cond-mat0608074/klt1.tex
1: %\documentclass[aps,prb,preprint,superscriptaddress]{revtex4}
2: %\documentclass[aps,prb,twocolumn,superscriptaddress]{revtex4}
3: %\documentclass[aps,prb,preprint]{revtex4}
4: \documentclass[aps,prb,twocolumn]{revtex4}
5: 
6: %\usepackage[textures]{graphicx,color}
7: %\usepackage[textures]{graphicx}
8: 
9: \usepackage[dvips]{graphicx,color}
10: 
11: %\setlength{\topmargin}{0mm}
12: 
13: \input epsf.sty
14: %\usepackage{graphicx}
15: 
16: \bibliographystyle{apsrev}
17: 
18: \begin{document}
19: 
20: %\preprint{Version 1 @ 2006/04/06}
21: 
22: 
23: %
24: %Title of paper
25: %
26: 
27: \title{Dielectric properties and lattice dynamics of 
28: Ca-doped K$_{0.95}$Li$_{0.05}$TaO$_{3}$}
29: 
30: \author{S. Wakimoto}
31: \email[Corresponding author: ]{wakimoto.shuichi@jaea.go.jp}
32: \affiliation{ Quantum Beam Science Directorate, 
33:    Japan Atomic Energy Agency, Tokai, Ibaraki 319-1195, Japan. }
34: 
35: \author{G. A. Samara, R. K. Grubbs and E. L. Venturini}
36: \affiliation{ Sandia National Laboratories, 
37:    Albuquerque, New Mexico 87185, USA }
38: 
39: %\author{E. L. Venturini}
40: %\affiliation{ Sandia National Laboratories, 
41: %   Albuquerque, New Mexico 87185, USA }
42: 
43: \author{L. A. Boatner}
44: \affiliation{ Oak Ridge National Laboratory, 
45:    Oak Ridge, Tennessee 37831, USA }
46: 
47: \author{G. Xu and G. Shirane~\footnote{Deceased.}}
48: \affiliation{ Department of Physics, Brookhaven National Laboratory,
49:    Upton, New York 11973, USA }
50: 
51: %\author{G. Shirane}
52: %\affiliation{ Department of Physics, Brookhaven National Laboratory,
53: %   Upton, New York 11973, USA }
54: 
55: \author{S.-H. Lee}
56: \affiliation{ NIST Center for Neutron Research, National Institute of
57:   Standards and Technology, Gaithersburg, MD 20899 }
58: \affiliation{ Department of Physics, University of Virginia, 
59:   Charlottesville, VA22904, USA }
60: 
61: 
62: \date{\today}
63: 
64: 
65: \begin{abstract}
66: 
67: %%% Ver 1 %%%%%%%%%%%%%%%%
68: Relaxor behavior and lattice dynamics have been studied by employing 
69: dielectric measurements and neutron scattering methods 
70: for a single crystal of K$_{1-x}$Li$_x$TaO$_3$ $(x=0.05)$, 
71: where a small amount of a Ca impurity ($\sim 15$~ppm) 
72: was incorporated during the single crystal growth procedure.  
73: The dielectric constant $\epsilon'(\omega, T)$ 
74: shows qualitatively similar behavior to that of Ca-free KLT 
75: with $x=0.043$ with both compositions exhibiting relaxational
76: properties with no evidence for a ferroelectric transition.
77: The absolute value of $\epsilon'(\omega, T=0)$ for the present 
78: crystal is larger by an order of magnitude than that of the 
79: Ca-free sample due to charge carriers 
80: induced by the Ca doping.  
81: %
82: This large value is shown to be due to a Maxwell-Wagner 
83: relaxation process associated with the low temperature ($< 8$~K) 
84: activation of frozen electronic carriers.
85: %
86: The dielectric loss tangent 
87: $\tan \delta$ reveals three Debye-like relaxations with 
88: Arrhenius activation energies of $80$, $135$, and $240$~meV 
89: that are assigned to Li$^+$ dipoles, Ca$^{2+}$-related relaxation 
90: and the Li$^+$-Li$^+$ dipolar pairs, respectively.
91: %
92: In the neutron scattering results, diffuse scattering ridges appear around the 
93: nuclear Bragg peaks along the [100] direction below $\sim 150$~K 
94: and phonon line broadening features start to appear at even higher 
95: temperatures suggesting that polar nano-regions 
96: (PNR's) start to form at these temperatures.
97: %
98: These results are supported by the dielectric data that reveal 
99: relaxor behavior starting at $\sim 200$~K on cooling.
100: %
101: From analyses of the diffuse intensities at different zones, 
102: we have derived atomic displacements in the PNR's.  The results 
103: suggest that the displacements include a uniform 
104: phase shift of all of the atoms in addition to the atomic 
105: displacements corresponding to a polarization vector of the 
106: transverse optic soft ferroelectric mode, a finding that is 
107: analogous to that in the prototypical relaxor material 
108: Pb(Mg$_{1/3}$Nb$_{2/3}$)O$_3$.  
109: %%%%%%%%%%%%%%%%%%%%%%%%%%
110: 
111: 
112: 
113: \end{abstract}
114: 
115: \pacs{}
116: 
117: \maketitle
118: 
119: 
120: 
121: 
122: \section{Introduction}
123: 
124: Relaxor ferroelectrics that have a relatively large dielectric 
125: permittivity over a wide temperature range have attracted a significant amount of 
126: attention due to their high potential for device applications 
127: as well as their scientific challenges.  It is now widely believed 
128: that relaxor behavior, whose universal signature is a frequency($\omega$)-dependent 
129: peak in the dielectric constant $\epsilon'(\omega, T)$, arises from the relaxational 
130: nature of randomly-oriented polarized nano-regions (PNR's).  
131: %
132: However, the mechanism for the PNR formation is not fully understood 
133: yet.  In particular, unlike the case for displacive ferroelectric materials, 
134: possible contribution of the zone-center transverse optic (TO) soft 
135: phonons to the PNR condensation still remains to be clarified.
136: 
137: 
138: The relaxor material system K$_{1-x}$Li$_x$TaO$_3$ (KLT($x$)) provides a unique 
139: opportunity for systematic research ranging from a relaxor state in the material at low Li
140: concentrations to a ferroelectric state at high concentrations.  
141: The parent compound KTaO$_3$ is an incipient ferroelectric 
142: material that does not undergo a ferroelectric 
143: transition~\cite{Wemple_65} although the TO soft 
144: mode softens remarkably at low temperature.~\cite{Axe_70}  On substituting at the K$^{+}$ 
145: site, the much smaller Li$^{+}$ ion occupies an off-center 
146: positions resulting in a large dipole moment that is the precursor for
147: PNR formation on cooling.  At high Li concentration, the PNR's percolate, 
148: and the system achieves ferroelectric order.
149: 
150: Although the system KLT($x$) has been studied extensively by 
151: Raman scattering,~\cite{Raman,Vogt_new,Prater_81} and its soft mode 
152: character is well-established, only a few neutron scattering studies have been 
153: carried out mainly to explore the role of the soft mode of the host crystal 
154: in the formation and growth of the PNR's.
155: %
156: Yong {\it et al.}~\cite{Toulouse_00} have studied the 
157: diffuse neutron scattering in KLT that originates from 
158: PNR formation.  
159: %
160: They reported a strong diffuse intensity at (110) and its absence 
161: at both (100) and (200).
162: %
163: If we assume a PNR condensation from the TO soft mode as in 
164: the case of the ordinary ferroelectrics, then the structure factors 
165: for the diffuse and the inelastic scattering by the TO mode should agree.
166: % 
167: However, it is found that the inelastic structure factor for the TO soft 
168: mode of KLT at (200) is much larger than that at (110) in 
169: disagreement with the diffuse scattering intensities.  
170: 
171: 
172: Such disagreements 
173: between the diffuse intensities and inelastic structure factors 
174: have also been reported in the relaxor material
175: Pb(Mg$_{1/3}$Nb$_{2/3}$)O$_3$ (PMN).~\cite{Naberezhnov_99}
176: %
177: Moreover, in this latter case, the atomic displacements determined 
178: from the diffuse intensities at several zones do not conserve the 
179: center-of-mass (CM).~\cite{Vakhrushev_95}
180: %
181: These facts suggest an apparent disconnection between the PNR 
182: condensation and the soft phonon mode in PMN.  However, 
183: Hirota {\it et al.}~\cite{Hirota_02} have reconciled these 
184: two observations by introducing a model of the soft mode condensation 
185: that incorporates an {\it additional uniform phase shift}, in which the total 
186: atomic displacements can be divided into two components.  
187: One component originates from the normal condensation of the TO 
188: soft mode that conserves the CM, and the other component represents a 
189: uniform shift of all of the atoms in the polarization direction.    
190: 
191: 
192: 
193: In the present paper, we report and discuss dielectric 
194: and neutron scattering experiments performed on KLT with 
195: $x=0.05$.
196: %
197: The results provide detailed views of the response and physics of 
198: this crystal.  An additional objective is to determine if the 
199: anomalous lattice dynamical features observed in the prototypical 
200: relaxor PMN are common to other ABO$_3$ relaxors.  This comparison 
201: offers us a unique opportunity to compare two different types of 
202: relaxors, one is KLT in which 
203: off-center impurity ions trigger a local polarization (PNR's) on the 
204: incipient ferroelectric background, and the other is PMN in which 
205: random occupation of the $B$-sites by two different ions with different 
206: valences produces the disorder that breaks long-range ferroelectric (FE) 
207: correlations and order resulting in the formation of a relaxor (R) state.  
208: 
209: %Our colleague and co-author Gen Shirane passed away after this work was started.  We dedicate this paper to his memory.
