Comment on ``Comparison of the $Cc$ and $R3c$ space groups for the superlattice phase of Pb(Zr$_{0.52}$Ti$_{0.48}$)O$_3$''
Extracted source text · arXiv cond-mat/0504432
%% ****** Start of file template.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.
%%
%
% This is a template for producing manuscripts for use with REVTEX 4.0
% Copy this file to another name and then work on that file.
% That way, you always have this original template file to use.
%
% 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[aps,prl,preprint,groupedaddress]{revtex4}
%\documentclass[aps,prb,preprint,superscriptaddress]{revtex4}
\documentclass[aps,prb,twocolumn,groupedaddress,showpacs,floatfix]{revtex4}
\usepackage{amsmath}
\usepackage{graphicx}
\usepackage{bm}
%\documentclass[aps,prl,twocolumn,groupedaddress]{revtex4}
% 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}
\title{Comment on ``Comparison of the $Cc$ and $R3c$ space groups
for the superlattice phase of Pb(Zr$_{0.52}$Ti$_{0.48}$)O$_3$''}
% repeat the \author\address pair as needed
\author{J. Frantti}
\affiliation{Materials and Structures Laboratory, Tokyo Institute of
Technology, 4259 Nagatsuta, Midori-ku, Yokohama, 226-8503, Japan}
\email{jfrantt1@msl.titech.ac.jp}
%\thanks{}
\date{\today}
\begin{abstract}
The two-phase (space groups $Cc$ and $Cm$) structural model recently
proposed for the modeling of the neutron powder diffaction pattern
collected on
Pb(Zr$_x$Ti$_{1-x}$)O$_3$ (PZT) sample with $x\approx 0.52$
[R. Ranjan, A. K. Singh, Ragini, and D. Pandey. \textit{Phys. Rev. B}
\textbf{71}, 092101 (2005)] is shown to neglect the $hkl$
dependent (anisotropic) line broadening. The most serious problem with
this model is that it assigns octahedral tilts to a wrong phase.
Instead of correctly taking
the anisotropic line broadening into account, this model used low
symmetry phases to minimize the residuals during Rietveld refinement.
The essential features of a model taking the $hkl$ dependent line
broadening into account are summarized. It has already been reported
that once the anisotropic line broadening, revealed by high resolution
neutron powder diffraction instrument, is correctly taken into account
the model with $R3c$ and $Cm$ space group symmetries describes all
Bragg peaks and their intensities well [J. Frantti, S. Eriksson, S.
Hull, V. Lantto, H. Rundl\"of, and M. Kakihana. \textit{J. Phys.:
Condens. Matter} \textbf{15}, 6031 (2003).]. It was further shown
that the $Cm+R3c$ model
is consistent with the structural features observed at other compositions
and temperatures, which is particularly important in the vicinity of the
phase boundary. Also problems related to sample preparation and data
collection are pointed out.
\end{abstract}
% insert suggested PACS numbers in braces on next line
\pacs{77.84.Dy 61.12.Ld 61.50.Ah 81.30.Dz}
\maketitle
Recently, based on previously published neutron powder diffraction data
\cite{Ranjan3,Hatch}, a model which used two monoclinic phases
(space group symmetries $Cc$ and $Cm$) was proposed to describe
the structure of
Pb(Zr$_x$Ti$_{1-x}$)O$_3$ (PZT) with $x\approx 0.52$ \cite{Ranjan}. This
model assigned a space group symmetries $P4mm$ (at room temperature) or $Cc$
(at low temperature) to the phase which was assigned to $Cm$ symmetry in
refs. \onlinecite{Noheda} and \onlinecite{FranttiPRB}. In addition, the phase,
which traditionally has been assigned to the rhombohedral symmetry, was
assigned to $Cm$ symmetry. To understand the problems related to such a model
and the reason why the use of space group $Cc$ was previously rejected
\cite{FranttiPRB} it is necessary to summarize the crucial role of the $hkl$
dependent line broadening for Rietveld refinement in the case of these
materials. There is a high risk that erroneous space group symmetry
assignments result in once this line broadening is compensated for by
reducing the space group symmetry. For example, one can make the difference
between measured and calculated diffraction intensities arbitrarily small
by simply using space group $P1$ and increasing the primitive cell size
until the difference vanishes below the desired value. In our opinion, a
structure model which
reduces the symmetries to 'model' the $hkl$ dependent line broadening takes
the disorder into account in a wrong way, since in this way the disorder is
assigned to be a periodical disturbance extending through the crystal,
obeying the space group symmetry. However, the $hkl$-dependent line broadening
is \emph{an inherent} property of PZT powders, and is already seen in
tetragonal Ti rich compositions. The same behaviour was observed in a closely
related Pb(Hf$_x$Ti$_{1-x}$)O$_3$ (PHT) system with $0.10\leq x \leq 0.40$,
where the peak widths of the $00l$ reflections were \emph{twice} as large as
the widths of $h00$ reflections\cite{FranttiSub}. It must be emphasized that
no signs of symmetry lowering from the $P4mm$ symmetry was observed in this
high- resolution neutron powder diffraction study. Light scattering
experiments revealed that both in the case of PZT and PHT samples
deviations from the average symmetry were observed
\cite{FranttiPRB2,FranttiJJAP2,FranttiBoston}. Although one could decrease
the average symmetry to take these observations into account, a more
realistic model is to assume that the \emph{average symmetry} over a
length scale of a few hundred nanometers is $P4mm$, and deviations from
this symmetry occur in
a local scale (of a few unit cells). The essential point here,
too, is that these deviations are not periodical! It is also important
to note that the $Cm$ phase was identified by studying the peak split
of the pseudo-cubic $110$ reflections \cite{Noheda}. This peak split
was beyond the resolution of the instrument used in ref.
