
The artificial coulomb interaction between cells should be then
more important in the $ab$-plane, but we observed that the lattice
parameter $c$ is more strongly affected by the charge state than $ab$.
\subsection{Structural relaxation and electronic structure.}
\subsubsection{Interstitials}
%We have considered incorporation of interstitial for each of the
%species present in zircon (Zr$_i$, Si$_i$ and O$_i$) in different
%charge states. In this section we will analyze the structural
%relaxation of these defects in zircon.
Initially, we place the interstitial atoms in regions with empty space
(see dashed circles in Fig. \ref{polyhedra}), and perform
conjugate gradient minimization
of the forces and stresses. We have explored different starting points,
showing here the relaxed structures with minimum energy.
For Zr$_i$ and Si$_i$ the interstitial remains along the $c$-axis
empty spaces (Fig. 1b). Nevertheless, the lattice
is strongly distorted and the SiO$_4$ and ZrO$_8$ polyhedra change
to accommodate to the presence of the interstitial.
There is an anisotropic change in the lattice parameters. While
Zr$_i$ produces large orthorhombic distortion in the basal plane
(with an increase of up to 3\% in the $a$ lattice parameter,
for a concentration of 1 defect in two unit cells, and smaller
distortions in the other parameters), the Si$_i$ distorts the
crystal mainly in the $c$-axis, but also in the other directions.
Upon removal of electrons from Si$_i$, new atomic structures are obtained,
with the interstitial causing oxygen atoms to approach, and neutralize the
charge around it. In this way, the coordination of Si$_i^{+n}$ increases
with $n$ and for Si$_i^{+4}$ a six-fold coordinated structure results,
with Si-O bonds slightly longer ($\sim$1.8\AA) than in the initial
tetrahedral geometry. In Fig. \ref{iSi} we show the electronic deformation
density (difference between the solid and atomic electronic clouds),
$\delta\rho$, around Si$_i$ in two different planes. The geometries
for positively charged Zr-interstitials are basically unchanged with
respect to the neutral structure.
\begin{figure}[b]
\includegraphics[scale=0.28]{i-Si.+q.eps}
\caption{({\it Color online}) Atomic configurations for silicon interstitial in the +4 charged state
and corresponding charge density-deformation
contour plot in the plane defined by Si$_i$, O$_A$ and O$_C$ (a),
and Si$_i$, O$_A$ and O$_B$ (b).
Distances given in \AA\ and the contour plots are drawn
between -0.45 and 0.45$e$, with a step of 0.015$e$. Solid (dashed) lines
correspond to negative (positive) values for the charge.}
\label{iSi}
\end{figure}
%Interstitial oxygen displaces from the $c$ channel and connects
%to another oxygen in the lattice.
In agreement with previous
theoretical studies\cite{Crocombette}, we observe that the neutral
O$_i$ forms a ``dumbbell''-like structure (see Fig. \ref{i-O}a),
with the axis almost
perpendicular to the mirror plane of the oxygen in the crystal
position. The distances
between both oxygen atoms, and the neighbouring cations are shown
in table \ref{dumbbell}. A similar structure was also observed in
calculations of interstitial O in ZrO$_2$\cite{Nieminen}. The high
ionicity of the crystal forces the interstitial to a position where
the electronic density screens the coulomb interactions. The covalent
bond to the interstitial oxygen reduces the strength of the Si-O bond,
and the SiO$_4$ tetrahedra loses part of its charge. Mulliken population
analysis shows that the charges in the interstitial and in the lattice oxygen
are similar, and slightly smaller (6\%) than in other oxygen atoms.
%The existence of charged O$_i$ defects can be energetically favoured.
\begin{figure*}[t]
\includegraphics[scale=0.58]{i-O.ps}
\caption{({\it Color online}) Atomic configurations for oxygen
interstitials in the neutral (a) and doubly negatively charged (b)
states, and corresponding deformation density
contour plots in the plane defined by Si$_A$, O$_A$ and O$_i$ (c and d
for neutral and charged configurations, respectively).