210: 
211: 
212: 
213: \section{Experimental details}
214: 
215: 
216: %%% Dielectric part %%%%%%%%%%%%%%
217: The KLT samples used in the present study were $<100>$ oriented 
218: single crystals cut from one boule that was grown by 
219: solidification from the melt.  The boule was nominally undoped, 
220: but our dielectric measurements (to be presented below) suggested 
221: the presence of an unintentional dopant leading to some electronic 
222: conduction and a high dielectric loss.  Subsequent analysis 
223: revealed the presence of 
224: %
225: only
226: %
227: Ca impurity (at a level of $\sim 15$~ppm) 
228: in the starting 
229: powder used in the crystal growth.~\cite{Lynn_note}  Clearly, some of the Ca was 
230: incorporated into the crystal, but at a sufficiently low level 
231: that was difficult to quantify.  The crystal was colorless 
232: with no indication of any significant free carrier absorption 
233: that usually produces a blue coloration in more highly conducting crystals of this type.  
234: Although the Ca dopant had a definite influence on the dielectric 
235: properties, normally it would not be expected to have a significant influence 
236: on the lattice dynamics measured by neutron scattering.
237: 
238: It is well established that not all of the Li in the starting 
239: mixture of the oxides is incorporated into the growing crystal during 
240: the solidification process due to segregation coefficient effects.  
241: It is estimated that 35~\% of the Li content in the growth charge 
242: is incorporated.~\cite{Klink_84}
243: The Li composition of our KLT crystal ($\simeq 5$ at.\%) was estimated 
244: from the amount of Li in the melt and from the established 
245: relationship between the peak temperature in the $\epsilon'(T)$ 
246: response and the Li content.~\cite{Samara_SSP}  Henceforth we shall 
247: designate our composition as KLT(5):Ca.
248: 
249: For the dielectric measurements, the large (100) faces of the crystal 
250: were polished using colloidal silica and were subsequently 
251: covered with vapor-deposited chromium followed by gold.  
252: The sample was investigated by dielectric spectroscopy with 
253: measurements of the real ($\epsilon'$) and loss ($\tan \delta$) 
254: parts of the dielectric function performed as functions of 
255: temperature ($4 - 300$~K), frequency ($10^2 - 10^6$~Hz) and 
256: hydrostatic pressure ($0 - 7.5$~kbar).  The pressure system 
257: consisted of a compressor and an intensifier that fed the 
258: pressure-transmitting medium (helium gas) into a cell placed 
259: in a low temperature Dewar.  The pressure was measured using a 
260: calibrated manganin gauge to an accuracy of better than 4\%. 
261: 
262: 
263: %%%% Wakimoto part %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
264: The neutron scattering experiments were performed using triple 
265: axis spectrometers SPINS at NIST, LTAS at JAEA, and TAS1 at 
266: JAEA.  The diffuse scattering was measured at SPINS and LTAS 
267: with an incident neutron energy of $E_i=5$~meV ($\lambda=4.04$~\AA) 
268: and a collimation sequence: Guide-80$'$-S-80$'$-open (S denotes 
269: the sample), and also at TAS1 with $E_i=14.7$~meV ($\lambda=2.36$~\AA) 
270: and 40$'$-40$'$-S-40$'$-80$'$.  The phonon modes were studied at 
271: TAS1 with a fixed final energy of $E_f=14.7$~meV and 
272: 40$'$-40$'$-S-80$'$-open.  Beryllium and pyrolytic graphite filters 
273: were utilized for the measurements with $\lambda=4.04$~\AA~ and 
274: $\lambda=2.36$~\AA, respectively, to remove those neutrons with higher 
275: harmonic wavelength $(\lambda/2, \lambda/3, etc.).$
276: 
277: A KLT(5):Ca crystal with dimensions of $6 \times 10 \times 
278: 13$~mm$^3$, cut from the same crystal boule that was used for 
279: the dielectric measurements, was prepared for the neutron 
280: scattering experiments.  The crystal was mounted in an unstrained manner on an 
281: aluminum post using aluminum foil, and the aluminum 
282: post was then masked by a cadmium plate to avoid any scattering 
283: from it.  The sample was sealed in an aluminum can filled 
284: with He gas, and placed in a He-gas closed-cycle refrigerator 
285: with the [100] and [010] axes in the horizontal scattering plane.
286: %
287: In the temperature range $10 \leq T \leq 295$~K, the crystal 
288: structure was cubic as evidenced 
289: in the neutron scattering resolution, with a lattice constant 
290: 3.95~\AA, corresponding to a reciprocal lattice unit (r.l.u.) of 
291: 1.59~\AA$^{-1}$.
292: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
293: 
294: 
295: \section{${\rm \bf Li}$ and ${\rm \bf Ca}$ substitution in ${\rm \bf KTaO}_3$}
296: 
297: Before presenting the dielectric and lattice dynamics 
298: properties, we briefly review in this section the effects of 
299: Li and Ca substitution on KTaO$_3$.
300: 
301: The substitution of Li$^{+}$ for K$^{+}$ in KLT raises an 
302: immediate issue.  The K$^{+}$ ion has a radius $r_i$ of 1.3~\AA~ and 
303: fits well in the lattice, occupying a centrosymmetric position 
304: in the oxygen ion cage that surrounds it in KTaO$_3$ (Fig. 1 (a)).  
305: The Li$^{+}$ ion, on the other hand, is too small ($r_i = 0.68$\AA), 
306: and the oxygen cage dimension is too large in comparison.  This is a circumstance that 
307: occurs relatively often in solids with the expected consequence 
308: that the small ion will shift to an off-center position.  
309: This is expected for Li in KLT, and indeed, this effect was observed early 
310: on.~\cite{Samara_SSP,Hochli_90,Yacoby_74}  The shift, as expected 
311: from the lattice symmetry, is in the [100] direction (Fig. 1 (b)) and 
312: is quite large (1.2~\AA).~\cite{Hochli_90}  
313: In this off-center position, the Li$^{+}$ ion has a relatively 
314: large dipole moment, which can interact with neighboring Li dipoles.  
315: Additionally, because of the high polarizability of the KTaO$_3$ host, 
316: the Li dipole can polarize adjacent unit cells resulting in polar 
317: nano-regions.  These are precisely the conditions that lead to 
318: dipolar glass and relaxor behavior.  Indeed, this is what is 
319: observed experimentally.~\cite{Samara_SSP,Hochli_90}
320: 
321: %dfigure1%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
322: \begin{figure}
323: \centerline{\epsfxsize=3.3in\epsfbox{figure01.eps}}
324: \caption{(a) The ABO$_3$ structure of KLT showing the off-center 
325: position of the Li ion at the K site.  The Li$^{+}$ hops among 6 
326: equivalent positions, 4 of which are shown in (b).  (c) shows the 
327: configuration of the co-linear Li$^{+}$-Li$^{+}$ ion pairs.}
328: \end{figure}
329: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
330: 
331: 
332: Leung~\cite{Leung_01} has performed a first-principles study of 
333: Ca$^{2+}$ substitution in KTaO$_3$.  Although his study is for 
334: KTaO$_3$ and not on KLT, we believe that the results are relevant 
335: to our Ca-doped KLT crystals.  The results show that Ca$^{2+}$ 
336: substitutes at both the K$^{+}$ and Ta$^{5+}$ sites.  
337: At the K$^{+}$ site, the Ca antisite, Ca$_{\rm K}^{+}$, would be 
338: expected to produce a K vacancy, but LeungÕs results find this 
339: circumstance highly unfavorable energetically.  Thus, the Ca$^{2+}$ 
340: simply donates an electron.  At the Ta$^{5+}$ site, the Ca$^{2+}$ 
341: binds to an oxygen vacancy (V$_{\rm O}$), thereby forcing the Ca$^{2+}$ 
342: off-center and resulting in a dipolar complex.  (Oxygen vacancies 
343: are formed on Ca$^{2+}$ substitution to preserve charge neutrality, 
344: and they are also native defects in perovskites).  The calculations 
345: predict that the activation energy for the reorientation of the 
346: (Ca-V$_{\rm O}$) dipole via oxygen-vacancy hopping within the first neighbor 
347: shell exceeds 2 eV.  This large energy precludes the direct 
348: involvement of the (Ca-V$_{\rm O}$) complex in the observed new relaxations 
349: that have much lower activation energies in the present KLT crystals. 
350: (See the next section.) 
351: 
352: Early work~\cite{Senhouse_66,Wemple_65} on Ca-doped KTaO$_3$ (KT) 
353: has suggested that the Ca is fully ionized at room temperature.  
354: Presuming that Ca$^{2+}$ substitutes at the K$^{+}$ site, Senhouse 
355: {\it et al.}~\cite{Senhouse_66} showed that the net ionized donor 
356: concentration is equal to the Ca concentration.  These results would 
357: suggest that most of the Ca$^{2+}$ resides at the K$^{+}$ site.  
358: However, impurities and defects (primarily oxygen vacancies that are 
359: prominent defects in ABO$_3$ oxides and in Ca-doped KT~\cite{Yacoby_74}) 
360: can also be sources of free electrons.  Wemple~\cite{Wemple_65} also 
361: showed that there is no ``freeze-out'' of the Ca-generated electrons in 
362: KT down to temperatures approaching liquid helium temperature.
363: 
364: When present in sufficient concentrations ($> \sim 10^{17} /cm^{3}$) 
365: the Ca-generated free electrons in Ca-doped KT and related crystals 
366: produce free carrier absorption and an associated blue coloration.  
367: In our present KLT(5):Ca crystal the Ca was present in small amounts, 
368: and consequently these crystals were colorless - containing relatively few free 
369: electrons and no noticeable free carrier absorption.  
370: %
371: However, this small amount of Ca had a significant influence on the 
372: amplitude of dielectric response and also a minor influence on the 
373: lattice dynamics.  (See later discussion.)