\onlinecite{Ranjan}
and there is no way to recover the lost information. It does not matter
how many 'new' refinements are carried out for the low-resolution data
collected with insufficient counting time (note that neutron powder
diffraction data in refs. \onlinecite{Ranjan3,Hatch} and
\onlinecite{Ranjan} are the same).
To solve the problem related to average symmetries it is necessary to carry
out experiments with an appropriate instrument and carefully prepared sample.
The role of the spatial composition
variation (which in practice cannot be eliminated) in the vicinity of
the phase boundary is an old and still a valid explanation for the two
phase 'co-existences' and also partially explains the $hkl$-dependent
line broadening. It was also worth of trying to find a simple model which
is consistent with other compositions and
temperatures\cite{FranttiPRB,FranttiJPCM}. These aspects
are shortly reviewed below.
\paragraph{Spatial composition variation.} Traditionally, the
two-phase 'co-existence', observed in room temperature PZT in the
vicinity of the morphotropic phase boundary (MPB), has been
explained by the composition variation. In the context of low
temperature symmetries the role of spatial composition variation was
discussed in refs. \onlinecite{FranttiPRB} and \onlinecite{FranttiJPCM}.
The crystal symmetry of Zr rich PZT ceramics is $R3c$
\cite{Michel,Corker,FranttiJPCM}, except for the compositions with
$x \approx 1$.
Since spatial composition variation cannot be completely eliminated, there
must exist two phases in the vicinity of MPB. To allow the existence of $Cc$
phase necessitates that there should be a narrow region in the $x-T$ plane
were this phase is stable or metastable.
Now, if we were to explain the existence of $Cc$ phase, two phase boundaries
located somewhere between $0.52 \leq x \leq 0.54$ should be assumed to exists,
see Fig. \ref{CompVar} (b).
This in turn, once the spatial composition variation is taken into account,
leads to three phase 'co-existence'. We preferred the simplest $Cm+R3c$
model (corresponding to Fig. \ref{CompVar} (a)), since it was able to
explain the experimental observations in simplest terms. At low temperature
(4 K\cite{FranttiJPCM} or 10 K \cite{FranttiPRB}) the phase fraction of
the $Cm$ phase was monotonically decreasing with $x$ increasing from $0.52$
to $0.54$, which in turn implies that two-phase 'co-existence' is
predominantly due to the spatial composition variation. Other factors
include stresses, which probably exist in the grain boundaries.
\begin{figure}
\includegraphics[width=8.6cm]{Figure1.eps}
\caption{\label{CompVar} The consequences of the spatial composition
variation (at fixed temperature and pressure) in the case of two samples
with average compositions $x_1$ and $x_2$ in the vicinity of (a) the phase
boundary separating $R3c$ and $Cm$ phases (at $X_1$) and (b) two phase
boundaries separating $R3c$ and $Cc$ (at $X_2$) and $Cc$ and $Cm$ phases
(at $X_3$).}
\end{figure}
It is worth to point out that the model proposed in ref.
\onlinecite{Ranjan} was
ruled out also in a recent paper, see the last paragraph
in ref. \onlinecite{Cox}.