(e) shows the contour plot in the plane through Si$_A$, O$_i$ and Si$_B$,
in O$_i^{-2}$. Distances given in \AA\ and the contour plots are drawn
between -0.45 and 0.45$e$, with a step of 0.015$e$. Solid (dashed) lines
correspond to negative (positive) values of the deformation density.}
\label{i-O}
\end{figure*}
\begin{table}[b]
\caption{Interatomic distances in the dumbbell structure (in \AA), for
different concentration of defects (supercells with 24, 48 and 96 atoms).
Charged defect is within the supercell containing 48 atoms.}
\begin{tabular*}{8cm}{c@{\extracolsep{\fill}}ccccc}
& 24 & 48 & 96 & Ref.[\onlinecite{Crocombette}] & O$_i^{-}$ \\
\hline
\hline
O$_i$-O$_A$ & 1.52 & 1.52 & 1.52 & 1.62 & 1.99 \\
O$_i$-Si & 1.96 & 1.69 & 1.70 & 1.83 & 1.69 \\
O$_i$-Zr$_1$ & 2.29 & 2.24 & 2.23 & 2.24 & 2.15 \\
O$_i$-Zr$_2$ & 2.22 & 2.84 & 2.82 & 2.57 & 2.98 \\
O$_A$-Si & 1.70 & 2.00 & 1.95 & 1.78 & 1.80 \\
O$_A$-Zr$_1$ & 2.24 & 2.23 & 2.30 & 2.20 & 2.32 \\
O$_A$-Zr$_2$ & 2.84 & 2.22 & 2.22 & 2.75 & 2.18 \\
\hline
\label{dumbbell}
\end{tabular*}
\end{table}
The structure of charged O$_i^{-}$ does not differ substantially
from the neutral defect, with a slight increase of the O$_i^{-}$-O
distance (see table \ref{dumbbell}). On the contrary, the doubly
charged interstitial displaces from the dumbbell structure, and
forms a bridge between neighbour silicon atoms, both having now
coordination five. In this structure, the relaxed O$_i^{-2}$-O$_A$
distance (2.19 \AA) does not admit a bond between the oxygens,
while Si$_A-$O$_i^{-2}$ and Si$_B-$O$_i^{-2}$ do (1.74 and 1.78
\AA, respectively). This can be seen in Fig. \ref{i-O}d, where
$\delta\rho$ around the interstitial oxygen differs from
the one in the neutral defect (Fig. \ref{i-O}c). The electronic cloud
along the Si$_A$-O$_i$-Si$_B$ directions (Fig. \ref{i-O}e) resembles
the Si-O bond in the crystalline configuration.
The relaxation energy from the initial dumbbell configuration is
of about 1.6 eV, while the energy of the neutral defect in the atomic
structure of O$_i^{-2}$ is 2.9 eV higher.
%This lowering of the energy that results from the capture of a
%second electron is known as Anderson's ``negative-U'' behaviour.
%It is interesting to note that the decay of two isolated O$_i^-$
%centre into a O$_i^{-2}$ and a neutral O$_i$ is energetically
%favorable, with a gain of $\sim$1.0 eV.
\subsubsection{Vacancies}
In the V$_O$, the silicon atom moves slightly in the direction
of the missing oxygen, and the remaining Si-O bond lengths are
increased by 4\%. A new electronic state is created in the gap,
localized around the vacancy. In the charged defect
V$_O^+$, the Si atom moves back toward its initial position, and the
Si-O bonds have deviations smaller than 1\% with respect to the
original Si-O tetrahedra.
The four oxygen atoms forming the tetrahedra around the silicon
vacancy remain in their position, with only minor deviations
towards their centre of mass ($\sim$-0.7\%). For the
V$_{\text Si}^{-n}$ configurations, the oxygen tetrahedron expands, with
the distance from the centre of mass increasing by 2\% for $n=1$ and
by 5\% for $n=2$. For the V$_{Zr}$ defect a similar effect is
observed for the oxygen atoms forming the dodecadeltahedra.
There is a considerable expansion of the V$_{Zr}-$O$^\alpha$
bond ($\sim$10\%), and a smaller change in the V$_{Zr}-$O$^\beta$
length ($\sim$+4\%), that explains the anisotropic change in
lattice parameters. The expansion of the V$_{Zr}-$O$^v$ bond increases for
the negatively charged vacancies. The V$_{Zr}-$O$^i$ length expansion
decreases (increases) for negatively (positively) charged vacancies.