374: 
375: 
376: \section{Results and discussion}
377: 
378: \subsection{Dielectric response}
379: 
380: Figure 2 shows the dielectric response of our KLT(5):Ca crystal at 1~bar.  
381: The inset in Fig. 2(a) shows $\epsilon'(\omega,T)$ for an undoped 
382: 4.3 at.\% Li KLT crystal~\cite{Prosandeev_01} - a composition whose 
383: dielectric response is seen to be very close to KLT(5).  The main 
384: overlapping peaks in this relaxational response are associated 
385: with the relaxations of isolated Li$^{+}$ dipoles (at the lower 
386: temperature) and with (Li$^{+}$ - Li$^{+}$) ion pairs with kinetic 
387: parameters $E \sim 80 (240)$~meV and $\omega_o \sim 1.5 \times 
388: 10^{12}$~s$^{-1}$ ($7 \times 10^{14}$~s$^{-1}$), 
389: respectively.~\cite{Prosandeev_01,Doussineau_93,Toulouse_98JKPS}  
390: The relaxational dielectric response of our KLT(5):Ca crystal in 
391: Fig. 2(a) has some similarities to that in the inset, but there 
392: are important differences.
393: 
394: 
395: %dfigure2%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
396: \begin{figure}
397: \centerline{\epsfxsize=3.3in\epsfbox{figure02.eps}}
398: \caption{The 1~bar $\epsilon'$ and $\tan \delta$ responses of the 
399: KLT(5):Ca crystal.  The inset is the response of undoped KLT(4.3) 
400: according to Prosandeev {\it et al.} (Ref.~\onlinecite{Prosandeev_01}.)  
401: Three primary relaxations denoted by A, B and C are noted in (b).}
402: \end{figure}
403: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
404: 
405: 
406: Starting at low temperatures, we note that the amplitude of $\epsilon'$ 
407: below $\sim 20$~K ($\sim 6000$) is over an order of magnitude 
408: larger than that of the undoped KLT crystal of about the same Li 
409: concentration as shown in the inset in Fig. 2 (a).  This strong low-$T$ 
410: enhancement of $\epsilon'$ in KLT(5):Ca is due to the Ca$^{2+}$ 
411: doping.  A feature in the KLT(5):Ca data in Fig. 2, namely a 
412: relaxational feature below 10~K, points to the mechanism for 
413: this enhancement.  This feature is shown in Fig. 3.  The onset 
414: of this relaxation is reflected in the simultaneous highly 
415: dispersive decrease in $\epsilon'$ and increase in $\tan \delta$ 
416: at 1~bar.  This relaxation is accompanied by strong frequency 
417: dispersion in both $\epsilon'(T)$ and $\tan \delta(T)$, a step 
418: in $\epsilon'(T)$ with $\epsilon'$ increasing from $\sim 600$ 
419: at the lowest temperatures (the expected value for this composition in 
420: the absence of Ca [See inset in Fig. 2 (a)]) to $\sim 6000$ at 
421: the plateau region, and a large peak in $\tan \delta(T)$.  These 
422: features are the signature of a capacitive barrier layer that, 
423: for our present case, could be due to either a Schottky barrier 
424: or space charge accumulation at the metal contacts.~\cite{Grubbs_05}
425: 
426: 
427: %dfigure3%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
428: \begin{figure}
429: \centerline{\epsfxsize=3.3in\epsfbox{figure03.eps}}
430: \caption{Low $T$ dielectric response at 1 bar and 7.5 kbar hints 
431: at the onset of a low $T$ ($< 8$~K) relaxation in KLT(5):Ca.}
432: \end{figure}
433: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
434: 
435: 
436: We speculate that the relaxation is, in fact, associated with the 
437: activation of the Ca-donated free carriers (electrons).  
438: Specifically, the carriers are frozen at the lowest temperatures and are 
439: activated at the onset of the relaxation.  To check this hypothesis, 
440: we have measured the dc resistivity ($\rho$) of our crystal (using In/Ga 
441: eutectic as contacts).  The results in Fig. 4 show that above the 
442: relaxation: $\rho < 10^5$ $\Omega$ cm, but that it increases sharply by 
443: over 3 orders of magnitude below 8~K, 
444: clear evidence of carrier freeze-out.  These results also 
445: support WempleÕs~\cite{Wemple_65} early finding for KTaO$_3$:Ca 
446: that carrier freeze-out occurs only at the lowest temperatures.  
447: The dielectric measurements were repeated at a pressure of 7.5~kbar.  
448: Pressure suppresses the frequency dispersion and shifts the 
449: relaxation to lower temperatures.  It also increases the 
450: amplitude of $\epsilon'(T)$ in the plateau region by 15~\% under 
451: 7.5~kbar.  In early high pressure work on KT and other perovskites, 
452: Wemple {\it et al.}~\cite{Wemple_private} concluded that for these 
453: materials the electron mobility is related to $\epsilon'$ by the 
454: relationship $\mu \propto 1/\epsilon' + constant$.  
455: Since $\epsilon'$ decreases with pressure as a result of the 
456: increase in soft mode frequency, $\mu$ can be expected to increase 
457: leading to an increase in the conductivity.  It is this increase 
458: in sample conductivity that is responsible for the increase in 
459: the amplitude of $\epsilon'(T)$ in the plateau region with 
460: pressure.  Simply stated, the higher the conductivity of the 
461: sample, the greater is the conductivity mismatch at the barrier 
462: layer and the higher is the relaxation step in $\epsilon'$.~\cite{Grubbs_05}
463: 
464: 
465: %dfigure4%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
466: \begin{figure}
467: \centerline{\epsfxsize=3.3in\epsfbox{figure04.eps}}
468: \caption{The temperature dependence of the resistivity, $\rho(T)$, 
469: of KLT(5):Ca showing carrier (electron) freeze-out below 10~K.  
470: It is the un-freezing of the carriers on heating that triggers 
471: the Maxwell-Wagner relaxation that results in the $\epsilon' = 6000$ 
472: plateau in Figs. 2 and 3.}
473: \end{figure}
474: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
475: 
476: 
477: A similarity between the results shown in Fig. 2 (a) for KLT(5):Ca and 
478: the undoped KLT(4.3) in the inset is represented by the shape of the $\epsilon'(T)$ 
479: response with its overlapping peaks, despite the fact that the 
480: absolute amplitudes of $\epsilon'(T)$ for the two materials are 
481: vastly different.  In fact, the response in Fig. 2 (a) is 
482: qualitatively like that of undoped KLT(5), but ``sitting on top of'' an 
483: $\epsilon' \sim 6000$~ baseline.  Remarkably, for the undoped 
484: crystal in the inset, $\epsilon'_{max}$ at the relaxational peak 
485: is $\sim 6 \times$ its value at low temperatures, and this ratio is about 
486: the same for the KLT(5):Ca crystal.  Pressure shifts the 
487: $\epsilon'(\omega,T)$ response to lower temperatures and sharpens 
488: it somewhat maintaining its relaxor character (Fig. 5). 
489: 
490: 
491: %dfigure5%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
492: \begin{figure}
493: \centerline{\epsfxsize=3.3in\epsfbox{figure05.eps}}
494: \caption{The dielectric $\epsilon'(T)$ and $\tan \delta$ responses 
495: of KLT(5):Ca at 7.5~kbar.  Pressure shifts the responses to lower $T$Õs 
496: and sharpens the relaxational features (compare to Fig. 2).}
497: \end{figure}
498: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
499: 
500: 
501: An additional feature in the $\epsilon'(T)$ response in Fig. 2 (a) 
502: is the structure at temperatures below $\epsilon'_{max}$.  This 
503: structure is revealed more distinctly in the $\tan \delta(\omega,T)$ 
504: response illustrated in Figs. 2 (b) and 5 (b).  Some of this response 
505: is probably due to some 
506: inhomogeneities or defects in the crystal, but there are three sets 
507: of relaxational peaks that we wish to draw attention to.  
508: These are labeled A, B and C in Figs. 2 (b) and 5 (b) 
509: and all are Debye-like, the relaxational frequency (or 
510: inverse relaxation time, $\tau$) obeying Arrhenius kinetics
511: %
512: \begin{equation}
513: \omega = \tau^{-1} = \omega_o \exp(-E/kT)
514: \end{equation}
515: %
516: as shown for A and C in Fig. 6.  The kinetic parameters are: $E = 79 (244)$~meV 
517: and $\omega_o = 4.2 \times 10^{12} (5 \times 10^{15})$~s$^{-1}$ for 
518: A (C), respectively.  These parameters are in close agreement with 
519: the above-cited values for the reorientation of the  isolated 
520: Li$^{+}$ and Li$^{+}$-Li$^{+}$ ion pair dipoles, respectively, in 
521: undoped KLT, and, thus, are unaffected by the light Ca doping.
522: 
523: 
524: 
525: 
526: The mechanism for the reorientation of the Li$^{+}$ dipole is by
527: $\pi/2$ $(90^{\circ})$ flips of the dipolar orientation among the 
528: six equivalent Li sites 
529: (Fig. 1 (b)).~\cite{Doussineau_93,Toulouse_98JKPS,Christen_91}  
530: At the 5~\% Li concentration in KLT(5):Ca, the Li$^{+}$ - Li$^{+}$ 
531: ion pairÕs signature is clearly present in both the $\epsilon'$ and 
532: $\tan \delta$ data.  The reorientation of this ion pair is believed 
533: to be associated with the reorientation of two nearest-neighbor Li$^{+}$ 
534: ions.~\cite{Doussineau_93,Toulouse_98JKPS,Wemple_private,Christen_91,Prosandeev_03}  
535: Doussineau {\it et al.}~\cite{Doussineau_93} have suggested that, 
536: in their two degenerate lowest energy configurations, the pair 
537: dipoles are collinear and aligned parallel to the line that joins 
538: their sites.  The relaxation mechanism involves a simultaneous 
539: reversal of both dipoles in $\pi$ (180$^{\circ}$) flips [Fig. 1 (c)].  