After all, the only difference between our model and the model proposed in
ref. \onlinecite{Cox} is that we used $R3c$ symmetry (corresponding to two
lattice parameters and four atomic coordinates), while they used $Cc$
symmetry (corresponding to four lattice parameters and twelve atomic
coordinates, if constraints are not used), a subgroup of $R3c$ space group.
Thus, the model proposed by us and the model proposed in ref.
\onlinecite{Cox} assigned octahedral tilts (evidenced by superlattice
reflections) to the same phase. Octahedral tilts are crucial once the
structural changes versus temperature, composition or pressure are studied.
Somewhat surprisingly, the space
group symmetries of the two phases were rather arbitrarily interchanged in
refs. \onlinecite{Ranjan} and \onlinecite{Ranjan2}. This model
immediately runs into trouble, as the octahedral tilts are assigned to a
wrong phase, see below.
In addition, there
have been reports claiming that dielectric measurements support $Cc$
phase (for example, see ref. \onlinecite{Ranjan3}). In this context we note
that the samples studied in
refs. \onlinecite{Ranjan} and \onlinecite{Ranjan2} contained a nonidentified
impurity phase(s), as was revealed by the peaks at around 28 and 35 two-theta
degrees (corresponding to $d$ spacings 3.18 \AA\mbox{ }and 2.56 \AA,
respectively). This implies that the composition was not well known and
there is no way to avoid compositional and structural inhomogeneities.
Although this non-perovskite phase was clearly observable, it was neglected
in the Rietveld refinement model considered in ref. \onlinecite{Ranjan2},
which in turn results in an error in the structural parameters of the
perovskite
phase(s). This is related to the fact that the
diffraction pattern shown in ref. \onlinecite{Ranjan2} (reported to
have $x=0.52$) is reminiscent to our diffraction pattern with $x=0.53$.
In contrast, our diffraction pattern with $x=0.52$ (see ref.
\onlinecite{FranttiJJAP}) and the diffraction pattern of the $x=0.52$
sample reported in refs. \onlinecite{Cox} and \onlinecite{Noheda} were
reminiscent. It should also be understood that dielectric measurements
do not suit for a space group determination. Although the dielectric
measurements can provide valuable information in the case of single
phase samples, particular care is necessary once multiphase samples are
studied. Anomalies observed in a dielectric constant versus temperature
measured from a \emph{two or three phase} ceramic bulk samples
(see ref. \onlinecite{Ragini3}) do
not provide reliable evidence that one of the phases has
undergone a phase transition from $Cm$ to $Cc$ phase.
\paragraph{Anisotropic line broadening.} The anisotropic line broadening
was ascribed to the spatial
composition variation resulting in 'microstrain' (\emph{i.e.}, Zr
substitution for Ti creates local strains which in turn contributes
to the $hkl$ dependent line broadening)\cite{FranttiJPCM}. In
addition, rather strong diffuse scattering is commonly
observed between the $h00$ and $00h$ Bragg reflections, which was assigned to
Pb ions displaced toward $\langle 110 \rangle$ directions. This latter feature
is somewhat puzzling, as it adds intensity to certain diffraction peaks, which
results in \emph{an asymmetric profile}. For example, the larger and smaller
$d$-spacings sides of the pseudo-cubic $200$ and $002$\mbox{ }reflections,
respectively, gain intensity. In ref. \onlinecite{Ranjan} $Cm$ phase was
practically used to model the intensity not only due to the crystalline phase,
but also due to the diffuse scattering and due to the main crystalline phase.
Despite the increased number of refined parameters, the residuals were still
rather high and the differences between different models compared were
marginal (for instance, $R_{exp}$ was lowest for the $R3c+Cm$ model which
was rejected in ref. \onlinecite{Ranjan}).
Although a reasonable structure refinement for medium resolution data
collected on PZT samples with $x\leq 0.50$ could be obtained using a
lineshape which ignores anisotropic line broadening, the situation was
quite different for high resolution facilities, particularly once the
data was collected on two phase samples (the present case). In such
a case it was essential to use an appropriate profile function.
\emph{Now, if the anisotropic line broadening is neglected}, the
fit in the case of certain weak peaks becomes slightly worse (as was
observed to be the case of weak superlattice reflections, which are
less weighted in the refinements). The reason for this is illustrated
in Fig. \ref{Asymmetry}:
in order to improve the fit corresponding to the strong pseudo-cubic
$200$ reflections, during the refinement the position of $R3c$
reflection is shifted toward higher $d$-spacings, which in turn
resulted in a small shift of the weak peaks at 2.44 and 1.06
\AA\mbox{ }(in a case of the data shown in ref. \onlinecite{Ranjan}
the latter peak was almost at the level of noise and the $Cc+Cm$ model
assigns more reflections than there are data points in this region).