%Again, there is a negative-U behaviour, and the energy of the decay
%reaction 2V$_O^+\Rightarrow$V$_O^{+2}+$V$_O$ is +1.7 eV.
\subsubsection{Frenkel pairs and anti-sites}
Frenkel pairs of Zr and Si are not favourable because of the strong
ionic interaction between the interstitial and the vacancy. There is
a trend for the interstitial to move towards the open space of the
vacancy, and a restitution was observed in the relaxation of
several initial structures considered.
%The formation energies for these
%defects are a factor of two smaller than the values given in
%Ref.\onlinecite{Crocombette}, where they are computed from the formation
%energies of the interstitial and the vacancy. Doing the same for the
%charged pairs (V$_{\text Zr}^{-n}+$Zr$_i^{+n}$ and
%V$_{\text Si}^{-n}+$Si$_i^{+n}$, for $n=1,2,3,4$) we obtain that the energy
%decreases from $\sim$20 eV to $\sim$11 eV for the
%V$_{\text Zr}^{-4}+$Zr$_i^{+4}$ pair, and the same for
%V$_{\text Si}^{-4}+$Si$_i^{+4}$. This suggests that the Frenkel pairs,
%Si$_{\text FP}$ and Zr$_{\text FP}$, that we actually simulated are
%more close to the charged configuration than to the neutral one, showing
%the high ionic character of these defects.
In the case of O$_{FP}$, the
interstitial and the vacancy do not interact strongly, and the final
configuration is similar to an isolated vacancy and a dumbbell interstitial,
even when both are in the same Si-O tetrahedra.
%The energy of the
%charged pairs (computed from the charged interstitial and vacancy)
%is slightly different, with V$_O^{+}+$O$_i^{-}$ 1.0 eV higher in energy,
%and V$_O^{+2}+$O$_i^{-2}$ 0.9 eV lower than the neutral pair.
In the Si$_{\text Zr}$ anti-site, the Si atom substitutes a Zr atom in the
center of the dodecadeltahedron, with a reduction of the lengths
(-4\% for Si-O$^i$ and -6\% for Si-O$^v$). In Zr$_{\text Si}$, the
Zr forms a tetrahedron with four oxygen around it, with bond lengths
of 1.9\AA (+17\% longer than the Si-O bonds). The Si-O distances increase
slightly ($\sim$+1\%) close to the defect, due to the increase in the
lattice parameters. Fig. \ref{Zr-O} shows
the contour plot of $\delta\rho$ in the plane defined by Zr$_{\text Si}$
and a pair of oxygens in the tetrahedron. In the charged configurations,
the bond lengths do not change substantially.
\begin{figure}[t]
\includegraphics[scale=0.38]{ZrSi.ps}
\caption{({\it Color online}) Charge redistribution density around Zr$_{\text Si}$ and Si, in
the tetrahedral plane.}
\label{Zr-O}
\end{figure}
\subsection{Vibrational properties}
The infrared (IR) reflection and absorption spectra of zircon is
considerably affected by radiation damage. Changes in peak frequency,
decrease in band intensity, and line broadening of the spectra of
crystalline zircon can be related to changes in the lattice bond lengths.
In addition, new features appear for highly damaged samples,
with an increase in the IR intensity with increasing dose. One of the
new spectral features is the appearance of an absorption sharp peak near
796 cm$^{-1}$ in moderately damaged zircon\cite{Zhang2000b}.
This peak is in a gap
of the vibrational density of states (VDOS) of crystalline zircon. Its
origin is unclear, and it has been proposed that it could be due
to a radiation-induced intermediate phase, or to distortions in the
boundaries between crystalline and amorphized regions\cite{Zhang2000b}.
We computed the vibrational properties for interstitials of O, Si and Zr,
and for the anti-site defect Zr$_{\text Si}$, in the relaxed structures
with 48 atoms.
IR active local vibrational modes associated to the defects were observed
for O$_i$ and Si$_i$ in a range of energies corresponding to the gap in
the VDOS (729 and 743 cm$^{-1}$ respectively). The eigenmodes are shown in
Fig. \ref{LVM}. They mainly involve the stretching of Si-O bonds close
to the interstitials.
Other localized modes were observed for these point defects, but
they are not IR active.
The crystal modes are affected by the presence of the defect.