540: Support for this mechanism comes from the absence of this 
541: relaxation in ultrasonic measurements.~\cite{Doussineau_93,Christen_91}  
542: Because the postulated Li$^{+}$ - Li$^{+}$ pairs are centro-symmetric, 
543: a flip leaves the quadrupole moment invariant, and this relaxation 
544: mode is not coupled to acoustic waves.  Support has also come from 
545: recent first-principles calculations on KLT 
546: supercells.~\cite{Prosandeev_03}  These calculations indicate that 
547: the pairÕs 180$^{\circ}$ reorientation proceeds via metastable 
548: intermediate steps.  Additionally, the calculated hopping barriers 
549: for both the isolated Li$^{+}$ dipole and the pair are found to be 
550: in reasonable agreement with those deduced from the dielectric 
551: measurements.
552: 
553: 
554: Relaxation B appears as a peak in $\tan \delta$ at 100~Hz, but 
555: is a shoulder at other frequencies.  This shoulder becomes more 
556: prominent and more separated from the A and C relaxations at high 
557: pressure [Fig. 5 (b)].  An approximate separation of the B peaks, 
558: shows that this relaxation is also Debye-like with the same 
559: activation energy $E \sim 135$~meV.  We attribute this new B 
560: relaxation in KLT(5):Ca to the Ca$^{2+}$ dopant, since it is not 
561: observed in undoped KLT.  But in any case, it is a minor feature 
562: in the response of KLT(5):Ca.
563: 
564: 
565: %dfigure6%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
566: \begin{figure}
567: \centerline{\epsfxsize=3.3in\epsfbox{figure06.eps}}
568: \caption{Arrhenius plots for the Li$^{+}$ [A] and Li$^{+}$-Li$^{+}$ 
569: pair [C] dipolar relaxations in KLT(5):Ca at 1~bar.}
570: \end{figure}
571: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
572: 
573: 
574: 
575: 
576: \subsection{Diffuse scattering and atomic shifts}
577: 
578: In the prototypical relaxor like Pb(Mg$_{1/3}$Nb$_{2/3}$)O$_3$ 
579: (PMN), it is well established that diffuse neutron 
580: scattering appears around the nuclear Bragg peaks below the 
581: so-called Burns temperature - below which the polar nano-regions
582: (PNR's) appear.~\cite{Naberezhnov_99}  Therefore, 
583: the diffuse scattering originates from the atomic displacements in 
584: PNR's, and its structure factor $F_{diff}({\rm \bf Q})$ can 
585: be written as
586: %
587: \begin{equation}
588: \label{f_diff} 
589: F_{diff}({\rm \bf Q}) = 
590: \sum_{\kappa} [{\rm\bf Q}\cdot \vec{\delta}_{\kappa}] 
591: b_{\kappa} e^{i{\rm\bf Q}\cdot{\rm\bf d}_{\kappa}} e^{-W_{\kappa}}, 
592: \end{equation}
593: %
594: where $\vec{\delta}_{\kappa}$, $b_{\kappa}$, ${\rm\bf d}_{\kappa}$, 
595: and $e^{-W_{\kappa}}$ are the atomic shift, neutron scattering 
596: length, coordination vector, and Debye-Waller factor 
597: (approximately 1) for $\kappa$ atoms, respectively.  
598: %
599: Note that the diffuse scattering intensity is proportional to $|F_{diff}|^2$.
600: %
601: For PMN, anomalous atomic shifts in the PNR's are reported; 
602: the total shift is composed of two components - one is 
603: a set of shifts that conserves the center-of-mass (CM) condition, 
604: and the other is a uniform phase shift.~\cite{Hirota_02}  
605: To verify if such anomalous shifts are also realized in the 
606: KLT system, we have carried out a study of the diffuse scattering from KLT(5):Ca.
607: 
608: 
609: %nfigure1%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
610: \begin{figure}
611: \centerline{\epsfxsize=3.3in\epsfbox{figure07.eps}}
612: \caption{(Color online) (a) Temperature dependence of the $(2, 0, 0)$ 
613: nuclear Bragg peak (black circles) and the diffuse scattering 
614: intensity measured at $(1.05, 1, 0)$.  (b)-(d) Contour 
615: maps of the diffuse intensity around $(1, 1, 0)$ at 11~K, 
616: 110~K, and 200~K.}
617: \end{figure}
618: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
619: 
620: 
621: The black circles in Fig. 7 (a) show the temperature dependence 
622: of the (200) nuclear Bragg peak.  Some relaxors that have an actual FE transition at 
623: $T_c$, such as KLT with higher Li concentration and Pb(Zn$_{1/3}$Nb$_{2/3}$)O$_3$ (PZN) 
624: have been reported to show an increase in the nuclear Bragg 
625: peaks below $T_c$ due to a release of extinction.  The 
626: present KLT(5):Ca crystal shows an increase of the (200) 
627: intensity below $75$~K.
628: %%
629: Although our crystal does not show a sharp peak in the dielectric 
630: response at 75~K, this temperature is consistent with the first hump 
631: in the $\epsilon'(T)$ response in Fig. 2 (a) that corresponds to the 
632: slowing down of the fluctuations of the Li$^{+}$ dipoles.
633: %%
634: Additionally, the increase of the nuclear Bragg intensity in the present 
635: crystal is only $\sim 50$~\% and it occurs gradually with decreasing 
636: temperature.  This is in contrast to the behavior of 
637: KLT crystals with higher Li content which have a well-defined 
638: FE transition in the dielectric measurements; for these latter crystals 
639: the nuclear peak 
640: intensity increases rapidly by more than an 
641: order of magnitude at $T_c$.~\cite{Toulouse_00,waki_unpub}  These facts 
642: suggest that the present crystal has a very small volume 
643: fraction of any ferroelectrically transformed area, and that a 
644: relaxor state is still dominant below 75~K, 
645: a conclusion supported by the dielectric results.
646: %%%%%%%%%%%%%%%%%%%
647: Hereafter, we define this temperature as $T_m$ (not $T_c$) since 
648: no bulk FE transition occurs in our crystal.  
649: 
650: 
651: 
652: 
653: Temperature variations of the diffuse scattering around 
654: the (110) position are shown in Figs. 7(b) - 7(d).  
655: At $T=11$~K, diffuse scattering ridges along the $H$ and $K$ 
656: directions are clearly observed.  At $T=110$~K, which is 
657: $\sim 35$~K higher than $T_m$, the diffuse ridges are 
658: still visible, and then, the diffuse scattering finally becomes 
659: unobservable at 200~K.  The temperature dependence of this 
660: scattering intensity measured at $(1.05, 1, 0)$ is 
661: indicated in Fig. 7 (a) by open circles and shows gradual 
662: increase starting at $\sim 150$~K, 
663: suggesting that the Burns temperature of this system 
664: should be above this temperature.  The diffuse 
665: scattering does not show the critical scattering behavior 
666: around $T_m$, that is observed in relaxor materials 
667: with a well-defined $T_c$, such as PZN~\cite{Stock_PRB04} 
668: and KLT with higher Li content.~\cite{Toulouse_00}  
669: Again, this is supporting evidence that the present sample does not 
670: have a ferroelectrically well-ordered state.
671: 
672: 
673: %nfigure2%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
674: \begin{figure}
675: \centerline{\epsfxsize=3.3in\epsfbox{figure08.eps}}
676: \caption{Comparison of diffuse scattering peaks at 10~K 
677: (open circles) and at 160~K (black circles) at various 
678: ${\rm \bf Q}$ positions.  Left and right figures show 
679: scans along the $H$ and $K$ directions, respectively.}
680: \end{figure}
681: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
682: 
683: 
684: %%% PNR shape and length scale %%%%%%%%%%%%%%
685: The observed diffuse scattering ridge along the [100] direction 
686: in the reciprocal space 
687: should correspond to a disk shape of PNR's which spreads in 
688: the $\{100\}$ plane in the real space, 
689: as discussed in Ref.~\onlinecite{Toulouse_00}.  
690: Similar {\it pancake} shaped PNR's in PMN and PZN-PT have been 
691: discussed in Ref.~\onlinecite{Xu_pancake}.  
692: From the width of the diffuse peaks, we estimate the dimensions 
693: of the disk PNR's as $52 (\pm 10)$~\AA~ in the \{100\} plane 
694: and $16 (\pm 3)$~\AA~ thickness in the [100] direction.
695: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
696: 
697: 
698: 
699: Diffuse scattering profiles at several zones are shown in Fig. 8.  
700: A clear diffuse component appears as wide intensity tails 
701: around (110), and a small diffuse intensity is 
702: observed around (210).  For (200), the longitudinal profile 
703: ($H$-scan) is somewhat peculiar; the Bragg peak itself has 
704: already a wide tail even at 160~K.  (We note that a small 
705: hump at $H = 1.93$ is a powder line from the aluminum 
706: surrounding the sample.)  
707: %% GAS statement SUSPICIOUS!! %%%%%%%%%%%%%
708: The presence of this tail at this high temperature may be related to 
709: the dielectric results in Fig. 2 (a) that show relaxation 
710: associated with the Li$^{+}$-Li$^{+}$ dipolar pairs that 
711: extends to $\sim 200$~K.
712: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
713: We also find that there is an intensity 
714: difference between the results obtained at 160~K and 10~K at $H \sim 2.05$.  
715: This difference, however, is suppressed quickly at $H=2.07$, 
716: whereas the diffuse scattering decays gradually up to $q=0.1$.  
717: Therefore, we conclude that the difference in intensities at 
718: $H = 2.05$ is not associated with diffuse scattering 
719: but results from possible broadening of the Bragg peak 
720: in the longitudinal direction below $T_m$.  
721: 
722: Apparently, there is no clear diffuse signature around (100).  