The anisotropic
line broadening was not limited to the $d$-spacing at around 2 \AA,
and similar mechanism was seen at other $d$-spacings.
\begin{figure}
\includegraphics[width=8.6cm]{Figure2.eps}
\caption{\label{Asymmetry} Left panel: Schematic illustration of the
way how the neglect of anisotropic line broadening affects the
position of the rhombohedral peaks: once the rightmost peak broadens,
it pushes the rhombohedral peak (middle) towards smaller $d$-spacing.
Righ panel: Observed and computed high resolution neutron powder
diffraction profile collected on PZT with $x=0.53$ at 4 K. Pseudo-cubic
$200$\mbox{ }reflection region is shown. Note that the $002_M$ peak
($Cm$ phase) at around 2.06 \AA\mbox{ }is significantly broader than
the $220_M$ ($Cm$ phase) peak at around 2.02 \AA. Now, the position
of the rhombohedral $20\bar{4}_R$ peak depends on the way the line
broadening is taken into account.}
\end{figure}
\paragraph{Previous high resolution neutron powder diffraction results.}
The anisotropic line broadening was the reason to carry out a
subsequent study using a high resolution neutron powder diffraction
instrument (which reveals the anisotropic line broadening particularly
clearly and
allows a distinction between average symmetry lowering and a defect
related anisotropic line broadening to be done, in contrast to the
low- or medium resolution constant wavelength neutron diffraction
facilities)\cite{FranttiJPCM}. In addition, we increased Zr content,
$x=0.54$\cite{FranttiJPCM} (although it definitely was possible to
fit \emph{all} reflections using the $Cm+R3c$ model also in the
case of $x=0.52$ and $x=0.53$\mbox{ }samples). This allowed a more
reliable refinement and symmetry identification to be done by studying
the \emph{changes in phase fractions versus temperature}. Importantly,
for these compositions it has been found that oxygen octahedra tilts
increase with increasing $x$ (ref. \onlinecite{Corker}) and decreasing
temperature (this feature is discussed in refs. \onlinecite{Thomas}
and \onlinecite{FranttiPRB}). We also note that previously $Cc$
was proposed to be space group symmetry corresponding to high
isotropic pressure\cite{Haines}, whereas
our work was concentrating on the determination of space group
symmetries
versus composition and temperature at ambient pressure. Now, PZT
sample with $x=0.54$ provided a test for
clarifying which phase is the preferred one at low temperature. To
model the peak profiles, GSAS\cite{GSAS} lineshape 4 by Stephens
\cite{Stephens} was used. As was seen in ref. \onlinecite{FranttiJPCM},
once the crystal contracts with decreasing temperature the $R3c$ phase
was favoured (and $Cm$ phase did not transform to $Cc$ phase, as one
could expect from the model proposed in ref. \onlinecite{Ranjan}).
Thus, when the oxygen octahedra had almost no chance to contract,
they were tilt towards opposite directions along the pseudo-cubic
$111$ axis (tilt system $a^-a^-a^-$). This tilt results in a decrease
of the volume of cuboctahedra around Pb ions. Changes evidencing
$Cc$ symmetry (and the corresponding tilt system) were not observed.
\emph{Indeed, the phase fraction of $Cm$ phase (which was assigned
to $Cc$ symmetry in ref. \onlinecite{Ranjan}) decreased with
decreasing temperature, whereas $R3c$ phase fraction significantly
increased with decreasing temperature. This can be confirmed even by a
naked eye, although the results of refinements were also given. The
intensity of the peak at around 1.06 \AA\mbox{ }was increasing with
decreasing temperature, which further confirmed that its origin is
the phase assigned to $R3c$ phase. Also the peak at around
1.06 \AA\mbox{ }was well fit by $R3c$ symmetry}.
As a summary, we do not find support for $Cm+Cc$ model proposed in ref.
\onlinecite{Ranjan}. Instead, we still consider that $Cm+R3c$ provides
a sufficient low temperature structural model for PZT with composition
in the vicinity of the morphotropic phase boundary.
%\begin{acknowledgments}
%\end{acknowledgments}
\begin{thebibliography}{99}
\bibitem{Ranjan3} R. Ranjan, S. K. Mishra, D. Pandey and K. Kennedy.
\textit{Phys. Rev. B} \textbf{65}, 060102 (2002).