The changes are more important in Si$_i$, probably due to a
stronger interaction with oxygen atoms around this interstitial.
For O$_i$, the frequencies of the crystalline modes are almost
unaffected (though some splittings are observed for the low frequency modes)
showing that the defective structure does not differ much from the
crystalline one. In the charged configuration (O$_i^{-2}$), the interstitial
is bonded to two silicon atoms, hence considerably affecting the vibrational
modes of the Si-O tetrahedra. Some weak IR active modes appear in the
gap for this defect. Zirconium interstitial affects mainly the low
energy part of the spectra (below 500 cm$^{-1}$), where the Zr-O
interactions are more important. The Si-O stretching modes are softer,
probably due the increase in the distance between silicon tetrahedra,
consequence of the large expansion in the $ab$-plane. For the anti-site
defect Zr$_{\text Si}$, the high energy modes, involving the stretching
of the tetrahedra bonds, are weaker. The low energy modes,
mainly related to relative displacements of the Zr and Si-tetrahedra
sublattices, are less affected.
\begin{figure}[b]
\includegraphics[scale=0.40]{LVM.ps}
\caption{({\it Color online}) Local Vibrational Modes for O$_i$ and Si$_i$ point defects.
(a) and (b) show the eigenmodes for oxygen interstitial at 729 cm$^{-1}$
(IR active) and 870 cm$^{-1}$ respectively. (c) shows the IR active
mode for silicon interstitial at 734 cm$^{-1}$. The relatively
high frequency of mode (b) is an indication of the bond strength
between the oxygens in the dumbbell. (Distances in \AA)}
\label{LVM}
\end{figure}
\section{Discussion}
Experimentally, the change in the lattice parameters can be directly
monitored using x-ray diffraction.
Point defects would be randomly distributed through the crystal lattice.
If the number of accumulated $\alpha$-decays is not too large
($\sim2\times10^{18}\alpha/g$), and we assume that $\sim$220 atoms are
displaced by each $\alpha$ particle\cite{Rios}, we can expect concentrations of
$2\times10^{21}$defects/cm$^{3}$. That is roughly one defect for each
two unit cells, the kind of concentration considered in this work.
Similar concentrations have been used to study the effect of doping
on the lattice parameters of semiconductors\cite{Si:As}, but the
changes observed are considerably smaller ($\sim$0.19\%).
We performed simulations of Si interstitials in the crystalline
silicon structure with a supercell of 64 atoms (roughly the same
concentration of defects), giving a volume swelling of 0.2\%, much
smaller than the values obtained for zircon with the same approach.
In ceramic materials used for nuclear applications radiation
swelling is more important ($\Delta V/V$ up to 8\% in SiC) and
both defect accumulation and ratio between amorphized and crystalline
phases play an important role\cite{SiC}.
When compared with experiments, our calculations at those
concentrations, single out three candidates as possible originators of the
lattice swelling: O$_i$, Si$_i$ and Zr$_{\text Si}$.
Some concentration of these three defects should be there to account for the
experimental findings, even if other defects can be present at lower
concentrations. In the following we discuss the compatibility of this
conclusion with other facts known about the system.
For weakly damaged zircon, the fraction of crystalline phase is larger
than the fraction of amorphized regions, and the diffraction maxima are
weaker than in undamaged crystals. The anisotropic thermal displacement
factors have been used to quantify the effect of the radiation damage in
zircon\cite{Rios}, indicating that the damage has more important
effects on Zr and O atoms than on Si, which remain inside the SiO$_4$
tetrahedral units due to the stronger binding of Si with O.
In contrast with these findings, calculations of threshold displacement
energies\cite{PKA} (E$_d$) show that for neutral atoms
E$_d$(O)$\ll$E$_d$(Si)<E$_d$(Zr).
However, it is important to emphasize that E$_d$ can change substantially
with the charge state. Charged defects can migrate through the lattice
with a lower migration energy than neutral ones. In fact, the atomic
configurations for oxygen interstitials are similar to the
ones observed for ZrO$_2$ and HfO$_2$\cite{Nieminen,Hafnia}. In these
materials, oxygen is incorporated and diffuses in atomic form, acting as
an electron trap, while neutral defects tend to form strong bonds with other
lattice oxygens. The negative-U behaviour favors the appearance
of charged interstitials that are less tightly bound to other
oxygens.