723: Based on Eq.~\ref{f_diff}, structure factors for both 
724: (100) and (210) can be written as 
725: %
726: $F_{diff}({\rm \bf Q}) = [{\rm \bf Q} \cdot \hat{\delta}](A - B -O)$
727: %
728: where 
729: %
730: $A = 0.95 \delta_{\rm K} b_{\rm K} + 0.05 \delta_{\rm Li} b_{\rm Li}$, 
731: $B = \delta_{\rm Ta} b_{\rm Ta}$, 
732: $O = \delta_{\rm O} b_{\rm O}$, 
733: %
734: and $\hat{\delta}$ is a unit vector.
735: This means that the difference in the diffuse intensity 
736: between (100) and (210) arises from $|{\rm \bf Q}|^2$.  
737: Our observation demonstrates an existence of diffuse 
738: intensity at (210), thus, the diffuse intensity at 
739: (100) should be simply too small to observe rather than being zero.
740: 
741: 
742: In order to derive the atomic shifts by using Eq.~\ref{f_diff}, we employ 
743: the relation $F_{diff}(110) = 1.93 F_{diff}(210)$ and 
744: $F_{diff}(200) = 0$.  The former is based on a comparison 
745: of the diffuse intensities at $(1.05, 1, 0)$ and 
746: $(2.05, 1, 0)$.  
747: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
748: %Then we assume that $\hat{\delta} = [100]$ 
749: %which is parallel to the Li off-center direction.  
750: Then we assume $\hat{\delta}$ to be parallel to the Li off-center direction,
751: either $[100]$ or $[010]$.
752: %%%%%%%%%%%%%%%%%%%%%
753: Thus the local crystal symmetry inside the PNR is tetragonal.  
754: %%%%%%%%%%%%%%%%%%%%%
755: The PNR's with $\hat{\delta} = [100]$ and $[010]$ are also assumed 
756: to be distributed with same population.
757: %%%%%%%%%%%%%%%%%%%%%
758: These conditions give $A = -1.66 O$ and $B=-1.34 O$.  
759: Here we use the quantities $b_{\rm K}=0.35$, $b_{\rm Ta}=0.70$, 
760: and $b_{\rm O}=0.58$, and, for simplicity, neglect the 
761: $\delta_{\rm Li}$ contribution since the Li content (5~\%) 
762: is relatively low.  Then atomic shifts that are normalized to 
763: the shift of the oxygen atom ($\delta_{\rm O}$) are obtained as 
764: %
765: \begin{eqnarray*}
766: \delta_{\rm K} &=& -2.89,\\
767: \delta_{\rm Ta} &=& -1.11,\\
768: \delta_{\rm O} &=& 1.00.
769: \end{eqnarray*}
770: 
771: Remarkably, these values do not satisfy the 
772: CM condition, $\sum_{\kappa} \delta_{\kappa} M_{\kappa} = 0$, 
773: where $M$ is an atomic mass, $M_{\rm K} = 39$, 
774: $M_{\rm Ta} = 181$, and $M_{\rm O} = 16$.  
775: %
776: A similar breaking of the CM rule has been found in the 
777: atomic shifts of PMN.~\cite{Vakhrushev_95}  
778: Hirota {\it et al.}~\cite{Hirota_02} proposed a 
779: model of TO phonon condensation with a uniform 
780: phase shift.  In this model, the atomic shift can 
781: be expressed as 
782: $\delta_{\kappa} = \delta_{\kappa}^{cm} + \delta_{\kappa}^{us}$, 
783: where $\delta_{\kappa}^{cm}$ originates from the
784: condensation of the TO soft mode conserving the CM condition, 
785: whereas $\delta^{us}$ is a uniform shift of PNR's 
786: relative to the surrounding cubic matrix.  
787: Following the same procedure employed by Hirota {\it et al.}, 
788: we obtain
789: %
790: \begin{eqnarray*}
791: \delta_{\rm K}^{cm} &=& -1.90,\\
792: \delta_{\rm Ta}^{cm} &=& -0.12,\\
793: \delta_{\rm O}^{cm} &=& 1.99,\\
794: \delta^{us} &=& -0.99.
795: \end{eqnarray*}
796: %
797: 
798: 
799: %nfigure3%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
800: \begin{figure}
801: \centerline{\epsfxsize=3.3in\epsfbox{figure09.eps}}
802: \caption{Zone center optic phonon profiles at $(2, 0, 0)$ 
803: (top) and $(1, 1, 0)$ (bottom).  Solid lines are fits to 
804: a resolution-convoluted Lorentzian function, while the dashed 
805: lines represent a Gaussian peak component at $\Omega = 0$.}
806: \end{figure}
807: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
808: 
809: 
810: It should be noted that, once the diffuse intensity ratios 
811: at different zones are given, the above quantities 
812: $\delta$, $\delta^{cm}$, and $\delta^{us}$ are uniquely 
813: derived only by a mathematical procedure.  
814: %
815: The validity of the model can be tested by checking to see if 
816: $\delta^{cm}$ can reproduce the intensity ratio of the 
817: zone center TO mode.  If $\delta_{\kappa}^{cm}$ 
818: originates from the condensation of the TO soft mode, 
819: it should be proportional to the phonon polarization 
820: vector $\vec{\xi}_{\kappa}$.  The inelastic structure factor 
821: $F_{inel}({\rm \bf Q})$ is expressed basically by the 
822: same formula as Eq.~\ref{f_diff} where $\vec{\delta}_{\kappa}$ 
823: should be replaced by $\vec{\xi}_{\kappa}$.  
824: %
825: From the obtained $\delta_{\kappa}^{cm} (\propto \xi_{\kappa}$), 
826: a ratio of the inelastic structure factors 
827: $|F_{inel}(200)|/|F_{inel}(110)|$ is calculated to 
828: be $\sim 2.9$.  The observed zone-center TO phonons at (200) and 
829: (110) are shown in Fig. 9 which were measured at 
830: room temperature.  The solid lines are fits to a Lorentzian 
831: phonon crosssection and a Gaussian peak at energy transfer $\Omega = 0$ 
832: convoluted with an instrumental resolution function.  (The actual 
833: formula is given in the next section.)  
834: Our fits give $|F_{inel}(200)|/|F_{inel}(110)| = 3.3 (\pm 0.3)$, 
835: which is in agreement with the value calculated 
836: from $\delta^{cm}$.  
837: 
838: 
839: Another check of the consistency of the data may be achieved by determining the 
840: contributions of the Slater and Last modes, which are two 
841: dominant optic modes in perovskite materials.  In case 
842: of KLT, the Slater mode corresponds to an atomic motion 
843: whereby the oxygen ions and Ta ions move in opposite 
844: directions while the K ions do not move.  The Last mode 
845: corresponds to the counter displacements of the TaO$_{6}$ octahedra and K 
846: ions.  In order to satisfy the CM 
847: condition, the amplitudes $(\xi_{\rm K}, \xi_{\rm Ta}, \xi_{\rm O})$ 
848: for the Slater and Last modes should be $\vec{s}_1 = (0, -0.265, 1)$ 
849: and $\vec{s}_2 = (-5.872, 1, 1)$, respectively.  Designating the contributions 
850: of the Slater and Last modes to $S_1$ and $S_2$, 
851: respectively, we obtain 
852: $S_1 \vec{s}_1 + S_2 \vec{s}_2 = (\xi_{\rm K}, 
853: \xi_{\rm Ta}, \xi_{\rm O})$, giving $S_1 \sim 1.6$ 
854: and $S_2 \sim 0.3$.  Harada {\it et al.}~\cite{Harada_70} 
855: have reported an almost 100~\% contribution of the Slater 
856: mode in the non-Li-doped KT, which is qualitatively 
857: consistent with the present observation giving an 85~\% 
858: contribution by the Slater mode.  In the Li-doped samples, 
859: it is possible that the K-ions can move more since the doped 
860: Li ions with their smaller radius effectively provide more space.  
861: This is possibly why we have a somewhat larger contribution of 
862: the Last mode in KLT than in KT.
863: %
864: (In the same mannar, the small amount of Ca ions in our crystal, 
865: which shift to the off-center positions in combination with the 
866: oxygen vacancies, may also cause the same effect.)
867: 
868: 
869: 
870: The diffuse scattering in the relaxor KLT(5):Ca, therefore, exhibits 
871: analogous behavior to that of the prototypical relaxor PMN.  
872: Remarkably, the atomic shifts in KLT, that are characterized 
873: by the diffuse intensities, can be consistently 
874: explained by the uniform phase shift model introduced 
875: by Hirota {\it et al.}  Accordingly, it is likely that the uniform 
876: phase shift is a common feature in relaxor systems 
877: and is a key to understanding the fundamentals of the relaxor mechanism.
878: 
879: 
880: 
881: The origin of the uniform phase shift remains somewhat of an open question.  Hirota {\it et al.}~\cite{Hirota_02} have hypothesized that the phase shift in PMN occurs as the polarized regions slide on an electric field gradient which is due to the existence of chemical ordered and disordered domains.  In the ordered domain, Mg$^{5+}$ and Nb$^{2+}$ ions are ordered in 1:1 ratio, which breaks the charge neutrality in the domains, and thence the electric field gradient is induced in a comparable length scale to the PNRÕs.  Although there are no such the electric field gradient in KLT, as Li$^{+}$ and K$^{+}$ have the same valence, a somewhat analogous mechanism is likely operable.  In this case, the large off-center displacements of Li ions produce two dipolar entities that are sources of random electric fields that can trigger the phase shift. Thus, we envision that for both KLT and PMN electric fields trigger the phase shift which is then superimposed on the ionic displacements associated with the FE soft mode resulting in the total measured displacements within the PNRs.