\bibitem{Hatch} D. M. Hatch, H. T. Stokes, R. Ranjan, Ragini,
S. K. Mishra, D. Pandey, and B. J. Kennedy,
\textit{Phys. Rev. B} \textbf{65}, 212101 (2002).
\bibitem{Ranjan} R. Ranjan, A. K. Singh, Ragini, and D. Pandey.
\textit{Phys. Rev. B} \textbf{71}, 092101 (2005).
\bibitem{Noheda} B. Noheda, D. E. Cox, G. Shirane, J. A. Gonzalo,
L. E. Cross and S-E. Park. \textit{Appl. Phys. Lett.} \textbf{74},
2059 (1999); B. Noheda, J. A. Gonzalo, L. E. Cross, R. Guo, S-E. Park,
D. E. Cox, and G. Shirane. \textit{Phys. Rev. B} \textbf{61}, 8687 (2000).
\bibitem{FranttiPRB} J. Frantti, S. Ivanov, S. Eriksson, H. Rundl\"of,
V. Lantto, J. Lappalainen, and M. Kakihana. \textit{Phys. Rev. B}
\textbf{66}, 064108 (2002).
\bibitem{FranttiSub} J. Frantti, Y. Fujioka, S. Eriksson, S. Hull and
M. Kakihana. Neutron powder diffraction study of Pb(Hf$_x$Ti$_{1-x}$)O$_3$
ceramics ($0.10 \leq x \leq 0.50$), submitted.
\bibitem{FranttiPRB2} J. Frantti, V. Lantto, S. Nishio and M. Kakihana.
\textit{Phys. Rev. B} \textbf{59}, 12 (1999).
\bibitem {FranttiJJAP2} J. Frantti, J. Lappalainen, S. Eriksson,
V. Lantto, S. Nishio, M. Kakihana, S. Ivanov, and H. Rundl\"of.
\textit{Jpn. J. Appl. Phys.} \textbf{38}, 5679 (1999).
\bibitem{FranttiBoston} J. Frantti, Y. Fujioka, S. Eriksson, V. Lantto,
and M. Kakihana. \textit{Journal of Electroceramics} \textbf{13}, 299 (2004).
\bibitem{Ranjan2} Ragini, R. Ranjan, S. K. Mishra, and D. Pandey,
\textit{J. Appl. Phys.} \textbf{92}, 3266 (2002).
\bibitem {FranttiJJAP} J. Frantti, J. Lappalainen, S. Eriksson,
V. Lantto, S. Nishio, M. Kakihana, S. Ivanov, and H. Rundl\"of.
\textit{Jpn. J. Appl. Phys.} \textbf{39}, 5697 (2000).
\bibitem{FranttiJPCM} J. Frantti, S. Eriksson, S. Hull, V. Lantto,
H. Rundl\"of, and M. Kakihana. \textit{J. Phys.: Condens. Matter}
\textbf{15}, 6031 (2003).
\bibitem{Michel} C. Michel, J. M. Moreau, G. D. Achenbach, R. Gerson, and
W. J. James, \textit{Solid State Commun.} \textbf{7}, 865 (1969).
\bibitem{Corker} D. L. Corker, A. M. Glazer, R. W. Whatmore, A. Stallard,
and F. Fauth. \textit{J. Phys. Condens. Matter} \textbf{10}, 6251 (1998).
\bibitem{Cox} D. E. Cox, B. Noheda, and G. Shirane. \textit{Phys. Rev. B}
\textbf{71}, 134110 (2005).
\bibitem{Haines} J. Rouquette, J. Haines, V. Bornand, M. Pintard, Ph.
Papet, W. G. Marshall, and S. Hull. \textit{Phys. Rev. B} \textbf{71},
024112 (2005).
\bibitem{Ragini3} Ragini, S. K. Mishra, D. Pandey, H. Lemmens, and G.
Van Tendeloo. \textit{Phys. Rev. B} \textbf{64},
054101 (2001).
\bibitem{Thomas} N. W. Thomas and A. Beitollahi.
\textit{Acta Crystallorg., Sect. B: Struct. Sci.} \textbf{50}, 549 (1994).
\bibitem{GSAS} A. C. Larson and R. B. Von Dreele \textit{General
Structure Analysis System}, LANSCE MS-H805, Los Alamos National
Laboratory, Los Alamos, NM 87545 (2000).
\bibitem{Stephens} P. W. Stephens. J. Appl. Crystallogr. \textbf{32}
281 (1999).
\end{thebibliography}
\end{document}