In a real material, we can expect all kind of defects being present,
with the swelling produced by some of them being compensated with the
deflation due to others. The presence of interstitials
requires that vacancies are formed somewhere in the lattice, but the effect
of these vacancies on the lattice parameters is smaller (see Fig. \ref{swelling}).
The formation of Zr$_{\text Si}$ leaves a V$_{\text Zr}$ and a interstitial
of Si. The combination of these three defects is compatible with a
larger swelling along the $c$-axis than in the $ab$-plane, as observed
experimentally.
%The large lattice expansion obtained for Zr$_{\text Si}$ and Si$_i$
%is in agreement with the experimental anisotropic lattice swelling.
The formation energies for the point defects studied\cite{Edefects},
indicate that the most probable
defects would be interstitials of oxygen and the anti-site
Zr$_{\text Si}$. Depending on the chemical potentials of each species
($\mu_{\text O}$, $\mu_{\text Si}$ and $\mu_{\text Zr}$)
interstitials of Si and Zr might be present.
The IR active mode at $\sim$730 cm$^{-1}$ would give further evidence for the
presence of oxygen, and maybe silicon, interstitials.
$^{29}$Si nuclear magnetic resonance studies of damaged zircon indicate
changes in the Si local environment towards a polimerization of the
structure\cite{nmr} during amorphization. This means that Si$-$O$-$Si
bonds are formed in the amorphous region and the
initially isolated SiO$_4$ tetrahedra are connected through an
oxygen bridge. This polymerization requires that on average the
number of oxygens per Si is smaller in the amorphized region, favouring the
presence of interstitial oxygens in the rest of the system.
As we have seen O$_i^{-2}$ forms
bridging structures between Si atoms. But also the presence charged
Si interstitials can give similar polymerization and change the chemical
shifts of crystalline Si.
%\begin{itemize}
%\item Vacancy migrations.... Flexibility of the lattice
% \item volume swelling
% Planes of octahedra (ZrO8) linked by edge-sharing in the basal plane,
% and connected to SiO4 tetrahedra along the c-axis. Then, there is
% more "flexibility" in the c-axis, than in the a-axis.
%
% Expansion of SiO4 tetrahedra, could explain the frequency shifts found in
% the Si-O stretching mode as a function of radiation dose.
%\end{itemize}
\section{Conclusions}
Simulations of high concentration of point defects in ZrSiO$_4$ have been
performed to study the effect of these on the lattice swelling under
radiation damage. Using different sizes for the host crystalline
supercell, we obtain various concentrations of periodically repeated
defects. A roughly linear dependence between the swelling and the
defect concentration has been observed for these ordered defective
crystal. Based on experimental evidence of anisotropic swelling,
%with a larger increase along the $c$ axis ($\sim$2\%) than in the $ab$-plane,
we have selected a set of defects as good candidates to be responsible
for the lattice expansion in
crystalline zircon. Vibrational properties in radiation damaged samples are
also used to justify our conclusions. These include interstitials of oxygen
and silicon, and the anti-site defect Zr$_{\text Si}$.
\section{Acknowledgments}
This work was supported by British Nuclear Fuels (BNFL) and NERC.
We would like to thank E. K. H. Salje, M. T. Dove, K. Trachenko, M. Yang, S. Rios,