882: 
883: 
884: 
885: 
886: 
887: 
888: \subsection{Lattice dynamics}
889: 
890: In relaxor materials, anomalies of the transverse optic (TO) and the 
891: transverse acoustic (TA) phonons are expected to accompany PNR formation.  
892: A feature that has recently attracted a significant amount of 
893: attention is the strong damping of the TO mode below 
894: the Burns temperature at small $q$, a feature 
895: observed in both PZN~\cite{Gehring_PRB01} and PMN~\cite{Gehring_PRL01}.  
896: This is interpreted as a result of PNR formation 
897: that disrupts the long wave-length, i.e. small-$q$, optic mode.  
898: Also, the line width of the TA mode has been found to 
899: broaden below the Burns 
900: temperature.~\cite{waki_02}  We have studied the TA and TO 
901: phonons in the present relaxor KLT(5):Ca crystal in order to understand 
902: the lattice dynamics and to learn if such phonon anomalies are universal 
903: features of relaxors.
904: %
905: It should be noted that the TO phonon in KLT originates mostly from the dynamics of the TaO$_{6}$ octahedra since the Slater mode is highly dominant as mentioned earlier.  On the other hand, the relaxational behavior of doped Li ions appears in the dielectric response as we see in Sec. IV A.
906: 
907: %nfigure4%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
908: \begin{figure}
909: \centerline{\epsfxsize=3.3in\epsfbox{figure10.eps}}
910: \caption{Zone center TO phonon measured at $(2, 0, 0)$ 
911: (left figures) and the TA phonon measured at 
912: $(2, -0.15, 0)$ (right figures).  Solid lines are fits to 
913: a resolution-convoluted Lorentzian function.}
914: \end{figure}
915: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
916: 
917: 
918: We have measured the phonons around the (200) Bragg peak since phonon 
919: structure factors at the other $Q$ positions that are accessible 
920: in the present condition of $E_f = 14.7$~meV are quite small.  
921: Figure 10 shows representative profiles of the TO phonons at the 
922: zone center (200) and of the TA phonons at $(2, -0.15, 0)$.   
923: It is seen that, with decreasing temperature, the phonon 
924: energy of the zone-center TO mode decreases and the line 
925: width apparently increases, but the phonon remains a well-defined 
926: excitation at all temperatures.  The TA phonon peak also broadens as 
927: the temperature decreases.
928: 
929: 
930: In order to extract more insight, all of the profiles have been analyzed by fitting to 
931: the resolution-convoluted phonon scattering function 
932: %
933: \begin{equation}
934: S({\rm \bf q}, \Omega) = [n(\Omega) + 1] \chi''({\rm \bf q}, \Omega),
935: \end{equation}
936: %
937: \begin{equation}
938: \chi''({\rm \bf q}, \Omega) = %[n(\Omega) + 1][
939: \frac{I}{(\Omega - \Omega_{ph}({\rm \bf q}))^2 + \Gamma (\Omega)} - 
940: \frac{I}{(\Omega + \Omega_{ph}({\rm \bf q}))^2 + \Gamma (\Omega)},
941: \end{equation}
942: %
943: where $\Omega$ is the neutron energy transfer $E_i - E_f$, 
944: $n(\Omega) = 1/(e^{\Omega / k_{B}T}-1)$ is the Bose factor, 
945: $\Omega_{ph}$ is the phonon energy, $\Gamma(\Omega)$ is the 
946: half-width-at-half-maximum (HWHM) of the phonon spectrum, 
947: and $I$ is the amplitude.  
948: %
949: To fit the TO spectra, the following dispersion relation was utilized;
950: %
951: \begin{equation}
952: \Omega_{ph}^2(q) = \Omega_0^2 + (Cq)^2,
953: \end{equation}
954: %
955: where $\Omega_0$ is a zone-center TO phonon energy, and 
956: $C = 40.3$~meV$^{2}$\AA$^2$ determined from the dispersion 
957: at room temperature.  For fitting of the TA profiles, 
958: the dispersion is taken to be linear in $q$.
959: 
960: 
961: %nfigure5%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
962: \begin{figure}
963: \centerline{\epsfxsize=3.3in\epsfbox{figure11.eps}}
964: \caption{(a) Dispersions of the TO and 
965: TA phonons measured in the $(2, 0, 0)$ 
966: zone.  (b) Temperature dependence of the line width $\Gamma$ 
967: of the TA phonon measured at $(2, -0.15, 0)$.  
968: (c) Zone-center TO phonon energy squared $\Omega_0^2$ 
969: (open circles) and phonon line width $\Gamma$ (black circles) 
970: as a function of temperature.}
971: \end{figure}
972: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
973: 
974: 
975: The fits thus obtained are shown by solid lines in Fig. 10 
976: where the flat background is also adjusted.  Dashed lines 
977: represent elastic peaks at $\Omega = 0$ which are 
978: assumed to be Gaussians.
979: %
980: The fitting parameters are summarized in Fig. 11.  As 
981: shown in Fig. 11 (a), the TO mode softens gradually 
982: with decreasing temperature, but the dispersion is 
983: well-defined at all temperatures - in contrast to those 
984: of PMN and PZN.  The TO mode softening is 
985: clearly seen in Fig. 11 (c) where the zone-center TO 
986: phonon energy squared $\Omega_{0}^2$ plotted by 
987: open circles decreases linearly with temperature.  
988: This implies that the system still maintains an 
989: incipient lattice instability towards the 
990: ferroelectric state even with dopant Li ions present, 
991: similar to KT~\cite{Axe_70} and several KLT compositions.~\cite{Vogt_new}
992: %
993: Although the TO dispersion is well-defined at all 
994: temperatures, the TO phonon peak shows a small 
995: broadening at low temperatures.  The line width 
996: $\Gamma$ of the TO mode at the zone center, 
997: represented by the closed symbols in Fig. 11 (c), 
998: increases with decreasing temperature, 
999: characteristic soft mode behavior for ABO$_3$ perovskites.
1000: 
1001: 
1002: The TA phonon dispersion at room temperature is 
1003: shown in Fig. 11 (a).  We have observed a small 
1004: softening by $\sim 1$~meV of the TA phonon energy at $q=0.15$~r.l.u. 
1005: from room temperature to 10~K 
1006: with a linear decrease in this energy.  
1007: The TA phonon peaks are sharp and nearly 
1008: resolution limited.  It is somewhat difficult 
1009: to evaluate an intrinsic line width by 
1010: deconvolution analysis which gives a large ambiguity 
1011: for the resolution-limited peaks.  
1012: Therefore, we have fit the TA profiles to a simple 
1013: Lorentzian function to determine an effective line width.  
1014: The line width, shown in 
1015: Fig. 11 (b), increases with decreasing temperature 
1016: even above $T_m$.  
1017: This temperature-dependence resembles that of 
1018: the diffuse intensity shown in Fig. 7 (a),
1019: implying that the TA line broadening accompanies
1020: the formation of PNR's.
1021: 
1022: 
1023: 
1024: 
1025: There are interesting similarities and differences 
1026: in the phonon behavior between the present KLT(5):Ca material 
1027: and the prototypical relaxors PMN and PZN.  
1028: %
1029: In the case of PMN and PZN, the TO phonon softens 
1030: following $\Omega_0^2 \propto T$ above the Burns 
1031: temperature $T_d$.  
1032: It then becomes strongly damped in the 
1033: temperature range of $T_c ({\rm or}~ T_m) < T < T_d$, and finally recovers 
1034: below $T_c$ ($T_m$) - and subsequently hardens with further decrease in 
1035: $T$ - again following 
1036: $\Omega_0^2 \propto T$.~\cite{Stock_PRB04,waki_02}
1037: %
1038: For the present KLT(5):Ca crystal, the TO phonon softens 
1039: in the same manner, although it does not harden 
1040: below $T_m$ because there is no macroscopically well-developed 
1041: ferroelectric state.  
1042: The softening at high 
1043: temperatures for all these materials means that the relaxor state 
1044: sets in ABO$_3$ lattices that have  
1045: incipient lattice tendencies towards ferroelectric 
1046: transitions. 
1047: %%% GAS %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1048: This is as expected because the high polarizability of these FE mode 
1049: lattices results in large correlation lengths for dipolar interactions 
1050: and favors the formation and growth of PNR's below the Burns 
1051: temperature.~\cite{Grubbs_05}
1052: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1053: No hardening of the TO phonon below 
1054: $T_m$ in the present crystal means that the sample 
1055: remains in the relaxor state at low temperatures.  
1056: Consistent with this conclusion, the crystal shows neither a well-defined 
1057: transition in the dielectric measurements nor a large 
1058: extinction release of the nuclear Bragg peaks.  
1059: 
1060: 
1061: Another observed difference between KLT and PMN (PZN) is the absence of strong damping 
1062: of the TO phonon in the present KLT(5):Ca crystal, 
1063: %
1064: which is frequently associated with the ``waterfall'' anomaly observed in PMN and PZN. 
1065: We speculate
1066: %
1067: that this 
1068: absence of strong damping 
1069: occurs because the PNR's in the KLT(5):Ca crystal 
1070: have small size and a small volume fraction.  Thus, these 
1071: PNR's are not robust enough to 
1072: dampen the TO phonon, but nevertheless can broaden the line width 
1073: as shown in Fig. 11 (c). 
1074: %
1075: An alternative explanation for the waterfall effect has come from a recent neutron scattering study of 60\%-PbTiO$_{3}$ doped PMN by Stock {\it et al.}~\cite{Stock_05NWF} that revealed the  existence of the waterfall anomaly even though the sample exhibits long-range order,  suggesting that the waterfall is unrelated to PNR's.  They have argued that the waterfall results from random fields due to the chemical disorder (i.e., valence fluctuations).  It is important to note, however, that although mixed PbTiO$_3$-PMN samples exhibit FE order, they are highly, compositionally disordered and still exhibit considerable relaxor character.  If random fields are responsible for the waterfall effect, then it should be seen in KLT where the Li dipoles produce strong random fields.  Therefore our conclusion at this time is that the PNR explanation for the waterfall effect should not be dismissed.  However, further work is necessary to firm up the origin of this interesting effect.
1076: 
1077: 
1078: As for the TA phonon behavior, in contrast to our KLT(5):Ca sample, 
1079: Toulouse and 
1080: Hennion~\cite{Toulouse_94} have reported that the TA 
1081: phonon of KLT(3.5) splits into two modes below $T_c=52$~K corresponding 
1082: to a tetragonal symmetry of the ferroelectric state.  
1083: (They claim that their KLT(3.5) sample shows an FE transition 
1084: at 52~K from the dielectric measurements.~\cite{Toulouse_94})
1085: %
1086: Our crystal shows only 
1087: TA phonon broadening and no splitting because 
1088: it does not have a ferroelectric transition.  
1089: %
1090: As shown in Fig. 11 (b), the TA broadening appears to 
1091: start above $T_m$, and possibly at $T_d$ which 
1092: should be around $200$~K as noted earlier based on the 
1093: dielectric measurements and the diffuse scattering.  
1094: Thus, the TA phonon is also affected by PNR 
1095: formation.
1096: %
1097: In contrast to this behavior, PMN and PZN have been reported to 
1098: show TA phonon broadening at temperatures in the range of 
1099: $T_c < T < T_d$.~\cite{waki_02}  
1100: %
1101: Such behavior has been recently 
1102: interpreted in the framework of a coupling of the TA mode 
1103: to the diffuse component since the TA broadening is strong
1104: at the {\rm\bf Q} positions where the diffuse scattering is 
1105: strong.~\cite{Hlinka_03,Stock_cm04}  
1106: %
1107: More detailed 
1108: measurements are necessary to determine if such coupling effects 
1109: also exist in KLT as a manifestation of a common feature of relaxors.
1110: 
1111: 
1112: %%%%% Discussion of the TO mode energy %%%%%%%%%%%%
1113: Finally we discuss the TO phonon energy that depends on the Li 
1114: concentration.  Comprehensive study of the TO soft mode using 
1115: hyper-Raman spectroscopy for various Li concentrations have been 
1116: carried out by Vogt.~\cite{Vogt_new}  It is clearly demonstrated 
1117: that the TO mode frequency $\Omega_{0}$ increases with Li content 
1118: $x$ following a linear relation 
1119: %
1120: \begin{equation}
1121: \Omega_{0}^{2}(x, T) = \Omega_{0}^{2}(0, T) + A x.  
1122: \label{Vogt_formula}
1123: \end{equation}
1124: %
1125: In fact, the $\Omega_{0}$ value of our KLT(5):Ca crystal is higher 
1126: than that of the pure KT measured by the neutron scattering by 
1127: Axe {\it et al.}~\cite{Axe_70}  However, it is even higher than 
1128: that of the $x=0.087$ sample measured by Raman spectroscopy by Vogt.  
1129: To address this problem, we test the relation of 
1130: Eq.~\ref{Vogt_formula} using the present KLT(5):Ca sample and 
1131: another KLT crystal with $x=0.10$.
1132: 
1133: %nfigure6%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1134: \begin{figure}
1135: \centerline{\epsfxsize=2.8in\epsfbox{figure12.eps}}
1136: \caption{Li concentration dependence of $\Omega_{0}^{2}$ at $T=10$~K 
1137: and $300$~K.  Data of pure KT (closed symbols) are adopted from 
1138: Ref.~\onlinecite{Axe_70}.  The lines are results of fits to 
1139: Eq.~\ref{Vogt_formula}.}
1140: \end{figure}
1141: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1142: 
1143: 
1144: The latter sample was grown in the same manner, but is free from 
1145: Ca impurity.  It has a clear FE transition at $T_c=113$~K evidenced 
1146: from dielectric measurements and neutron scattering experiments.  
1147: (Details of this sample will be published elsewhere.~\cite{waki_unpub})  
1148: The quantities of $T_m = 75$~K for KLT(5):Ca and $T_c = 113$~K for 
1149: KLT(10) are consistent with {\it freezing temperatures}, $T_g$, 
1150: summarized in Ref.~\onlinecite{Klink_83}, suggesting a validity of 
1151: the estimation of the Li contents.
1152: 
1153: The zone-center TO phonon energy squared $\Omega_{0}^{2}(x, T)$ at 
1154: $T=10$~K and $300$~K are shown in Fig. 12, together with those of 
1155: pure KT referred from Ref.~\onlinecite{Axe_70}.  It is clearly 
1156: demonstrated that the linear relation of Eq.~\ref{Vogt_formula} 
1157: stands for the series of our samples.  This is further evidence 
1158: of proper Li concentrations of our samples, and also, it implies 
1159: that the small amount of Ca impurity has only a minor effect on 
1160: the lattice dynamics properties.
1161: %
1162: We believe that the disagreement of $\Omega_{0}$ with those of 
1163: Raman data by Vogt is due to a systematic difference of the Li 
1164: content evaluation. 
1165: 
1166: 
1167: 
1168: 
1169: 
1170: 
1171: \section{Concluding remarks}
1172: 
1173: 
1174: The origins of some aspects of the relaxor behavior in ABO$_3$ oxides remain as somewhat open questions.  For the prototypical relaxors PMN and PZN, however, it is clear that the chemical (valence) disorder of the Mg and Nb ions and the associated random fields play an important role by disrupting FE correlations leading to the relaxor state.  In the KLT system Li dipoles also produce random fields that are thought by some to be responsible for the relaxor behavior (e.g., see Ref.~\onlinecite{Kleemann_87}).  We note, however, that the situation in KLT is different from that in most ABO$_3$ relaxors in that the parent crystal KT has no FE order to disrupt.  In fact, at very low Li concentration in KT the Li dipoles exhibit essentially dipolar glass response with no correlations among dipoles.  Such correlations develop with increasing Li concentration leading to the evolution of a relaxor state
1175: 
1176: Combining the relaxational behavior and the neutron scattering results, we can draw the following picture for KLT(5).  At the Burns temperature, PNR's start to form triggered by off-center displacements of the Li ions.  The symmetry of these polar regions is tetragonal, and the observed ionic displacements consist of the displacements associated with the soft FE mode on which is superimposed a uniform phase shift.  
1177: 
1178: Below the Burns temperature, the dynamics of Li atoms appears as the relaxational behavior in the dielectric response. As temperature decreases dipolar fluctuations are reduced and stronger correlations develop among the PNRs that grow in volume, as evidenced by the growth of diffuse scattering intensity. Slowing down of dipolar fluctuations sets in at a temperature corresponding to the peak temperature in the dielectric constant and continues on further cooling.  Although the present sample shows no bulk FE transition, the increase of the Bragg intensity below $\sim 75$~K, where the Li-Li pair becomes well static ($1000/T \sim 13$ in Fig. 6) is indicative of larger PNRs  (which are nano FE domains) or even the presence of larger FE domains in a sea of relaxor phase.
1179: 
1180: 
1181: In summary, 
1182: we have measured the dielectric response and lattice dynamics 
1183: in the relaxor KLT(5):Ca where a small amount of Ca was incorporated 
1184: into the KLT crystal.  
1185: Both types of results affirm that the present crystal remains in the 
1186: relaxor state and does not develop a ferroelectric state at 
1187: low temperatures.  Dielectric constants that are sensitive 
1188: to impurities exhibit strong enhancement attributed to free 
1189: electrons from the Ca dopant, however 
1190: the relaxational features including Arrhenius parameters of 
1191: Li$^{+}$ ion and Li$^{+}$-Li$^{+}$ pair dipolar relaxations 
1192: are not affected by the Ca.  
1193: 
1194: Neutron scattering results for KLT(5):Ca compared with those 
1195: of the relaxor PMN highlight common features in relaxors.  
1196: Lattice displacements in PNR's derived from the diffuse 
1197: intensities suggest that the model of the normal soft-mode 
1198: condensation with an additional uniform phase shift within 
1199: the PNR's is also relevant in KLT.  
1200: %
1201: This means that we cannot rule out a contribution of the 
1202: zone-center TO soft mode to the formation of PNR's.
1203: %
1204: However, the TO soft mode remains at relatively high energy compared to 
1205: that of PMN, and it also remains well-defined in contrast to the heavily-damped 
1206: nature characteristic of PMN.
1207: %%%
1208: We believe that this difference reflects the fact that KLT(5):Ca 
1209: is a weak relaxor, essentially a dipolar glass, whereas PMN is 
1210: the prototypical strong relaxor. 
1211: %%%
1212: A more comprehensive study will be required in order to 
1213: extract the precise nature of the specific coupling 
1214: between the TO phonon and the relaxor mechanism in this system.
1215: 
1216: 
1217: 
1218: \begin{acknowledgments}
1219: 
1220: 
1221: The authors thank B. Burton, T. Egami, R. Erwin, P. M. Gehring, 
1222: K. Hirota, K. Kakurai, and N. Metoki 
1223: for invaluable discussions.  
1224: This work was partially supported by the US-Japan Cooperative 
1225: Research Program on Neutron Scattering. 
1226: Work at Sandia National Laboratory was supported by 
1227: the Division of Material Sciences and Engineering, Office of Basic 
1228: Energy Sciences, U.S. Department of Energy under Contract 
1229: DE-AC04-94AL85000.  
1230: Research sponsored in part by the Division of Material Sciences 
1231: and Engineering, Office of Basic 
1232: Energy Sciences, U.S. Department of Energy under Contract 
1233: DE-AC05-00OR22725 with Oak Ridge National Laboratory, managed and
1234: operated by UT-Battelle, LLC.
1235: Finally, work at Brookhaven National Laboratory is supported by 
1236: financial support from the U. S. Department of Energy under 
1237: Contract No. DE-AC02-98CH10886.