I. Farnan for helpful discussions on radiation damage.
\begin{thebibliography}
\bibitem
\bibitem{Weber98}{W. J. Weber {\it et. al.}, J. Mater. Res. {\bf 13}, 1434 (1998).}
\bibitem{Grimes}{K. E. Sickafus, L. Minervini, R. W. Grimes, J. A. Valdez, M. Ishimaru, F. Li, and K. J. McClellan, Science {\bf 289}, 5479 (2000).}
\bibitem{Wilk}{G. D. Wilk and R. M. Wallace, Appl. Phys. Lett. {\bf 76}, 112 (2000).}
\bibitem{Ewing95}{R. C. Ewing, W. Lutze, and W. J. Weber, J. Mater. Res. {\bf 10}, 243 (1995).}
\bibitem{H&G}{H. D. Holland and D. Gottfried, Acta Cryst. {\bf 8}, 291 (1955).}
\bibitem{Weber93}{W. J. Weber, J. Am. Ceram. Soc. {\bf 76}, 1729 (1993).}
\bibitem{Rios}{S. R\'{\i}os, T. Malcherek, E. K. H. Salje, and C. Domeneghetti, Acta Cryst. {\bf B56}, 947 (2000).}
\bibitem{Crocombette}{J.-P. Crocombette, Phys. Chem. Minerals {\bf 27}, 138 (1999).}
\bibitem{PKA}{B. Park, W. J. Weber, L. R. Corrales, Phys. Rev. B {\bf 64}, 174108 (2001).}
\bibitem{siesta} {P. Ordej\'on, E. Artacho, and J. M. Soler, Phys. Rev. B {\bf 53}, 10441 (1996); J. M. Soler, E. Artacho, J. D. Gale, A. Garc{\'{\i}}a, J. Junquera, P. Ordej\'on and D. S\'anchez-Portal, J. Phys.: Condens. Matter {\bf 14}, 2745 (2002).}
\bibitem{DFT}{P. Hohenberg and W. Kohn, Phys. Rev. \textbf{136}, B864 (1964); W. Kohn and L. J. Sham, Phys. Rev. \textbf{140}, A1133 (1965).}
\bibitem{ca}{D. M. Ceperley and B. J. Alder, Phys. Rev. Lett. {\bf 45}, 566 (1980).}
\bibitem{tm2}{N. Troullier and J. L. Martins, Phys. Rev. B {\bf 43}, 1993 (1991).}
\bibitem{kb}{L. Kleinman and D.M. Bylander, Phys. Rev. Lett. {\bf 48}, 1425 (1982).}
\bibitem{basis}{J. Junquera, O. Paz, D. S\'anchez-Portal, and E. Artacho, Phys. Rev. B, {\bf 64}, 235111 (2001).}
\bibitem{optZr}{The basis was optimized to minimize the energy of ZrO$_2$ with a fictitious pressure of 0.2 GPa with the method described in E. Anglada, J. M. Soler, J. Junquera, and E. Artacho, Phys. Rev. B, {\bf 66}, 205101 (2002).}
\bibitem{payne}{G. Makov, and M. C. Payne, Phys. Rev. B {\bf 51}, 4014 (1995).}
\bibitem{Edefects}{M. Pruneda and E. Artacho, (to be published).}
\bibitem{Zhang2001}{M. Zhang and E. K. H. Salje, J. Phys. Condens. Matter {\bf 13}, 3057 (2001).}
\bibitem{LOTO}{X. Gonze and C. Lee, Phys. Rev. B {\bf 55}, 10355 (1997).}
\bibitem{Vanderbilt}{R. D. King-Smith and D. Vanderbilt, Phys. Rev. B {\bf 47}, 1651 (1993).}
\bibitem{Gonze}{G.-M. Rignanese, X. Gonze, A. Pasquarello, Phys. Rev. B {\bf 63}, 104305 (2001).}
\bibitem{latt}{K. Robinson, G. V. Gibbs, P. H. Ribbe, Am. Mineral. {\bf 56}, 782 (1971).}
\bibitem{Nieminen}{A. S. Foster, V. B. Sulimov, F. Lopez Gejo, A. L Shluger and R. M. Nieminen, Phys. Rev. B {\bf 64}, 224108 (2001).}
\bibitem{Zhang2000b}{M. Zhang, E. K. H. Salje, R. C. Ewing, I. Farnan, S. R\'{\i}os, J. Schl\"uter, P. Leggo, J. Phys. Condens. Matter {\bf 12}, 5189 (2000).}
\bibitem{Si:As}{G. S. Cargill, III, J. Angilello, and K. L. Kavanagh, Phys. Rev. Lett. {\bf 61}, 1748 (1988).}
\bibitem{SiC}{F. Gao and W. J. Weber, Nucl. Instrum. and Meth. B {\bf 207}, 10 (2003).}
\bibitem{nmr}{I. Farnan and E. K. H. Salje, J. Appl. Phys. {\bf 89}, 2084 (2001).}
\bibitem{Hafnia}{A. S. Foster, F. Lopez Gejo, A. L Shluger, and R. M. Nieminen, Phys. Rev. B {\bf 65}, 174117 (2002).}
\end{thebibliography}
\end{document}