1238: 
1239: 
1240: \end{acknowledgments}
1241: 
1242: 
1243: 
1244: 
1245: 
1246: 
1247: 
1248: %\newpage
1249: 
1250: \begin{references}
1251: 
1252: 
1253: 
1254: % 07
1255: \bibitem{Wemple_65} S. H. Wemple, Phys. Rev. {\bf 137}, A1575 (1964).
1256: 
1257: % 26
1258: \bibitem{Axe_70} J. D. Axe, J. Harada, and G. Shirane,
1259: Phys. Rev. B {\bf 1}, 1227 (1970).
1260: 
1261: % 28
1262: \bibitem{Raman} P. Calvi, P. Camagni, E. Giulotto, and L. Rollandi, 
1263: Phys. Rev. B {\bf 53}, 5240 (1996).
1264: 
1265: %%% Added by George [4] [5] %%%%%%%%%%
1266: \bibitem{Vogt_new} H. Vogt, J. Phys:Cond. Matter {\bf 7}, 5913 (1995); 
1267: Ferroelectrics {\bf 184}, 31 (1996); {\bf 202}, 157 (1997).
1268: 
1269: \bibitem{Prater_81} R. L. Prater, L.L. Chase and L. A. Boatner, 
1270: Phys. Rev. B {\bf 23}, 5904 (1981).
1271: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1272: 
1273: % 17
1274: \bibitem{Toulouse_00} G. Yong, J. Toulouse, R. Erwin, S. M. Shapiro,
1275: and B. Hennion, Phys. Rev. B {\bf 62}, 14736 (2000).
1276: 
1277: % 14
1278: \bibitem{Naberezhnov_99} A. Naberezhnov, S. Vakhrushev, B. Doner, D.
1279:   Strauch, and H.  Moudden Eur. Phys. J. B {\bf 11}, 13 (1999)
1280: 
1281: % 20
1282: \bibitem{Vakhrushev_95} S. B. Vakhrushev, A. A. Naberezhnov, 
1283: N. M. Okuneva, and B. N. Savenko, Fiz. Tverd. Tela (St. Petersburg) 
1284: {\bf 37}, 3621 (1995) [Phys. Solid State {\bf 37} 1993 (1995)].
1285: 
1286: % 21
1287: \bibitem{Hirota_02} K. Hirota, Z.-G. Ye, S. Wakimoto, P. M. Gehring, 
1288: and G. Shirane, Phys. Rev. B {\bf 65}, 104105 (2002).
1289: 
1290: \bibitem{Lynn_note} We have performed inductively coupled plazma (ICP) analysis on the starting powder materials.  The following metals were sought: Ca, Bi, Cr, Fe, Mn, Sr, V. Sb, B, Co, Pb, Mo, Si, Te, Zn, As, Cd, Cu, Ni, Ag, Sn, Zr, Ba, In, Mg, P, Na, and Ti.  The ICP analytical results did reveal the presence of $< 15$~ppm of only Ca in the potassium carbonate.
1291: 
1292: % 01
1293: \bibitem{Klink_84} J. J. van der Klink and F. Borsa, 
1294: Phys. Rev. B {\bf 30}, 52 (1984).
1295: 
1296: % 02
1297: \bibitem{Samara_SSP} G. A. Samara in Solid State Physics, 
1298: Vol.56 edited by H. Ehrenreich and F. Spaepen, Academic Press, 
1299: New York (2001) p. 239 and references therein.
1300: 
1301: % 03
1302: \bibitem{Hochli_90} U. T. Hochli, K. Knorr and A. Loidl, 
1303: Adv. Phys. {\bf 39}, 405 (1990).
1304: 
1305: % 04
1306: \bibitem{Yacoby_74} Y. Yacoby and S. Just, 
1307: Solid State Commun. {\bf 12}, 715 (1974).
1308: 
1309: % 05
1310: \bibitem{Leung_01} K. Leung, Phys. Rev. B {\bf 65}, 012102 (2001). 
1311: 
1312: % 06
1313: \bibitem{Senhouse_66} L.S. Senhouse, Jr., M.V. DePaolis, Jr. and 
1314: T.L. Loomis, Appl. Phys. Lett. {\bf 8}, 173 (1966).
1315: 
1316: % 08
1317: \bibitem{Prosandeev_01} S. A. Prosandeev, V.A. Trepakov, 
1318: M.E. Savinov, J. Jastradik and S.E. Kapphan, 
1319: J. Phys.: Condens. Matter {\bf 13}, 9749 (2001).
1320: 
1321: % 09
1322: \bibitem{Doussineau_93} P. Doussineau, Y. Farssi, C. Frenos, 
1323: A. Levelut, K. McEnaneu, J. Toulouse and S. Ziolkiewicz, 
1324: Europhys. Lett. {\bf 24}, 415 (1993). 
1325: 
1326: % 10
1327: \bibitem{Toulouse_98JKPS} J. Toulouse and R. Pattnaik, 
1328: J. Korean Phys. Soc. {\bf 32}, S942 (1998).
1329: 
1330: %%% Added by George [19] %%%%%%%%%%%%%%%%
1331: \bibitem{Grubbs_05} R. K. Grubbs, E. L. Venturini, P. G. Clem, 
1332: J. J. Richardson, B.A. Tuttle and G. A. Samara, 
1333: Phys. Rev. B {\bf 72}, 104111 (2005).  
1334: See also J. Volger in {\it Progress in Semiconductors}, edited by 
1335: A. F. Gibson (John Wiley and Sons, New York 1960), vol. 4, p. 207.
1336: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1337: 
1338: % 11
1339: \bibitem{Wemple_private} S. H. Wemple, M. DiDomenico, Jr. 
1340: and A. Jayaraman, Phys. Rev. {\bf 180}, 547 (1969).
1341: 
1342: % 12
1343: \bibitem{Christen_91} H. M. Christen, U.T. Hochli, A. Chatelain 
1344: and S. Ziolkiewicz, J. Phys.: Condens. Matter {\bf 3}, 8387 (1991).
1345: 
1346: % 13
1347: \bibitem{Prosandeev_03} S.A. Prosandeev, E. Cockayne and B.P. Burton, 
1348: Phys. Rev. B {\bf 68}, 014120 (2003).
1349: 
1350: 
1351: % 18
1352: \bibitem{waki_unpub} S. Wakimoto, G. A. Samara, and L. A. Boatner, unpublished.
1353: 
1354: % 19
1355: \bibitem{Stock_PRB04}C. Stock, R.J. Birgeneau, S. Wakimoto, J.S. Gardner, 
1356: W. Chen, Z.-G. Ye, and G. Shirane, Phys. Rev. B {\bf 69}, 094104 (2004).
1357: 
1358: %%% Pancake PNR
1359: \bibitem{Xu_pancake} G. Xu, Z. Zhong, H. Hiraka, and G. Shirane,
1360: Phys. Rev. B {\bf 70}, 174109 (2004).
1361: 
1362: % 22
1363: \bibitem{Harada_70} J. Harada, J. D. Axe, and G. Shirane, Acta Cryst. 
1364: {\bf A 26}, 608 (1970).
1365: 
1366: % 23
1367: \bibitem{Gehring_PRB01} P. M. Gehring, S.-E. Park, and G. Shirane,
1368: Phys. Rev. B {\bf 63}, 224109 (2001).
1369: 
1370: % 24
1371: \bibitem{Gehring_PRL01} P. M. Gehring, S. Wakimoto, Z.-G. Ye, and G. Shirane,
1372: Phys. Rev. Lett. {\bf 87}, 277601 (2001).
1373: 
1374: % 25
1375: \bibitem{waki_02} S. Wakimoto, C. Stock, R. J. Birgeneau, Z.-G. Ye, 
1376: W. Chen, W. J. L. Buyers, P. M. Gehring, and G. Shirane,
1377: Phys. Rev. B {\bf 65}, 172105 (2002).
1378: 
1379: % 27
1380: \bibitem{Toulouse_94} J. Toulouse and B. Hennion,
1381: Phys. Rev. B {\bf 49}, 1503 (1994).
1382: 
1383: %% cold neutron PZN-8PT
1384: %% butterfly diffuse
1385: %% coupling TA and diffuse
1386: \bibitem{Hlinka_03} J. Hlinka, S. Kamba, J. Petzelt, J. Kulda, 
1387: C. A. Randall, and S. J. Zhang,
1388: J. Phys.: Condes. Matter {\bf 15}, 4249 (2005).
1389: 
1390: %% neutron PMN
1391: %% strong coupling between diffuse and TA
1392: \bibitem{Stock_cm04} C. Stock, H. Luo, D. Viehland, J. F. Li, 
1393: I. Swainson, R. J. Birgeneau, and G. Shirane,
1394: J. Phys. Soc. Jpn. {\bf 74}, 3002 (2005).
1395: 
1396: 
1397: \bibitem{Klink_83} J. J. van der Klink, D. Rytz, F. Borsa, and 
1398: U. T. H{\"o}chli, Phys. Rev. B {\bf 27}, 89 (1983).
1399: 
1400: \bibitem{Stock_05NWF} C. Stock, D. Ellis, I. P. Swainson, G. Xu, H. Hiraka, Z. Zhong, H. Luo, X. Zhao, D. Viehland, R. J. Birgeneau, and G. Shirane, Phys. Rev. B {\bf 73}, 064107 (2006).
1401: 
1402: \bibitem{Kleemann_87} W. Kleemann S. K\"utz, and D. Rytz, Europhys. Lett. {\bf 4}, 239 (1987).
1403: 
1404: 
1405: \end{references}
1406: 
1407: 
1408: 
1409: 
1410: 
1411: 
1412: 
1413: 
1414: 
1415: 
1416: 
1417: 
1418: \end{document} 
1419: