1: % Version of Oct. 31,2005
2: \documentclass[prb,showpacs,twocolumn,floatfix]{revtex4}
3: %\documentclass[prb,showpacs,preprintnumbers,preprint]{revtex4}
4: \usepackage{graphicx}
5: \sloppy
6: \begin{document}
7: \def\rhov{{\mbox{\boldmath{$\rho$}}}}
8: \def\tauv{{\mbox{\boldmath{$\tau$}}}}
9: \def\Lambdav{{\mbox{\boldmath{$\Lambda$}}}}
10: \def\sigmav{{\mbox{\boldmath{$\sigma$}}}}
11: \def\xiv{{\mbox{\boldmath{$\xi$}}}}
12: \def\oh{{\scriptsize 1 \over \scriptsize 2}}
13: \def\of{{\scriptsize 1 \over \scriptsize 4}}
14: \def\tf{{\scriptsize 3 \over \scriptsize 4}}
15: \title{Towards a Microscopic Model of Magnetoelectric Interactions in
16: ${\bf Ni_3V_2O_8}$}
17:
18: \author{A. B. Harris,$^1$ T. Yildirim,$^2$ A. Aharony,$^{3,4}$ and
19: O. Entin-Wohlman,$^{3,4}$}
20:
21: \affiliation{
22: (1) Department of Physics and Astronomy, University of Pennsylvania,
23: Philadelphia, PA 19104\\(2) NIST Center for Neutron
24: Research, National Institute of Standards and Technology,
25: Gaithersburg, Maryland 20899\\ (3) School of Physics and
26: Astronomy, Raymond and Beverly Sackler Faculty of Exact Sciences,
27: Tel Aviv University, Tel Aviv 69978, Israel\\(4) Department of
28: Physics, Ben Gurion University of the Negev, Beer Sheeva 84105, Israel}
29: %%% ----------------------------------------------------------------------
30: \date{\today}
31:
32: \begin{abstract}
33: %TYTY Begin
34: We develop a microscopic magnetoelectric coupling in Ni$_3$V$_2$O$_8$
35: (NVO) which gives rise to
36: the trilinear phenomenological coupling used previously to explain
37: the phase transition in which magnetic and ferroelectric order
38: parameters appear simultaneously. Using combined neutron scattering
39: measurements and first-principles calculations of the phonons in NVO,
40: we determine eleven phonons which can induce the observed spontaneous
41: polarization. Among these eleven phonons, we find that a few of them
42: can actually induce a significant dipole moment. Using the calculated atomic
43: charges, we find that the required distortion to induce the observed
44: dipole moment is very small (~0.001 \AA) and therefore it would be
45: very difficult to observe the distortion by neutron-powder diffraction.
46: Finally, we identify the derivatives of the
47: exchange tensor with respect to atomic displacements which are
48: needed for a microscopic model of a spin-phonon coupling in NVO
49: and which we hope will be obtained
50: from a fundamental quantum calculation such as LDA+U.
51: We also analyze two toy
52: models to illustrate that the Dzyaloskinskii-Moriya interaction
53: is very important for coexisting of magnetic and ferroelectric
54: order but it is not the only mechanism when the local site
55: symmetry of the system is low enough.
56: %TYTY END
57: \end{abstract}
58: \pacs{75.25.+z, 75.10.Jm, 75.40.Gb}
59: \maketitle
60:
61: \section{Introduction}
62:
63: Recently studies have identified
64: a family of multiferroics which display a phase transition
65: in which there simultaneously develops long-range incommensurate
66: magnetic and uniform ferroelectric order. Perhaps the most detailed
67: studies have been carried out on the systems Ni$_3$V$_2$O$_8$
68: (NVO)\cite{Rogado,LawesKenzelmann,FERRO,EXPT} and
69: TbMnO$_3$ (TMO).\cite{TMO1,TMO2} (For a review, see Ref. \onlinecite{REV}.)
70: This phenomenon has been explained\cite{FERRO} on
71: the basis of a phenomenological model which invokes a Landau
72: expansion in terms of the order parameters describing the
73: incommensurate magnetic order and the order parameter describing
74: the uniform spontaneous polarization. Already from this
75: treatment it was clear that a microscopic model would have to
76: involve a trilinear interaction Hamiltonian proportional to the
77: product of two spin variables and one displacement variable.
78: Furthermore, the symmetry requirements of the phenomenological model
79: would naturally be realized by a proper microscopic model.
80: Accordingly, in this paper we present a detailed combined neutron
81: scattering study and first-principles calculations of the
82: optical phonons of NVO and thereby identify those having the right
83: symmetry to induce a dipole moment. We also consider the expansion of the
84: exchange tensor to first order in the generalized displacement coordinates.
85: The aim of this programme is to determine which of the generalized
86: displacements are relevant and which corresponding elements of the
87: exchange tensor are needed for this microscopic calculation.
88: This paper will therefore set the stage for a separate quantum
89: calculation of the exchange tensor and its derivatives with respect
90: to atomic displacements. On general grounds, one might expect the
91: Dzyaloshinskii-Moriya\cite{DM1,DM2} (DM) interaction to play an
92: important role in this calculation,\cite{DAGOTTO}
93: and indeed, we will find this to be the case.
94: However, the conclusion is a bit more subtle, in that
95: displacement derivatives of many elements of the exchange
96: tensor (and not just the antisymmetric one of the DM
97: interaction) are needed for the microscopic calculation.
98: We will show that simplistic models are somewhat misleading
99: in that they may lead one to believe that the only exchange tensor
100: elements whose derivatives are important are the DM ones.
101: Although our calculations may seem complicated, they
102: are essential if one wishes to actually relate the
103: magnetoelectric coupling to detailed quantum calculations
104: (via LDA\cite{LDA} or LDA$+U$\cite{LDAU} schemes) of the strain-dependence
105: of the exchange tensors. The methodology of the present paper can
106: be extended in a straightforward way to TMO, for instance.
107:
108: Briefly this paper is organized as follows. In Sec. II we
109: give an overview of the calculation.
110: In Sec. III we discuss the first-principles calculations of
111: the zone-center phonons and identify those phonons which
112: transform like a vector and which are thus candidates to
113: produce a spontaneous polarization. In this section we
114: also present the neutron scattering measurements of the phonon
115: density of states (DOS), which is found to be in good agreement
116: with the calculated spectrum. Next in Sec. IV we
117: summarize the results of the determination of the spin
118: structure.\cite{LawesKenzelmann,EXPT} In Sec. V we then use the
119: symmetry operations of the crystal to show how the phonon
120: derivatives of the various exchange tensors in the unit cell
121: are related to one another. Then in Sec. VI we show that a mean-field
122: treatment of this spin-phonon coupling leads to the results
123: obtained previously\cite{LawesKenzelmann,EXPT} in a phenomenological
124: model. Here we give expressions for the spontaneous polarization
125: in terms of graidents of the excahnge tensor whose evaluation
126: remains to be done. It would be nice to have a simple model to illustrate these
127: results. However, our studies of two ``toy models'' in Sec. VII
128: indicate that they do not reproduce some essential features of our
129: complete calculation. Finally, our conclusions are summarized
130: in Sec. VIII.
131:
132: \section{Overview of Calculation}
133:
134: Here we give a brief qualitative overview of the calculation.
135: First we review the symmetry
136: of the space group of NVO, Cmca (No. 64 in Ref. \onlinecite{HAHN}).
137: The space group operations (apart from primitive translations) are
138: specified in Table \ref{PUC}. (Here and below, sites within the unit cell
139: are given as fractions of the sides of the {\it conventional} unit cell,
140: so that $(x,y,z)$ denotes $(xa,yb,zc)$.) We now describe the sets of
141: crystallographically equivalent sites which the various atoms occupy.
142: (Such a set of crystallographically equivalent
143: sites is called a {\it Wyckoff orbit}.) There are six such orbits as
144: shown in Table~\ref{orbit}. The first two are those of the Ni atoms,
145: the first consisting of the two Ni(1) (a) sites (which we call
146: ``cross-tie'' sites) and the second consisting of the four
147: Ni(2) (e) sites (which we call ``spine'' sites). The four V (f) sites
148: comprise the third orbit and the oxygen sites are distributed into
149: two (f) orbits, one containing four O(1) atoms, the other containing four
150: O(2) atoms, and a (g) orbit containing eight O(3) atoms. (The letters
151: a, e, f, g classify the site symmetry according to the convention of
152: Ref. \onlinecite{HAHN}. The number of sites in the orbit as
153: listed in Ref. \onlinecite{HAHN} is twice what we give here because here
154: we consider the primitive unit cell rather than the conventional
155: unit cell.) The locations of these sites are specified in the second
156: column of Table \ref{orbit}.
157: Note that there are two formula units of NVO per unit cell.
158: The Ni sites form buckled planes which resemble a kagom\'{e}
159: lattice and two such adjacent planes are shown in Fig. 1 where
160: the buckling is omitted for simplicity. There one sees that the
161: Ni spine sites form chains along the $x$-direction between which
162: the Ni cross-tie sites are situated, with bonds to nearest-neighboring
163: spine sites which form a cross tie.
164: To illustrate the use of Table \ref{orbit} we find that the eight
165: operations of Table \ref{PUC} acting on $(0,0,0)$
166: generate four copies of each of the 2 a sites
167: which are at $(0,0,0)$ and $(1/2,0,1/2)$. Similarly one can
168: generate the eight g sites by applying in turn the eight operations
169: of Table \ref{PUC} to the site at $(x,y,z)$. (In each case it may be
170: necessary to bring the site back into the original unit cell via
171: a primitive translation vector.)
172:
173: \begin{table}
174: \begin{tabular} {||c | c ||}
175: \hline
176: ${\rm E}{\bf r} =(x,y,z)$
177: & $2_z{\bf r} =(\overline x + 1/2, \overline y , z + 1/2)$ \\
178: $2_y {\bf r} = (\overline x + 1/2, y, \overline z +1/2)$
179: & $2_x {\bf r} = (x, \overline y , \overline z)$ \\
180: ${\cal I} {\bf r} =(\overline x,\overline y,\overline z)$
181: &$m_z {\bf r} =(x+\oh , y, \overline z + \oh )$ \\
182: $m_y {\bf r} = (x+\oh , \overline y, z +1/2)$
183: &$m_x{\bf r} = (\overline x, y , z)$ \\
184: \hline \end{tabular}
185: \caption{\label{PUC}General positions within the primitive unit cell for
186: Cmca which describe the symmetry operations of this space group.
187: $2_\alpha$ is a two-fold rotation (or screw) axis and
188: $m_\alpha $ is a mirror (or glide) plane.
189: The primitive translation vectors are ${\bf a}_1=(a/2)\hat i + (b/2)\hat j$,
190: ${\bf a}_2=(a/2) \hat i - (b/2)\hat j$, and ${\bf a}_3=c\hat k$,
191: where $a=5.92170\AA$, $b=11.37105\AA$, and
192: $c=8.22638\AA$.\protect{\cite{EXPT,SAUER}}}
193: \end{table}
194:
195: %TYTYTY BEGIN replaced the old Table II with below...
196: \begin{table}
197: \begin{tabular}{|c|c|c|c|}\hline\hline
198: Atoms & (x/a,y/b,z/c) & Wyckoff & Decomposition \\ \hline
199: Ni(1) & (0,0,0) & 2a & A$_u$+2B$_{1u}$+2B$_{2u}$+B$_{3u}$ \\ \hline
200: Ni(2) & (1/4,y/b,1/4) & 4e & A$_u$+2B$_{1u}$+B$_{2u}$+2B$_{3u}$ \\
201: & y=0.1298 (0.1304) & & A$_{g}$+2B$_{1g}$+B$_{2g}$+2B$_{3g}$\\ \hline
202: V(1) & (0,y,z) & 4f & A$_u$+2B$_{1u}$+2B$_{2u}$+B$_{3u}$ \\
203: & y=0.3762 (0.3762) & & 2A$_{g}$+2B$_{1g}$+B$_{2g}$+2B$_{3g}$\\
204: & z=0.1197 (0.1196) & & \\ \hline
205: O(1) & (0,y/b,z/c) & 4f & A$_u$+2B$_{1u}$+2B$_{2u}$+B$_{3u}$ \\
206: & y=0.2481 (0.2490) & & 2A$_{g}$+2B$_{1g}$+B$_{2g}$+2B$_{3g}$\\
207: & z=0.2308 (0.2301) & & \\ \hline
208: O(2) & (0,y/b,z/c) & 4f & A$_u$+2B$_{1u}$+2B$_{2u}$+B$_{3u}$ \\
209: & y=0.0011(0.0008) & & 2A$_{g}$+2B$_{1g}$+B$_{2g}$+2B$_{3g}$\\
210: & z=0.2444 (0.2441) & & \\ \hline
211: O(3) & (x/a,y/b,z/c) & 8g & 3A$_u$+3B$_{1u}$+3B$_{2u}$+3B$_{3u}$ \\
212: & x=0.2656(0.2703) & & 3A$_{g}$+3B$_{1g}$+3B$_{2g}$+3B$_{3g}$\\
213: & y=0.1192 (0.1184) & & \\
214: & z=0.0002 (0.0012) & & \\ \hline \hline
215: \end{tabular}
216: \caption{\label{orbit}
217: In column 2 we give the Wyckoff positions, the fractional
218: coordinates (in coumn 2) and their multiplicity (in column 3)
219: of atoms in the Wyckoff orbits of NVO listed in column 1.
220: In column 2 we give the values of the structural parameters
221: (e.g. $x$, $y$, and $z$) as deduced from
222: diffraction data\protect{\cite{EXPT,SAUER}}
223: and the corresponding values we find
224: from the structural minimization are given in parentheses. In
225: column 4 we give the symmetry decomposition of the displacements
226: of atoms in each of the Wyckoff orbits.}
227: \end{table}
228:
229: We now review briefly the nature of the ordered phases which occur
230: as the temperature $T$ is lowered at zero external magnetic
231: field.\cite{Rogado,LawesKenzelmann,EXPT}
232: At high temperatures the system is paramagnetic.
233: When $T$ is lowered through the value $T_{PH} \approx 9.1$K,
234: an incommensurate phase appears (which we call the high-temperature
235: incommensurate or HTI phase) in which the Ni spins on
236: the spine chains are oriented very nearly along the $x$-axis with
237: a modulation vector also along $\hat i$. (The axes are denoted
238: either ${\bf a}$, ${\bf b}$, and ${\bf c}$, or $x$, $y$, and $z$ and
239: corresponding unit vectors are denoted $\hat i$, $\hat j$, and $\hat k$.)
240: As the temperature is further
241: lowered through the value $T_{HL}\approx 6.3$K transverse order
242: appears at the same incommensurate wavevector and also order
243: appears on the cross-tie sites, as shown in Fig. 1. We call
244: this phase the low-temperature incommensurate or LTI phase. Within
245: experimental uncertainty, these two ordering transitions are
246: continuous. As the temperature is lowered through the value
247: $T_{LC} \approx 4$K, a discontinuous transition occurs in which
248: a commensurate antiferromagnetic phase appears. In this
249: phase antiferromagnetism results from the arrangement of
250: spins within the unit cell in such a way that the magnetic
251: unit cell remains identical to the paramagnetic unit cell.
252:
253: \begin{figure}
254: \includegraphics[height=7.0cm]{NEW.eps}
255: \caption{\label{SSTR} Schematic diagram showing
256: the $x$- and $y$-components of the spins in the LTI
257: phase. We used the parameters: $q=0.28(2 \pi /a)$,
258: $a_{s,x}=1.6$, $a_{c,y}=1.4$, $b_{s,y}=1.3$, $b_{c,x}=-2.2$, and
259: $\phi_L=\phi_H + \pi/2$. The small $z$-components of spin
260: are not represented.
261: The planes are buckled, so that alternately spine chains are displaced
262: above and below the planes shown (but this buckling is not shown).
263: In the HTI phase the cross-ties have negligible moments and the spine
264: chains have the incommensurately modulated longitudinal moments similar
265: to those shown. The a spin components are odd under $2_x$ and
266: the b spin components are even under $2_x$, where $2_x$ is a
267: two-fold rotation about the $x$ axis.}
268: \end{figure}
269:
270: Perhaps surprisingly, it was found that ferroelectricity coincides with
271: the existence of LTI order, and this behavior was explained by
272: a phenomenological model based on a Landau expansion
273: in powers of order parameters describing ferroelectricity
274: and those needed to describe the magnetic ordering of the
275: HTI and LTI phases.\cite{FERRO}
276: The phenomenological interaction $V$ between the magnetic and
277: ferroelectric order parameters was given as
278: \begin{eqnarray}
279: V&=& \sum_{\alpha} \sum_{X,Y=L,H} a_{\alpha, X,Y} \sigma_X({\bf q})
280: \sigma_Y (-{\bf q}) P_\alpha \ ,
281: \label{VINT} \end{eqnarray}
282: where $\sigma_{X}({\bf q})$ is the order parameter describing
283: incommensurate order at wavevector ${\bf q}$ characteristic of
284: the HTI phase (for X=H) or for the additional ordering
285: appearing in the LTI phase (for X=L) and
286: $\alpha$ labels the Cartesian component of the uniform spontaneous
287: polarization vector ${\bf P}$. Using the symmetry properties of
288: the order parameters $\sigma_H({\bf q})$ and $\sigma_L({\bf q})$ it was
289: shown that only $a_{b,L,H}$ and $a_{b,H,L}$ are nonzero. This result
290: provides a phenomenological explanation for the experimental finding
291: that a nonzero polarization is induced by incommensurate magnetism
292: only in the LTI phase and then only with {\bf P} along the {\bf b} axis.
293:
294: This phenomenological theory elucidates the symmetry of the
295: magnetically induced ferroelectric state. To develop an
296: analogous microscopic theory, one could treat a Hubbard-like Hamiltonian
297: involving the 3d electrons of the Ni ions and also the 2p electrons
298: of the O ions. From such a treatment one can obtain the
299: {\it spin Hamiltonian} which describes the ``low energy''
300: sector of this Hubbard-like Hamiltonian. This low-energy
301: sector is obtained by removing states whose energy relative
302: to the low energy sector involves Coulomb integrals. In this
303: approach one develops a canonical transformation to eliminate
304: hopping matrix elements. This approach is obviously most
305: appropriate for insulators. This type of calculation was
306: formalized by Anderson\cite{PWA} who thereby obtained the
307: antiferromagnetic Heisenberg model from a superexchange mechanism.
308: More recently this approach has been applied
309: to LaCu$_2$O$_4$\cite{TY,OE} and LaTiO$_3.$\cite{RS1,RS2}
310: Here we will use the phenomenological model to deduce the form
311: of the low energy spin Hamiltonian which describes
312: magnetically induced ferroelectricity in NVO. We are
313: currently analyzing the more basic Hubbard-like Hamiltonian
314: to show how it gives rise to the spin Hamiltonian.
315:
316: From the form of Eq. (\ref{VINT}), it is clear that the spin-phonon
317: Hamiltonian we seek must be of the form
318: \begin{eqnarray}
319: V_{\rm sp-ph} &=& \sum_{i,j,k} \sum_{\alpha \beta \gamma}
320: b_{\alpha \beta \gamma_k} S_\alpha (i) S_\beta (j) Q_{\gamma_k} \ ,
321: \label{MICRO}
322: \end{eqnarray}
323: where ${\bf S}(i)$ is the vector spin operator for site $i$ and
324: $Q_{\gamma_k}$ is the $k$th normal mode amplitude at zero
325: wavevector which transforms like the $\gamma$ component of a
326: first rank tensor (vector). We will discuss the normal modes in some
327: detail in the next section. To implement the interaction of
328: Eq. (\ref{MICRO}) it is convenient to classify both the normal modes
329: and the spin components according to their transformation properties.
330: This interaction represents a linear potential, {\it i. e.}
331: a force on the phonon coordinate $Q_{\gamma_k}$. Up to
332: quadratic order in the displacements the terms in the elastic
333: potential energy $V_{\rm el}$ which depend on the $Q_{\gamma_k}$ are
334: \begin{eqnarray}
335: V_{\rm el} &=& {1 \over 2} \sum_{\gamma, k}
336: \omega_{\gamma_k}^2 Q_{\gamma_k}^2 + V_{\rm sp-ph} \ .
337: \label{HPH} \end{eqnarray}
338: Minimization of this elastic energy with respect to the phonon
339: coordinates leads to a phonon displacement which is proportional to
340: the product of two spin functions, whose symmetry we analyze below.
341: Furthermore, since the displaced ions carry an
342: electric charge (albeit an effective charge), these displacements
343: will give rise to a spontaneous polarization providing the
344: necessary spin components are nonzero. This calculation will recover
345: the symmetry properties of the phenomenological treatment.
346:
347: \section{Zone-center Phonons; Neutron Scattering Measurements and First
348: Principles Calculations}
349:
350: \subsection{Generalized Displacements}
351:
352: Since the normal modes are complicated linear combination of atomic
353: displacements we start by giving a qualitative discussion of
354: generalized displacements (GD's), which are linear combinations of
355: atomic displacements which transform according to the various irreducible
356: representations (irreps), whose characters are given in Table \ref{CHAR},
357: for the paramagnetic space group of NVO. The actual phonon modes
358: are linear combinations (which we will give later) of these basis
359: functions or GD's.
360:
361: \begin{table}
362: \begin{tabular}{||cc|cccccccc|c|}\hline\hline
363: & &$1$ & $2_y$ & $2_x$ & $2_z$ & $\cal I$ & $m_y$ &
364: $m_x$ & $m_z$ &Function\\
365: \hline
366: & $A_g$ & 1 & 1 & 1 & 1 & 1 & 1 & 1 & 1 & x$^2$,y$^2$,z$^2$ \\
367: & $A_u$ & 1 & 1 & 1 & 1 & -1 & -1 & -1 & -1 & xyz \\
368: & $B_{2g}$ & 1 & 1 & -1 & -1 & 1 & 1 & -1 & -1 & xz \\
369: & $B_{2u}$ & 1 & 1 & -1 & -1 & -1 & -1 & 1 & 1 & y\\
370: & $B_{3g}$ & 1 & -1 & 1 & -1 & 1 & -1 & 1 & -1 & yz \\
371: & $B_{3u}$ & 1 & -1 & 1 & -1 & -1 & 1 & -1 & 1 & x \\
372: & $B_{1g}$ & 1 & -1 & -1 & 1 & 1 & -1 & -1 & 1 & xy \\
373: & $B_{1u}$ & 1 & -1 & -1 & 1 & -1 & 1 & 1 & -1 & z \\
374: \hline\hline
375: \end{tabular}
376: \caption{\label{CHAR}Irreducible representation of the
377: paramagnetic space group of NVO. The vector representations
378: are $B_{1u}$, $B_{2u}$, and $B_{3u}$ which transform like
379: $z$, $y$, and $x$, respectively.}
380: \end{table}
381:
382: We are only interested in GD's which transform according to the
383: vector irreps $B_{nu}$, for $n=1,2,3$. We now
384: give a qualitative discussion of these vector GD's.
385: It is actually quite easy to generate the GD's which
386: transform like vector components $x$, $y$, and $z$. First of all,
387: one sees that assigning all atoms of a given Wykoff orbit the
388: {\it same} displacement along the $\alpha$-axis will give a GD
389: which transforms under the symmetry operations of the
390: space group (given in Table \ref{PUC}) like the $\alpha$ component
391: of a vector. Since NVO has 6 crystallographically inequivalent sites
392: this construction gives six GD's along each of the
393: three coordinate axes, which we denote $x_n$, $y_n$, and $z_n$, for
394: $n=1, 2, \dots 6$. We now discuss the construction of the less trivial
395: GD's which transform like a vector and which are
396: shown in Figs. \ref{GENDIS}, \ref{GENDIS2}, and \ref{GENDIS3}.
397: To generate these results, one can start with
398: an arbitrarily chosen site to which a vector displacement along one
399: of the three coordinate axes is specified. Then one generates the
400: displacements of the other sites in the Wykoff orbit
401: so as to reproduce the desired transformation properties.
402: For example, to construct a $z$-like mode on the spines, we could
403: assign the lower left spine site (in the lower left panel of Fig.
404: \ref{GENDIS}) a displacement along the $x$-axis.
405: To be a $z$-like mode the pattern of displacements should be even
406: under $m_x$, which fixes the displacement of the lower right spine
407: to be that shown. Such a $z$-like mode should be odd under a two-fold
408: rotation about an $x$-axis passing through the center of the cell.
409: Applying this operation to the two lower spine sites fixes the orientation
410: of the displacements of the upper spine sites.
411: The other four symmetry operations give these
412: same displacements. Had we started with a spine site with a displacement
413: along the $y$-axis, we would have gotten a null displacement because
414: this symmetry with displacements along the $y$-axis is not allowed. Had
415: we fixed the first site to have a displacement along the $z$-axis we would
416: have found the trivial mode in which all sites are displaced in parallel.
417: The other GD's shown in the figures were generated in
418: this way. Of course, the actual normal modes (phonons) consist of linear
419: combinations of basis functions having the same irrep label and they will
420: be discussed in the next subsection.
421:
422: \begin{figure}
423: \includegraphics[height=6cm]{PH.eps}
424: \caption{\label{GENDIS} Generalized displacements
425: $y_7$, $z_7$, $z_8$, and $x_7$ which transform like the
426: components of a vector, for cross-tie sites (upper panels)
427: and spine sites (lower panels). Atomic displacements (for
428: the GD's indicated by the labels) in the
429: $x$-$z$ plane are represented by arrows, whereas those in
430: the $+y$ direction ($-y$ direction) are indicated by
431: filled (open) circles.
432: It is easy to see that these modes couple to the uniform
433: displacements. For instance, consider the mode $z_8$ shown
434: here and in particular consider how the ionic displacements
435: of the spine sites which are shown affect
436: the cross-tie sites (not shown) at
437: the corners and centers of the square. Imagine the ion-ion
438: interactions to be repulsive. Then the nearest neighbors
439: at negative $z$ relative to each cross-tie get closer to the
440: cross tie and the nearest neighbors at positive $z$ relative to
441: each cross tie get farther away from the cross tie. Thus, all cross-ties
442: are squeezed towards positive $z$. This same reasoning also shows
443: that even though this motion is confined to the $x$-direction, it will
444: induce a dipole moment along the $z$-direction.}
445: \end{figure}
446:
447: \begin{figure}
448: \includegraphics[height=3cm]{PH2.eps}
449: \caption{\label{GENDIS2} As Fig. \protect{\ref{GENDIS}}. Here we illustrate
450: schematically the vector GD's $z_9$, $z_{10}$, $z_{11}$, $y_8$, $y_9$, and
451: $y_{10}$ for f sites. The
452: placement of the sites reproduces the symmetry of an f site and
453: is not quantitative for either V or O atoms. Three distinct f sites
454: are occupied, one by a V atom and the other two by O atoms.}
455: \end{figure}
456:
457: \begin{figure}
458: \includegraphics[height=5.5cm]{PH3.eps}
459: \caption{\label{GENDIS3} As Fig. \protect{\ref{GENDIS2}}. Here we illustrate
460: schematically the vector GD's $y_{11}$, $x_8$, $x_9$, $z_{12}$, $y_{12}$, and
461: $z_{13}$ for g sites.}
462: \end{figure}
463:
464: \subsection{Normal Modes}
465:
466: In this section we present inelastic neutron scattering (INS)
467: measurements of the phonon density of states (DOS) in NVO
468: along with the first-principles calculations of the zone-center
469: phonons. We will identify the phonons which have the correct
470: symmetry to induce a spontaneous polarization. We will also
471: attempt to estimate the local distortion which gives rise to
472: the observed dipole moment in NVO.
473:
474: The INS measurements were performed using the filter analyzer spectrometer (FANS)
475: located on beamline BT4 at the NIST Center for Neutron Research \cite{fans}.
476: For energies above 40 meV, a Cu(220) monochromator, surrounded by
477: $60'-40'$ horizontal collimation and
478: combined with a cooled polycrystalline beryllium filter analyzer was used.
479: For the low energy spectrum (i.e. $ E < 40 $ meV), a graphite (PG) monochromator
480: with $20'-20'$ collimation was used. The relative energy resolution of the FANS
481: instrument is approximately 5\% in the energy range probed. The powder
482: Ni$_3$V$_2$O$_8$ sample (about 20 grams) was held at 12 K (paramagnetic phase)
483: and 8 K (HTI phase) with a helium-filled aluminum can using a closed-cycle
484: He$^3$ refrigerator.
485:
486: The first-principles total-energy and phonon calculations were performed
487: by the plane-wave implementation of the spin-polarized generalized gradient
488: approximation (SP-GGA) to density functional theory (DFT).\cite{pwscf}
489: We used $4\times 4\times 3$ k-points according to Monkhorst-pack scheme
490: and Vanderbilt ultra-soft pseudopotentials for which a cutoff energy
491: of 400 eV was found to be enough for total energies to converge within 0.5 eV/atom.
492: We considered the primitive unit cell of the NVO which contains 26 atoms
493: as listed in Table~II. Experimental lattice parameters were used in the
494: calculations but the atomic positions were optimized to eliminate the
495: forces down to 0.02 eV/\AA. The optimized positions are listed in
496: Table.~II, which are in excellent agreement with the experimental positions.
497: Using the optimized structure, we next calculated the zone-center phonons
498: and the corresponding INS one-phonon spectrum as described in Ref.
499: \onlinecite{taner_phonon}.
500:
501: \begin{figure}
502: \includegraphics[width=8cm]{nvodipole.eps}
503: \caption{\label{ins_dos} The observed and calculated INS spectrum. The black
504: vertical bars show the B$_{2u}$ phonons whose intensity is proportional
505: to the induced dipole moment when the system is distorted by the
506: zero-point rms values of the modes.}
507: \end{figure}
508:
509:
510: The measured INS spectrum along with the calculation is shown in Fig.~\ref{ins_dos}.
511: We observed almost identical spectra in the paramagnetic ($T=12$ K) and
512: the HTI ($T=8$ K) phases and therefore show only the combined spectrum.
513: The agreement of the calculations to the observed spectrum is quite good,
514: giving further confidence that the first-principle calculations capture
515: the main physics. It also suggests that the phonon modes in NVO have small
516: dispersion with wavevector. This is because the INS spectrum is approximately
517: averaged over a large range of ${\bf Q}$ and the calculations are only for $Q=0$.
518: The biggest difference between the INS spectrum and calculation
519: is for the observed feature near 80 meV, which is calculated to be around
520: 70 meV. As we discuss in detail below, interestingly this phonon has
521: the right symmetry and the atomic displacement vector to induce a large dipole moment.
522: Hence, maybe the disagreement for the energy of this mode could be some indication
523: of strong spin-phonon coupling.
524:
525: In order to identify the right phonon modes that can induce the observed
526: dipole moment along the {\bf b}-axis in NVO, we carried out the symmetry analysis of the
527: zone center phonons. Table \ref{CHAR} shows the character table for the
528: irreducible representations (irreps) of the group $G_{\bf v}$ for optical
529: phonons at zero wavevector ${\bf v}$. (For a review of group theory see Ref.
530: \onlinecite{GROUP}.) There are 26 atoms in the primitive unit cell and
531: the representation $\Gamma_u$ induced by the vector space of these
532: $26 \times 3 = 78$ atomic displacements has the decomposition
533: \begin{eqnarray}
534: \Gamma_u &=& 10 A_g + 8 A_u + 8 B_{1g} + 13 B_{1u} \nonumber \\
535: && \ + 7 B_{2g} + 12 B_{2u} + 11 B_{3g} + 9 B_{3u} \ .
536: \end{eqnarray}
537: One can check that the vector representations
538: which transform like $x$, $y$, and $z$ are $B_{3u}$,
539: $B_{2u}$, and $B_{1u}$, respectively. To discuss the
540: spontaneous polarization these
541: are the only irreps we will need to consider.
542: Among the 78 phonons, twelve have B$_{2u}$ symmetry,
543: and can therefore produce the observed\cite{FERRO} spontaneous polarization along
544: the ${\bf b}$-axis. One of the these twelve modes is acoustic (i.e.
545: all atoms move uniformly along the {\bf b}-axis) and will not be considered any further.
546: To calculate the phonon energies and wavefunctions we found the
547: eigenvalues $\omega^2_n$ of the matrix ${\bf W}(q=0)$, which is related to the
548: Fourier transform of the potential energy matrix by
549: $W_{\tau, \alpha ; \tau' , \alpha'}(q=0) = \sqrt{M_\tau} V_{\tau,\alpha;
550: \tau',\alpha'} (q=0) \sqrt{M_{\tau'}}$, where $\tau$ labels sites within
551: the primitive unit cell, $\alpha$ labels Cartesian components, and
552: $M_\tau$ is the mass of the atom at site $\tau$. The
553: corresponding eigenvectors $O_{\tau, \alpha}^{(n)}$ are given in Table \ref{b2u}.
554: Thus the displacements within the unit cell are given in terms of the
555: normal modes operators $Q_n$ as
556: \begin{eqnarray}
557: u_{\tau, \alpha} &=& \sum_n O_{\tau,\alpha}^{(n)} M_\tau^{-1/2} Q_n \nonumber \\
558: &=& \sum_n O_{\tau,\alpha}^{(n)} \sqrt{ {\hbar \over 2 M_\tau \omega_n}}
559: (a_n^\dagger + a_n ) \ ,
560: \end{eqnarray}
561: where $a_n^\dagger$ is a phonon creation operator. The energies of the
562: eleven optical modes are shown in Fig.~\ref{ins_dos} by black bars whose height
563: for mode $n$ is proportional to the average polarization ${\bf P}^{(n)}_y$ of
564: that mode along the $y$ axis. The polarization vector of the $n$th mode
565: is estimated from the following formula:
566: \begin{equation}
567: P_{{\rm rms},\alpha}^{(n)} = \frac{1}{\Omega_{\rm uc}} \sum_{\tau} q_\tau
568: O_{\tau ,\alpha}^{(n)} Q_{\rm rms} M_\tau^{-1/2} \ ,
569: \end{equation}
570: where $\Omega_{\rm uc}$ is the volume of the unit cell and
571: $Q_{\rm rms}=\sqrt{\hbar/2\omega}$
572: is the average zero-point fluctuation. As we can see from
573: Fig.~\ref{ins_dos}, half of the B$_{2u}$ modes induce relatively small
574: dipole moments. This is due to the fact that for these
575: phonons, atoms mainly oscillate along the ${\bf c}$-axis and the
576: $b$-component of the motion is only a second order effect.
577: However for the other half, the motion is directly along the
578: ${\bf b}$-axis and therefore the induced dipole moment is significant.
579: Animations of these modes and more information can be obtained at
580: Ref. \onlinecite{taner_web}. We note that two particular phonons, one at 36 meV
581: and the other around 70 meV, induce a significantly large dipole moment.
582: The eigenvectors of these and the other $B_{2u}$ modes along with
583: the magnitude of the induced dipole moments are given in Table~\ref{b2u}.
584: \begin{table*}
585: \begin{tabular}{|c|cc|c|cc|cc|cc|ccc|c|c|c|}\hline\hline
586: %%
587: & \multicolumn {2} {|c|} {Ni(1)} & Ni(2) &
588: \multicolumn{2} {|c|} {V(1)} & \multicolumn {2} {|c|} {O(1)} &
589: \multicolumn{2} {|c|} {O(2)} & \multicolumn {3} {|c|} {O(3)} &
590: $\omega$ & $\frac{Q_{rms}}{\sqrt{m_p}}$& P$_{rms}$ \\
591: %%
592: & \multicolumn{2}{|c|} {$q=0.90$e} & $q=0.86$e &
593: \multicolumn{2} {|c|} {$q=1.18$e} & \multicolumn {2} {|c|} {$q=-0.62$e} &
594: \multicolumn {2} {|c|} {$q=-0.68$e} & \multicolumn{3} {|c|} {$q=-0.59$e} &
595: (meV) & (\AA)& ($10^{-4}$ C/m$^{2}$) \\ \hline\hline
596: %%%%
597: Mode& \multicolumn {2} {|c|} {$(0,y,z)$} & $(0,y,0)$ &
598: \multicolumn{2} {|c|} {$(0,y,z)$} & \multicolumn {2} {|c|} {$(0,y,z)$} &
599: \multicolumn{2} {|c|} {$(0,y,z)$} & \multicolumn {3} {|c|} {$(x,y,z)$} &
600: & & \\
601: %%%%
602: & $y$ & $z$ & $y$ & $y$ & $z$ & $y$ & $z$ & $y$ & $z$ & x & $y$ & $z$ &
603: & & \\ \hline
604: %%%%
605: \#4 & 0.023 & -0.446 & -0.119 & 0.040 & 0.169 &-0.032 & -0.015 & 0.025 & -0.314&
606: -0.024 & 0.005 & 0.048 & 9.2 & 0.47 & 2.5 \\ \hline
607: %
608: \#16 & 0.551 & -0.007 & -0.209 &-0.094 &-0.063 &-0.099 & -0.111 &-0.061 & 0.069&
609: 0.013 & 0.073 & -0.001 &21.3 & 0.31 & 1.7 \\ \hline
610: %
611: \#27 &-0.205 & 0.058 & -0.110 & 0.206 & 0.172 &-0.006 & -0.354 & 0.023 & -0.061&
612: 0.071 & 0.041 & 0.009 &30.4 & 0.26 & 3.3 \\ \hline
613: %
614: \#29 &-0.008 & -0.428& 0.019 & 0.010 &-0.265 & 0.090 & 0.050 &-0.053 & 0.086&
615: -0.020 &-0.040 &-0.178 &31.1 & 0.26 & 6.1 \\ \hline
616: %
617: \#34 &-0.164 & 0.066 & -0.286 & 0.073 & 0.049 & 0.109 & 0.163 & 0.103 & 0.077&
618: -0.073 & 0.188 &-0.068 &36.2 & 0.24 &66.1 \\ \hline
619: %
620: \#40 & 0.178 & 0.113 & -0.166 & 0.162 &-0.012 & 0.157 & 0.076 & 0.171 & -0.012&
621: 0.029&-0.237 & 0.007 &38.9 & 0.23 &18.7 \\ \hline
622: %
623: \#49 & 0.069 & 0.307 & 0.021 &-0.056 &-0.138 & 0.023 & 0.057 &-0.069 & -0.348&
624: -0.068 & 0.015 &-0.139 &44.4 & 0.22 & 0.5\\ \hline
625: %
626: \#53 &-0.014 & -0.022 & -0.037 &-0.043 & 0.165 & 0.143 & 0.182 &-0.112 & -0.080&
627: 0.212 & 0.066 & 0.155 &49.8 & 0.20 & 16.3 \\ \hline
628: %
629: \#64 & 0.037 & 0.041 & -0.030 & 0.264 & 0.110 & 0.015 & 0.047 &-0.396 & 0.042&
630: -0.038 &-0.027 &-0.010 &69.2 & 0.17 & 46.1 \\ \hline
631: %
632: \#70 & 0.011 & -0.011 & 0.001 &-0.155 & 0.017 & 0.365 & -0.169 &-0.081 & 0.011&
633: -0.143 & -0.004 & 0.090 &93.3 & 0.15 & 23.0 \\ \hline
634: %
635: \#78 & 0.010 & 0.019 & 0.004 &-0.089 & 0.238 & 0.150 & -0.099 &-0.029 & 0.000&
636: 0.203 & 0.009 &-0.187 &103.0& 0.14 & 11.3\\ \hline
637: \end{tabular}
638: \caption{\label{b2u} Mass-weighted atomic displacements of B$_{2u}$ phonons which induce
639: a dipole moment along the ${\bf b}$ axis (normalized so that the sum of the
640: squares of the components equals unity for each mode). (The acoustic B$_{2u}$
641: is not tabulated.) Each component of the
642: mass-weighted displacement represents the atomic
643: displacement divided by the square root of the respective atomic mass. The
644: mass-weighted displacements are given for the sites listed in column
645: \#2 of Table \ref{orbit}. The displacements of the remaining atoms in
646: each Wyckoff orbit are fixed so that the mode transforms like B$_{2u}$, i. e.
647: like the $y$-component of a vector. The calculated atomic charges, magnitude
648: of the rms displacement $Q_{\rm rms}$ (where $m_p$ is the proton mass),
649: the rms dipole moment, and the mode energy is also given.}
650: \end{table*}
651:
652: \begin{figure}
653: \includegraphics[width=8cm]{mode34_64_v2.eps}
654: \caption{\label{mode34_64} A schematic representation of the top view
655: of two particularly interesting B$_{2u}$ modes whose displacement vectors
656: are given in Table~\ref{b2u}.}
657: \end{figure}
658:
659:
660: Figure~\ref{mode34_64} shows schematically
661: how the oxygen atoms move in these two particularly interesting phonons.
662: For the low-energy mode at 36 meV, the two oxygen atoms
663: connecting the spine-spins move in the same direction. Therefore
664: while one of the Ni-O-Ni bond angle
665: decreases, the other Ni-O-Ni bond angle increases.
666: Hence at first order, we do not expect large changes in the Ni-O-Ni
667: superexchange due to this phonon. On the other hand, for the $E=70$
668: meV mode, only one oxygen (which is connected to the cross-tie Ni spin)
669: moves along the ${\bf b}$-axis. Hence, in
670: this case, only one of the Ni-O-Ni bond angles
671: changes from nearly $90^{\rm o}$ and
672: therefore we expect this phonon to have important effects on the
673: Ni-O-Ni superexchange interaction. Interestingly, the biggest disagreement
674: between the experimental data
675: and the calculated phonon energies happens for this phonon, which further
676: suggests that it may have strong spin-phonon coupling.
677:
678: Finally, we discuss what kind of distortion is needed in order to
679: induce the experimentally observed dipole moment whose magnitude is
680: about $P_{\rm exp}=1.25\times10^{-4}$ C/m$^{2}$. We note that
681: this induced dipole moment is much smaller than the calculated rms dipole
682: moment
683: ($P_{\rm rms}~ \sim 46-70 \times10^{-4}$ C/m$^{2}$) listed in Table~\ref{b2u}.
684: This suggests that the local distortion should be at the order of
685: $P_{\rm exp}/P_{\rm rms}$, which suggests that atoms move less than
686: 0.001 \AA. This is a quite small distortion and would be very difficult
687: to observe directly by neutron powder diffraction.
688:
689: \section{Symmetry of the Spin Structure}
690:
691: In Refs. \onlinecite{FERRO}, \onlinecite{EXPT}, \onlinecite{REV}, and
692: \onlinecite{THEORY} the application of
693: representation theory to the determination and characterization of
694: magnetic structures is discussed in detail. In Table \ref{CHARQ} we
695: give the character table for the irreps for an arbitrary wavevector
696: of the form $(q,0,0)$. Because the irreps are one dimensional,
697: each spin basis function is an eigenvector of the symmetry operator
698: with the listed eigenvalue.
699:
700: \begin{table}
701: \begin{tabular}{||cc|cccc||}\hline\hline
702: & &$1$ & $2_x$ & $\tilde m_y$ & $\tilde m_z$ \\ \hline
703: & $\Gamma^1$ & 1 & 1 & 1 & 1 \\
704: & $\Gamma^2$ & 1 & 1 & $-1$ & $-1$ \\
705: & $\Gamma^3$ & 1 & $-1$ & $1$ & $-1$ \\
706: & $\Gamma^4$ & 1 & $-1$ & $-1$ & $1$ \\ \hline\hline
707: \end{tabular}
708: \caption{\label{CHARQ}Irreducible representations of
709: the group $G_{\bf v}$ for the incommensurate magnetic structure with
710: ${\bf v}=(q,0,0)$. Here it is simplest to use the symmetry operations
711: $\tilde m_y$ and $\tilde m_z$, such that $\tilde m_y {\bf r}=
712: (x, \overline y + \oh , z + \oh )$ and $\tilde m_z {\bf r} =
713: (x , y + \oh , \overline z + \oh )$.}
714: \end{table}
715:
716: In these references it is shown that
717: the HTI phase is described by a set of five complex amplitudes
718: associated with the irrep $\Gamma_4$. Here we call these
719: $a_{s,x}$, $ia_{s,y}$, and $a_{s,z}$ to describe
720: the orientation of the spine spins and $a_{c,y}$ and $a_{c,z}$ to
721: describe the orientation of the cross-tie spins. When the LTI phase is
722: entered additional variables associated with the irrep $\Gamma_2$
723: become nonzero. These LTI variables are here denoted
724: $ib_{s,x}$, $b_{s,y}$, and $ib_{s,z}$ to describe
725: the orientation of the spine spins and $b_{c,z}$ to describe the
726: orientation of the cross tie spins. Because the crystal
727: is centrosymmetric, it is shown\cite{EXPT,REV,THEORY}
728: that within a given representation all these complex
729: structural parameters (the $a$'s and $b$'s)
730: can be written in terms of a real amplitude times a complex
731: phase factor which is the same for all variables of the same
732: irrep, $\Gamma_4$ or $\Gamma_2$, in the sense that
733: \begin{eqnarray}
734: a_{t,\alpha} &=& \pm |a_{t,\alpha}| e^{i \phi_{HTI}} \ , \ \ \
735: b_{t,\alpha} = \pm |b_{t,\alpha}| e^{i \phi_{LTI}} \ ,
736: \label{PHASEEQ} \end{eqnarray}
737: where $t$ denotes either spine or cross-tie. It is further
738: expected\cite{EXPT,THEORY} that
739: $\phi_{\rm HTI} - \phi_{\rm LTI}$ is $\pm \pi /2$.
740: Thus in these two phases we may use the results of Table VIII in
741: Ref. \onlinecite{EXPT} to write the spin components of
742: the six Ni ions in the unit cell as
743: \begin{eqnarray}
744: S_x^{(1)}({\bf R}_1) &=& (a_{s,x}+ib_{s,x}) e^{i{\bf q} \cdot {\bf R}_1 }
745: \nonumber \\ && \ + (a_{s,x}^* -ib_{s,x}^*) e^{-i{\bf q}\cdot {\bf R}_1 }
746: \label{SPINS1} \\
747: S_y^{(1)}({\bf R}_1) &=& (ia_{s,y}+b_{s,y}) e^{i{\bf q} \cdot {\bf R}_1 }
748: \nonumber \\ && \ + (- ia_{s,y}^*+b_{s,y}^*) e^{-i{\bf q}\cdot{\bf R}_1 } \\
749: S_z^{(1)}({\bf R}_1) &=& (a_{s,z}+ib_{s,z}) e^{i{\bf q} \cdot {\bf R}_1 }
750: \nonumber \\ && \ + (a_{s,z}^*-ib_{s,z}^*) e^{-i{\bf q} \cdot {\bf R}_1 } \\
751: S_x^{(2)}({\bf R}_2) &=& (-a_{s,x}+ib_{s,x}) e^{i{\bf q} \cdot{\bf R}_2 }
752: \nonumber \\ && \ + (- a_{s,x}^* -ib_{s,x}^*)e^{-i{\bf q}\cdot{\bf R}_2 } \\
753: S_y^{(2)}({\bf R}_2) &=& (ia_{s,y}-b_{s,y})e^{i{\bf q} \cdot{\bf R}_2 }
754: \nonumber \\ && \ + (-i a_{s,y}^*-b_{s,y}^*)e^{-i{\bf q} \cdot{\bf R}_2 } \\
755: S_z^{(2)}({\bf R}_2) &=& (a_{s,z}-ib_{s,z}) e^{i{\bf q} \cdot{\bf R}_2 }
756: \nonumber \\ && \ + (a_{s,z}^*+ib_{s,z}^*) e^{-i{\bf q} \cdot{\bf R}_2 } \\
757: S_x^{(3)}({\bf R}_3) &=& (a_{s,x}-ib_{s,x}) e^{i{\bf q} \cdot{\bf R}_3 }
758: \nonumber \\ && \ + (a_{s,x}^*+ib_{s,x}^*) e^{-i{\bf q} \cdot{\bf R}_3 } \\
759: S_y^{(3)}({\bf R}_3) &=& (-ia_{s,y}+b_{s,y}) e^{i{\bf q} \cdot{\bf R}_3 }
760: \nonumber \\ && \ +(i a_{s,y}^*+b_{s,y}^*) e^{-i{\bf q} \cdot{\bf R}_3 } \\
761: S_z^{(3)}({\bf R}_3) &=& (a_{s,z}-ib_{s,z}) e^{i{\bf q} \cdot{\bf R}_3 }
762: \nonumber \\ && \ + (a_{s,z}^*+ib_{s,z}^*) e^{-i{\bf q} \cdot{\bf R}_3 } \\
763: S_x^{(4)}({\bf R}_4) &=& (-a_{s,x}-ib_{s,x}) e^{i{\bf q} \cdot{\bf R}_4 }
764: \nonumber \\ && \ + (-a_{s,x}^*+ib_{s,x}^*) e^{-i{\bf q} \cdot{\bf R}_4 } \\
765: S_y^{(4)}({\bf R}_4) &=& (-ia_{s,y}-b_{s,y}) e^{i{\bf q} \cdot{\bf R}_4 }
766: \nonumber \\ && \ +(ia_{s,y}^*-b_{s,y}^*) e^{-i{\bf q} \cdot{\bf R}_4 } \\
767: S_z^{(4)}({\bf R}_4) &=& (a_{s,z}+ib_{s,z}) e^{i{\bf q} \cdot{\bf R}_4 }
768: \nonumber \\ && \ + (a_{s,z}^*-ib_{s,z}^*) e^{-i{\bf q} \cdot{\bf R}_4 } \\
769: S_x^{(5)}({\bf R}_5) &=& b_{c,x} e^{i{\bf q} \cdot {\bf R}_5}
770: + b_{c,x}^* e^{-i{\bf q} \cdot {\bf R}_5} \\
771: S_y^{(5)}({\bf R}_5) &=& a_{c,y} e^{i{\bf q} \cdot {\bf R}_5}
772: + a_{c,y}^* e^{-i{\bf q} \cdot {\bf R}_5} \\
773: S_z^{(5)}({\bf R}_5) &=& a_{c,z} e^{i{\bf q} \cdot {\bf R}_5}
774: + a_{c,z}^* e^{-i{\bf q} \cdot {\bf R}_5} \\
775: S_x^{(6)}({\bf R}_6) &=& -b_{c,x} e^{i{\bf q} \cdot {\bf R}_6}
776: -b_{c,x}^* e^{-i{\bf q} \cdot {\bf R}_6} \\
777: S_y^{(6)}({\bf R}_6) &=& -a_{c,y} e^{i{\bf q} \cdot {\bf R}_6}
778: - a_{c,y}^* e^{-i{\bf q} \cdot {\bf R}_6} \\
779: S_z^{(6)}({\bf R}_6) &=& a_{c,z} e^{i{\bf q} \cdot {\bf R}_6}
780: + a_{c,z}^* e^{-i{\bf q} \cdot {\bf R}_6} \ .
781: \label{SPINS2} \end{eqnarray}
782: Here the superscripts on 1, 2, 3, 4 on ${\bf S}$ label the spine
783: sites and 5 and 6 label the cross-tie sites c and c$^\prime$,
784: respectively, (as in Fig. \ref{SPSP}, below)
785: and ${\bf R}_n$ is the position of a spin $n$. Also
786: ${\bf q}=q\hat i$ with $q \approx 0.28(2 \pi /a)$.\cite{LawesKenzelmann,EXPT}
787:
788: \section{Relations for the Strain Derivative of the Exchange Tensor}
789:
790: In this section we obtain explicit forms for the most important spin-phonon
791: coupling matrices. For this purpose we start by introducing notation
792: for the principal exchange interactions. We write the interactions between
793: spins on sites $i$ and $j$ as
794: \begin{eqnarray}
795: {\cal H} (i,j) &=& \sum_{\alpha \beta} X_{\alpha \beta} (i,j)
796: S_\alpha (i) S_\beta (j) \ ,
797: \end{eqnarray}
798: where $X_{\alpha \beta} (i,j) = X_{\beta \alpha}(j,i)$, of course.
799: For nearest neighbor (nn) interactions between spine spins we set
800: $X_{\alpha \beta} (i,j)=U_{\alpha \beta} (i,j)$, for next nearest neighbor
801: (nnn) interactions between spine spins we set
802: $X_{\alpha \beta} (i,j)=V_{\alpha \beta} (i,j)$, and for nn interactions
803: between spine and cross-tie spins we set
804: $X_{\alpha \beta} (i,j)=W_{\alpha \beta} (i,j)$.
805: We may further decompose the exchange tensor into its symmetric
806: and antisymmetric parts. For example, for nn spine-spine interactions
807: we write (omitting the site labels $i$ and $j$)
808: \begin{eqnarray}
809: {\bf U} &=& \left[ \begin{array} {c c c}
810: J_{xx} & J_{xy} + D_z & J_{xz} - D_y \\
811: J_{xy} - D_z & J_{yy} & J_{yz} + D_x \\
812: J_{xz} + D_y & J_{yz} - D_x & J_{zz} \\
813: \end{array} \right] \ ,
814: \end{eqnarray}
815: where ${\bf D}$ is the Dzyaloshinskii-Moriya (DM) vector.\cite{DM1,DM2}
816: Similar decompositions will be made for ${\bf V}$ and ${\bf W}$ in terms
817: of symmetric tensors ${\bf K}$ and ${\bf L}$, respectively, and the
818: DM vectors ${\bf E}$ and ${\bf F}$, respectively.
819:
820: Now we consider the gradient expansion of these exchange tensors.
821: From the above development we write the displacement ${\bf u}$
822: of the $\tau$th atom in the unit cell in terms of the amplitude
823: $Q_{\gamma_p}$ of the GD labeled $\gamma_p$ as
824: \begin{eqnarray}
825: u_\alpha ({\bf R}, \tau ; \gamma_p) &=&
826: Q_{\gamma_p} a(\gamma_p; \tau, \alpha ) \ ,
827: \end{eqnarray}
828: where $a(\gamma_p; \tau, \alpha )$ is the $\alpha$-component of the displacement
829: of atom $\tau$ in unit cell at ${\bf R}$ for the GD
830: labeled $\gamma_p$ (which are shown in Figs. \ref{GENDIS},
831: \ref{GENDIS2}, and \ref{GENDIS3}). If $Z$
832: represents a component of an exchange tensor, then we write
833: \begin{eqnarray}
834: {\partial Z \over \partial Q_{\gamma_p}} &=& \sum_{\alpha {\bf R} \tau}
835: {\partial Z \over \partial u_\alpha ({\bf R}, \tau; \gamma_p)}
836: a(\gamma_p; \tau, \alpha ) \ .
837: \end{eqnarray}
838: The aim of the present paper is to determine which such derivatives are
839: required to completely determine the trilinear spin-phonon coupling.
840: The actual calculation of these coefficients will be undertaken separately.
841: Thus, for each GD $\gamma_p$, we consider the interaction
842: \begin{eqnarray}
843: {\cal H}_{\gamma_p} &=& Q_{\gamma_p} \sum_{\alpha \beta} \sum_{ij}
844: {\partial X_{\alpha \beta} (i,j) \over \partial Q_{\gamma_p} }
845: S_\alpha (i) S_\beta (j) \ .
846: \label{FEINT} \end{eqnarray}
847: Our objective is to express the results for the spontaneous polarization
848: due to the trilinear coupling in terms of the parameters
849: $\partial X_{\alpha \beta} (i,j) / \partial Q_{\gamma_p}$.
850:
851: Here we analyze the gradients of the nn interactions between spine spins.
852: (Similar analyses of next-nearest neighbor spine interactions
853: and of nn spine-cross tie interactions are given in Appendices.)
854: We introduce the coupling between spine sites \#1 and
855: \#4 in Fig. \ref{SPSP}:
856: \begin{eqnarray}
857: {\partial U_{\alpha \beta} (1,4) \over \partial Q_{\gamma_p}} &\equiv &
858: U_{\alpha \beta}^{\gamma_p} \ .
859: \end{eqnarray}
860: Clearly ${\bf U}^{\gamma_p}$ has to be invariant under the symmetry operations
861: of the crystal. But $m_x$ takes the bond in question into itself
862: (and interchanges indices).
863: This indicates that these interaction matrices must satisfy
864: \begin{eqnarray}
865: \sigmav_x {\bf U}^{x_p} \sigmav_x &=& - \tilde {\bf U}^{x_p} \nonumber \\
866: \sigmav_x {\bf U}^{y_p} \sigmav_x &=& \tilde {\bf U}^{y_p} \nonumber \\
867: \sigmav_x {\bf U}^{z_p} \sigmav_x &=& \tilde {\bf U}^{z_p} \ ,
868: \label{SIGEQ} \end{eqnarray}
869: where tilde indicates transpose,
870: \begin{eqnarray}
871: \sigmav_x &=& \left[ \begin{array} {c c c} -1 & 0 & 0 \\ 0 &1 & 0
872: \\ 0 & 0 & 1 \\ \end{array} \right] \ , \nonumber
873: \end{eqnarray}
874: and later
875: \begin{eqnarray}
876: \sigmav_y &=& \left[ \begin{array} {c c c} 1 & 0 & 0 \\ 0 & -1
877: & 0 \\ 0 & 0 & 1 \\
878: \end{array} \right] \ , \nonumber
879: \end{eqnarray}
880: \begin{eqnarray}
881: \sigmav_z &=& \left[ \begin{array} {c c c} 1 & 0 & 0 \\ 0 & 1 & 0
882: \\ 0 & 0 & -1 \\ \end{array} \right] \ .
883: \end{eqnarray}
884: Here we used the fact that the GD's were constructed to have known symmetry:
885: \begin{eqnarray}
886: m_\alpha Q_{\beta_p} = (1 - 2 \delta_{\alpha ,\beta}) Q_{\beta_p}
887: \label{SIGEQ1}
888: \end{eqnarray}
889: and
890: \begin{eqnarray}
891: 2_\alpha Q_{\beta_p} &=& (-1 + 2 \delta_{\alpha , \beta}) Q_{\beta_p} \ ,
892: \label{SIGEQ2}
893: \end{eqnarray}
894: where $\delta_{a,b}$ is unity if $a=b$ and is zero otherwise.
895:
896: \begin{figure}
897: \includegraphics[height=5.5cm]{SPSP.eps}
898: \caption{\label{SPSP} Diagram of an ${\bf a}-{\bf c}$ plane
899: used to specify nearest-neighbor and next-nearest-neighbor interactions
900: along a single spine. Circles are spine sites and square are cross-tie
901: sites and $d=0.13b$ (see Table \protect{\ref{orbit}}).
902: The dashed rectangle indicates the unit cell. Interactions in
903: other ${\bf a}-{\bf c}$ planes are obtained by using translation symmetry.}
904: \end{figure}
905:
906: In view of Eq. (\ref{SIGEQ}), we have
907: \begin{eqnarray}
908: {\bf U}^{x_p} &=& \left[ \begin{array} {c c c}
909: 0 & J_{xy}^x & J_{xz}^x \\ J_{xy}^x & 0 & D_x^x \\ J_{xz}^x & -D_x^x & 0 \\
910: \end{array} \right] \ , \nonumber \\
911: {\bf U}^{y_p} &=& \left[ \begin{array} {c c c}
912: J_{xx}^y & D_z^y & -D_y^y \\ -D_z^y & J_{yy}^y &
913: J_{yz}^y \\ D_y^y & J_{yz}^y & J_{zz}^y \\ \end{array} \right] \ ,
914: \nonumber \\ {\bf U}^{z_p} &=& \left[ \begin{array} {c c c}
915: J_{xx}^z & D_z^z & -D_y^z \\ -D_z^z & J_{yy}^z & J_{yz}^z \\
916: D_y^z & J_{yz}^z & J_{zz}^z \\ \end{array} \right] \ ,
917: \end{eqnarray}
918: where the index $p$ on the superscripts of $J$ and $D$ are left implicit
919: and $J_{\alpha \beta}^{\gamma_p}$ and $D_\alpha^{\gamma_p}$
920: (and similarly later for superscripts on ${\bf K}$, ${\bf L}$, ${\bf E}$,
921: and ${\bf F}$) are defined to be
922: \begin{eqnarray}
923: J^{\gamma_p}_{\alpha \beta} &\equiv &
924: {\partial J_{\alpha \beta} (1,4) \over \partial Q_{\gamma_p}} \ , \ \ \
925: D^{\gamma_p}_\alpha \equiv
926: {\partial D_\alpha (1,4) \over \partial Q_{\gamma_p}} \ .
927: \end{eqnarray}
928: Then we obtain the \#2-\#3 interaction from the above by $2_x$, a two-fold
929: rotation about the $x$-axis, so that
930: \begin{eqnarray}
931: {\partial U_{\alpha \beta} (2,3) \over \partial Q_{x_p}} &= &
932: \sigmav_y \sigmav_z {\bf U}^{x_p} \sigmav_y \sigmav_z \nonumber \\ &=&
933: \left[ \begin{array} {c c c} 0 & -J_{xy}^x & - J_{xz}^x \\
934: -J_{xy}^x & 0 & D_x^x \\
935: -J_{xz}^x & - D_x^x & 0 \\ \end{array} \right] \ ,
936: \end{eqnarray}
937: \begin{eqnarray}
938: {\partial U_{\alpha \beta} (2,3) \over \partial Q_{y_p}} &= &
939: - \sigmav_y \sigmav_z {\bf U}^{y_p} \sigmav_y \sigmav_z \nonumber \\ &=&
940: \left[ \begin{array} {c c c}
941: -J_{xx}^y & D_z^y & -D_y^y \\ -D_z^y & -J_{yy}^y &
942: -J_{yz}^y \\ D_y^y & -J_{yz}^y & -J_{zz}^y \\ \end{array} \right] \ ,
943: \end{eqnarray}
944: \begin{eqnarray}
945: {\partial U_{\alpha \beta} (2,3) \over \partial Q_{z_p}} &= &
946: - \sigmav_y \sigmav_z {\bf U}^{z_p} \sigmav_y \sigmav_z \nonumber \\ &=&
947: \left[ \begin{array} {c c c}
948: -J_{xx}^z & D_z^z & -D_y^z \\ -D_z^z & -J_{yy}^z & -J_{yz}^z \\
949: D_y^z & -J_{yz}^z & -J_{zz} \\ \end{array} \right] \ .
950: \end{eqnarray}
951: where we used Eq. (\ref{SIGEQ2}).
952:
953: We obtain the \#4-\#1' interactions by applying the glide operation $m_y$
954: to the \#2-\#3 interaction, so that
955: \begin{eqnarray}
956: {\partial U_{\alpha \beta} (4,1') \over \partial Q_{x_p}} &= &
957: \sigmav_y {\partial {\bf U}(2,3) \over \partial Q_{x_p}} \sigmav_y
958: \nonumber \\ &=&
959: \left[ \begin{array} {c c c} 0 & J_{xy}^x & - J_{xz}^x \\
960: J_{xy}^x & 0 & -D_x^x \\ -J_{xz}^x & D_x^x & 0 \\ \end{array} \right] \ ,
961: \end{eqnarray}
962: \begin{eqnarray}
963: {\partial U_{\alpha \beta} (4,1') \over \partial Q_{y_p}} &= &
964: - \sigmav_y {\partial {\bf U}(2,3) \over \partial Q_{y_p}} \sigmav_y
965: \nonumber \\ &=&
966: \left[ \begin{array} {c c c}
967: J_{xx}^y & D_z^y & D_y^y \\ -D_z^y & J_{yy}^y &
968: -J_{yz}^y \\ -D_y^y & -J_{yz}^y & J_{zz}^y \\ \end{array} \right] \ ,
969: \end{eqnarray}
970: \begin{eqnarray}
971: {\partial U_{\alpha \beta} (4,1') \over \partial Q_{z_p}} &= &
972: \sigmav_y {\partial {\bf U}(2,3) \over \partial Q_{z_p}} \sigmav_y
973: \nonumber \\ &=&
974: \left[ \begin{array} {c c c}
975: -J_{xx}^z & -D_z^z & -D_y^z \\ D_z^z & -J_{yy}^z & J_{yz}^z \\
976: D_y^z & J_{yz}^z & -J_{zz}^z \\ \end{array} \right]\ ,
977: \end{eqnarray}
978: and finally we get the \#3-\#2' interaction by applying a two-fold
979: rotation about the $x$-axis to the \#4-\#1' interaction to get
980: \begin{eqnarray}
981: {\partial U_{\alpha \beta} (3,2') \over \partial Q_{x_p}} &= &
982: \sigmav_y \sigmav_z {\partial {\bf U} (4,1') \over \partial Q_{x_p}}
983: \sigmav_y \sigmav_z \nonumber \\ &=&
984: \left[ \begin{array} {c c c} 0 & -J_{xy}^x & J_{xz}^x \\ -J_{xy}^x & 0 & -D_x^x \\
985: J_{xz}^x & D_x^x & 0 \\ \end{array} \right] \ ,
986: \end{eqnarray}
987: \begin{eqnarray}
988: {\partial U_{\alpha \beta} (3,2') \over \partial Q_{y_p}} &= &
989: - \sigmav_y \sigmav_z {\partial {\bf U} (4,1') \over \partial Q_{y_p}}
990: \sigmav_y \sigmav_z \nonumber \\ &=&
991: \left[ \begin{array}{c c c}
992: -J_{xx}^y & D_z^y & D_y^y \\ -D_z^y & -J_{yy}^y &
993: J_{yz}^y \\ -D_y^y & J_{yz}^y & -J_{zz}^y \\ \end{array} \right]\ ,
994: \end{eqnarray}
995: and
996: \begin{eqnarray}
997: {\partial U_{\alpha \beta} (3,2') \over \partial Q_{z_p}} &= &
998: - \sigmav_y \sigmav_z {\partial {\bf U} (4,1') \over \partial Q_{z_p}}
999: \sigmav_y \sigmav_z \nonumber \\ &=&
1000: \left[ \begin{array}{c c c}
1001: J_{xx}^z & -D_z^z & -D_y^z \\ D_z^z & J_{yy}^z &
1002: -J_{yz}^z \\ D_y^z & -J_{yz}^z & J_{zz}^z \\ \end{array} \right]\ .
1003: \end{eqnarray}
1004:
1005: \section{Mean Field Spin-Phonon Hamiltonian}
1006:
1007: \subsection{Mean Field Results}
1008:
1009: Here we treat the nn spine-spine interactions in detail. Analogous
1010: calculations for the nnn spine-spine and nn spine-cross tie
1011: interactions are treated in an Appendix.
1012: We evaluate the spin-phonon Hamiltonian ${\cal H}_{\gamma_p}$
1013: of Eq. (\ref{FEINT}) at the mean-field level. In other words,
1014: for the spin operators we simply substitute their average
1015: values as given in Eqs. (\ref{SPINS1})-(\ref{SPINS2}).
1016: One sees that ${\cal H}_{x_p}$ for nn spine-spine interactions, for
1017: instance, will consist of
1018: contributions proportional to $J_{xy}^{x_p}$, to
1019: $J_{xz}^{x_p}$, and to $D_x^{x_p}$. To illustrate the
1020: calculation we will explicitly evaluate the first of these,
1021: which we denote ${\cal H}_{x_p}(J_{xy})$:
1022: \begin{eqnarray}
1023: {\cal H}_{x_p}(J_{xy})&=& Q_{x_p} J_{xy}^{x_p}
1024: \sum_{uc} \left[ S_x^{(1)} S_y^{(4)} + S_y^{(1)} S_x^{(4)} \right.
1025: \nonumber \\ && \ - S_x^{(2)} S_y^{(3)} - S_x^{(3)} S_y^{(2)}
1026: + S_x^{(4)} S_y^{(1')} \nonumber \\ && \ \left.
1027: + S_x^{(1')} S_y^{(4)} - S_x^{(3)} S_y^{(2')}
1028: - S_x^{(2')} S_y^{(3)} \right] \ ,
1029: \end{eqnarray}
1030: where, since we included all interactions within a single unit
1031: cell, the sum is over all $N_{\rm u c}$ unit cells. (In this
1032: summation only terms involving both ${\bf q}$ and $-{\bf q}$
1033: survive.) Thus
1034: \begin{eqnarray}
1035: {\cal H}_{x_p}(J_{xy})&=&
1036: 2 N_{\rm u c} Q_{x_p} J_{xy}^{x_p} e^{-iqa/2} \nonumber \\
1037: && \ \times \bigl[
1038: (a_{s,x} + i b_{s,x})(ia_{s,y}^* - b_{s,y}^*)
1039: \nonumber \\ && \ +
1040: (ia_{s,y} + b_{s,y})(-a_{s,x}^* + ib_{s,x}^*)
1041: \nonumber \\ && \
1042: - (-a_{s,x} + i b_{s,x})(ia_{s,y}^* + b_{s,y}^*)
1043: \nonumber \\ && \
1044: - (ia_{s,y} - b_{s,y})(a_{s,x}^* + ib_{s,x}^*) \bigr] + {\rm c. \ c.}
1045: \nonumber \\ &=&
1046: 16 N_{\rm u c} Q_{x_p} J_{xy}^{x_p} \cos (qa/2) \nonumber \\
1047: && \ \times \Im
1048: [a_{s,x}^* a_{s,y} + b_{s,x} b_{s,y}^*] \ .
1049: \end{eqnarray}
1050: The other terms proportional to $Q_{x_p}$ are
1051: \begin{eqnarray}
1052: {\cal H}_{x_p}(J_{xz})&=& 16 N_{\rm u c} Q_{x_p} J_{xz}^{x_p} \sin (qa/2)
1053: \nonumber \\ && \ \times
1054: \Im [a_{s,x} a_{s,z}^* + b_{s,x} b_{s,z}^*] \ .
1055: \end{eqnarray}
1056: \begin{eqnarray}
1057: {\cal H}_{x_p}(D_x)&=& 16 N_{\rm u c} Q_{x_p} D_x^{x_p} \cos (qa/2)
1058: \nonumber \\ && \ \times
1059: \Im [-a_{s,y} a_{s,z}^* + b_{s,y} b_{s,z}^*] \ .
1060: \end{eqnarray}
1061: In view of Eq. (\ref{PHASEEQ}) all these terms involving $Q_{x_p}$
1062: vanish, as was found from the phenomenological formulation.
1063: Similarly, all the terms involving $Q_{z_p}$ also vanish, again
1064: in conformity with the phenomenological argument.
1065:
1066: We are thus only left with terms involving $Q_{y_p}$. In Appendix
1067: C we find that the strain dependence of the nn spine interactions give
1068: \begin{eqnarray}
1069: {\cal H}_{y_p} &=& 16 N_{uc} Q_{y_p} \sum_{\mu , \nu = x,y,z}
1070: \Lambda_{\mu \nu}^{(nn)} \ \Im \left[ a_{s,\mu}^* b_{s,\nu} \right] \ ,
1071: \label{YPEQ} \end{eqnarray}
1072: where
1073: \begin{eqnarray}
1074: \Lambdav^{(nn)} &=& \left[ \begin{array} {c c c}
1075: J_{xx}^{y_p} c & D_z^{y_p} s & D_y^{y_p}c \\
1076: - D_z^{y_p} s & -J_{yy}^{y_p} c &
1077: - J_{yz}^{y_p} s \\ D_y^{y_p} c &
1078: J_{yz}^{y_p} s& - J_{zz}^{y_p} c \\
1079: \end{array} \right] \ ,
1080: \label{G1EQ} \end{eqnarray}
1081: where $c\equiv \cos (qa/2)$ and $s \equiv \sin (qa/2)$.
1082: Using the results of Appendix C we find that the nnn
1083: interactions give a result of the form of Eq. (\ref{YPEQ})
1084: but with
1085: \begin{eqnarray}
1086: \Lambdav^{(nnn)} &=& \left[ \begin{array} {c c c}
1087: - K_{xx}^{y_p} c' & - E_z^{y_p} s' & - K_{xz}^{y_p}c' \\
1088: E_z^{y_p} s' & K_{yy}^{y_p} c' &
1089: -E_x^{y_p} s' \\ -K_{xz}^{y_p} c' &
1090: E_x^{y_p} s' & -K_{zz}^{y_p} c' \\
1091: \end{array} \right] \ ,
1092: \label{YPEQ2} \end{eqnarray}
1093: where $c' \equiv \cos (qa)$ and $s' \equiv \sin (qa)$.
1094: Using the results of Appendix C we have the results for
1095: the spine-cross tie exchange gradients:
1096: \begin{eqnarray}
1097: V_{y_p}&=& 16 N_{\rm u c} Q_{y_p} \left(
1098: \sum_{\alpha=y,z} \sum_\beta \Lambda_{\alpha \beta}^{sx}
1099: \Im \left[ a_{c \alpha} b_{s \beta}^* \right] \right.
1100: \nonumber \\ && \ \left.
1101: + \sum_{\alpha=x} \sum_\beta \Lambda_{\alpha \beta}^{sx}
1102: \Im \left[ b_{c \alpha} a_{s, \beta}^* \right] \right) \ ,
1103: \label{YPEQ3} \end{eqnarray}
1104: where $\Lambdav^{sx}$ is
1105: \begin{eqnarray}
1106: \left[ \begin{array} {c c c}
1107: L_{xx}^{y_p} {s}^{\prime \prime} & \left[ L_{xy}^{y_p} + F_z^{y_p}
1108: \right] c'' &
1109: \left[ L_{xz}^{y_p}-F_y^{y_p} \right] {s}^{\prime \prime} \\
1110: \left[ L_{xy}^{y_p} - F_z^{y_p} \right] {c}^{\prime \prime} &
1111: L_{yy}^{y_p} {s}^{\prime \prime} &
1112: \left[ L_{yz}^{y_p} + F_x^{y_p} \right] {c}'' \\
1113: \left[ L_{xz}^{y_p}+F_y^{y_p} \right] {c}'' &
1114: \left[ L_{yz}^{y_p}-F_x^{y_p} \right] {s}'' & L_{zz}^{y_p} {c}'' \\
1115: \end{array} \right] \ ,
1116: \label{G3EQ} \end{eqnarray}
1117: where $c'' \equiv \cos (qa/4)$ and $s'' \equiv \sin (qa/4)$.
1118: These results agree with the phenomenological model, in that
1119: $V_{y_p}$
1120: is only nonzero when both the ``$a$" and the ``$b$" irreps are
1121: simultaneously present and they can not have the same phase
1122: (lest $a^*b$ be real).
1123:
1124: \subsection{Summary} \label{RESULTS}
1125:
1126: Here we show how the above results lead to an evaluation of the
1127: spontaneous polarization. If we combine the results of Eqs.
1128: (\ref{YPEQ}), (\ref{YPEQ2}), and (\ref{YPEQ3}), we may write
1129: \begin{eqnarray}
1130: V_{y_p} = N_{\rm uc} Q_{y_p} \lambda_{y_p} \ , \ \ \
1131: p=1,12 \ .
1132: \end{eqnarray}
1133: But the GD's are related to the normal modes via
1134: \begin{eqnarray}
1135: Q_{y_p} &=& \sum_{n=1,11} o_{y_p}^{(n)} M_p^{-1/2} Q_ n \ ,
1136: \end{eqnarray}
1137: where the orthogonal transformation ${\bf o}$ is essentially that
1138: given in Table \ref{b2u} apart from normalization factors due
1139: to replacing a Wyckoff orbit by one of its atoms. Note that the GD
1140: $y_p$ is associated with atoms all having the same mass $M_p$. The
1141: magnetoelectric trilinear interaction in terms of normal mode operators is
1142: thus
1143: \begin{eqnarray}
1144: V &=& N_{uc} \sum_{n,p} \lambda_{y_p} o_{y_p}^{(n)} M_p^{-1/2} Q_n \ ,
1145: \end{eqnarray}
1146: Since the unperturbed elastic energy is ${\cal H}_0 = (N_{\rm uc}/2)
1147: \sum_n \omega_n^2 Q_n^2$, we see that the trilinear interaction
1148: induces the displacements
1149: \begin{eqnarray}
1150: \langle Q_n \rangle = \sum_p
1151: \lambda_{y_p} o_{y_p}^{(n)} M_p^{-1/2} \omega_n^{-2} \ ,
1152: \end{eqnarray}
1153: from which we obtain the induced spontaneous polarization as
1154: \begin{eqnarray}
1155: P_y &=& {1 \over \Omega_{\rm uc}} \sum_{n, p}
1156: q_p o_{y_p}^{(n)} \langle Q_n \rangle M_p^{-1/2} \ ,
1157: \label{PEQ} \end{eqnarray}
1158: where the atomic charges $q_p$ are given in Table \ref{b2u}.
1159: The only ingredients we do not have for this evaluation are
1160: the various gradients of the exchange tensors.
1161:
1162: \section{Toy Models} \label{TOYSEC}
1163:
1164: In this section we will consider two toy models. In the first one, we consider a single spine
1165: with one Ni per unit cell, but with two oxygen atoms symmetrically placed on either side
1166: of the spine. In this version, all atoms lie in a single plane. In the second version,
1167: the size of the unit cell is doubled. In one plaquette the oxygen atoms are both displaced
1168: equally perpendicularly to the original atomic plane and in the next plaquette the oxygens
1169: are oppositely displaced. In this version one of the mirror planes becomes a glide plane.
1170: These models illustrate the simplifications which arise when the system has higher
1171: symmetry than the buckled kagom\'{e} lattice of NVO.
1172:
1173: \subsection{Unit cell with two atoms}
1174:
1175: In this section we consider the toy model shown in the left panel of Fig. \ref{TOY}.
1176: We first analyze the symmetry of the strain derivatives of the exchange
1177: tensor. By translational symmetry all nearest neighbor interactions are
1178: equivalent. So we set
1179: \begin{eqnarray}
1180: && {\partial J_{\alpha \beta} (n,n+1) \over \partial Q_{\gamma_p}} \equiv
1181: {\bf J}_{\alpha \beta}^{\gamma_p} \nonumber \\ && \ \ =
1182: \left[ \begin{array} {c c c} J_{xx}^\gamma & J_{xy}^\gamma + D_z^\gamma &
1183: J_{xz}^\gamma - D_y^\gamma \\ J_{xy}^\gamma - D_z^\gamma &
1184: J_{yy}^\gamma & J_{yz}^\gamma + D_x^\gamma \\ J_{xz}^\gamma
1185: + D_y^\gamma & J_{yz}^\gamma - D_x^\gamma & J_{zz}^\gamma \\
1186: \end{array} \right] \ ,
1187: \end{eqnarray}
1188: where $J_{\alpha \beta}^\gamma \equiv \partial J_{\alpha \beta} /
1189: \partial Q_{\gamma_p}$, $D_\alpha^\gamma \equiv \partial D_\alpha
1190: / \partial Q_{\gamma_p}$, and the index $p$ is left implicit. The
1191: Hamiltonian is invariant under mirror reflections with respect to
1192: each coordinate axis. Taking account of the symmetry of the displacement
1193: coordinate and the fact that $m_x$
1194: interchanges indices of the exchange tensor, we have that
1195: \begin{eqnarray}
1196: \sigmav_x {\bf J}^\gamma \sigmav_x &=&
1197: (1 - 2 \delta_{x, \gamma}) \tilde {\bf J}^\gamma \nonumber \\
1198: \sigmav_y {\bf J}^\gamma \sigmav_y &=&
1199: (1 - 2 \delta_{y, \gamma}) {\bf J}^\gamma \nonumber \\
1200: \sigmav_z {\bf J}^\gamma \sigmav_z &=&
1201: (1 - 2 \delta_{z, \gamma}) {\bf J}^\gamma \ .
1202: \label{SYMEQS} \end{eqnarray}
1203: \begin{figure}
1204: \includegraphics[width=8cm]{TOY.eps}
1205: \caption{\label{TOY} The two toy models. In each case the unit cell
1206: is bounded by a dashed line. The filled circles are Ni sites
1207: and the squares are O sites.}
1208: \end{figure}
1209:
1210: \noindent
1211: As a result of this symmetry we have that ${\bf J}^x = 0$,
1212: \begin{eqnarray}
1213: {\bf J}^y&=& \left[ \begin{array} {c c c}
1214: 0 & D_z^y & 0 \\ -D_z^y & 0 & 0 \\ 0 & 0 & 0 \\ \end{array}
1215: \right] \ , \nonumber \\
1216: {\bf J}^z&=& \left[ \begin{array} {c c c}
1217: 0 & 0 & -D_y^z \\ 0 & 0 & 0 \\ D_y^z & 0 & 0 \\ \end{array}
1218: \right] \ .
1219: \end{eqnarray}
1220: We construct the trilinear spin-phonon interaction, $V$, as
1221: \begin{eqnarray}
1222: && V = \sum_p \left( Q_{y_p} D_z^{y_p} C_y + Q_{z_p} D_y^{z_p} C_z
1223: \right) \ ,
1224: \end{eqnarray}
1225: where
1226: \begin{eqnarray}
1227: C_\alpha &=& \sum_n [ S_x(n) S_\alpha (n+1) - S_\alpha (n) S_x(n+1) ] .
1228: \end{eqnarray}
1229: Now we replace the spins by their equilibrium values. We note that
1230: each spin component belongs to a separate representation and we write
1231: \begin{eqnarray}
1232: S_x(n) = S_x(q) e^{inqa} + S_x(q)^* e^{-inqa} \ , \nonumber \\
1233: S_y(n) = S_y(q) e^{inqa} + S_y(q)^* e^{-inqa} \ , \nonumber \\
1234: S_z(n) = S_z(q) e^{inqa} + S_z(q)^* e^{-inqa} \ .
1235: \end{eqnarray}
1236: Then, keeping only those terms which survive the sum over $n$ we have
1237: \begin{eqnarray}
1238: V&=& 4N \sin (qa) \sum_p \left[ Q_{y_p} D_z^{y_p} r_x(q) r_y(q) \sin(\phi_x
1239: -\phi_y) \right. \nonumber \\ && \ \left.
1240: + Q_{z_p} D_y^{z_p} r_z(q) r_x(q) \sin(\phi_x - \phi_z) \right] \ ,
1241: \end{eqnarray}
1242: where $N$ is the total number of Ni spins and
1243: we set $S_\alpha (q) = r_\alpha (q) e^{i \phi_\alpha}$, where
1244: $r_\alpha (q)$ is real. As found before\cite{FERRO,EXPT,REV} this
1245: interaction is only nonzero when a) two different representations
1246: are condensed and b) the two representation have different phases $\phi$.
1247: In view of our previous results, it is clear that the appearance of
1248: only strain derivatives of the DM vector is an artifact of the
1249: rather high symmetry of this coplanar model.
1250:
1251: \subsection{Unit cell with four atoms}
1252:
1253: Now we consider the noncoplanar toy model shown in the right panel of
1254: Fig. \ref{TOY}. The first two symmetry relations of Eq. (\ref{SYMEQS})
1255: remain valid, but the third one now results from the glide plane which
1256: involves a translation along the chain. If ${\bf J}^\alpha_-$ denotes
1257: the strain derivative of the exchange tensor for coupling sites $2n$ and
1258: $2n+1$, and ${\bf J}_+^\alpha$ that for sites $2n+1$ and $2n+2$, then we have
1259: \begin{eqnarray}
1260: {\bf J}_\pm^x &=& \left[ \begin{array} {c c c} 0 & 0 & \pm J_{xz}^x \\
1261: 0 & 0 & 0 \\ \pm J_{xz}^x & 0 & 0 \\ \end{array} \right] \ ,
1262: \end{eqnarray}
1263: \begin{eqnarray}
1264: {\bf J}_\pm^y &=& \left[ \begin{array} {c c c} 0 & D_z^y & 0 \\
1265: -D_z^y & 0 & \pm J_{yz}^y \\ 0 & \pm J_{yz}^y & 0 \\ \end{array} \right] \ , \end{eqnarray}
1266: \begin{eqnarray}
1267: {\bf J}_\pm^z &=& \left[ \begin{array} {c c c} \pm J_{xx}^z & 0 & -D_y^z \\
1268: 0 & \pm J_{yy}^z & 0 \\ D_y^z & 0 & \pm J_{zz}^z \\ \end{array} \right] \ .
1269: \end{eqnarray}
1270:
1271: \begin{table}
1272: \caption{\label{BASIS}Basis spin functions for sites \#1 and \#2 in the unit cell
1273: in terms of the complex-valued Fourier components $S_\alpha (q)$ for irreps
1274: characterized by the eigenvalues of $m_y$ and the glide operation $m_z$.}
1275:
1276: \vspace{0.2 in}
1277: \begin{tabular} {||c | c |c || c |c||}
1278: \hline $\Gamma$ & $m_y$ & $m_z$ & ${\bf S}$(\#1) & ${\bf S}$(\#2) \\ \hline
1279: $\Gamma_1$ & $+$ & $+$ & $(S_x(q),0,S_z(q))$ & $(S_x(q),0,-S_z(q))$ \\
1280: $\Gamma_2$ & $+$ & $-$ & $(S_x(q),0,S_z(q))$ & $(-S_x(q),0,-S_z(q))$ \\
1281: $\Gamma_3$ & $-$ & $+$ & $(0,S_y(q),0)$ & $(0,S_y(q),0)$ \\
1282: $\Gamma_4$ & $-$ & $-$ & $(0,S_y(q),0)$ & $(0,-S_y(q),0)$ \\ \hline
1283: \end{tabular}
1284: \end{table}
1285:
1286: \begin{figure}
1287: \includegraphics[width=8cm]{TOYPH.eps}
1288: \caption{\label{TOYPH} Generalized displacements which transform like vectors.
1289: Open circles (circles with inscribed "x") are displacements out of (into) the page.
1290: The sites at positive or negative $z$ are indicated. The upper panels show
1291: modes in which the atoms move only in the $z$-direction.
1292: Upper left: an $x$-like mode $x_3$. Upper right: a $y$-like mode $y_3$.
1293: The lower panels show $z$-like modes. Left: $z_3$ with motion only along the
1294: $x$-axis and right: $z_4$ with motion only along the $y$-axis.}
1295: \end{figure}
1296:
1297: Next we characterize the spin structure. There are four irreps for which
1298: the basis functions are listed in Table \ref{BASIS}. Thus we write
1299: for spins \#1 (${\bf S}_1)$ and \#2 $({\bf S}_2)$ in the unit cell,
1300: \begin{eqnarray}
1301: S_{1,x}(X) &=& [S_x^{(1)}(q) + S_x^{(2)}(q)] e^{iqX} \nonumber \\
1302: && \ +
1303: [S_x^{(1)}(q)^* + S_x^{(2)}(q)^*] e^{-iqX} \ , \\
1304: S_{1,y}(X) &=& [S_y^{(3)}(q) + S_y^{(4)}(q)] e^{iqX} \nonumber \\
1305: && \ + [S_y^{(3)}(q)^* + S_y^{(4)}(q)^*] e^{-iqX} \ , \\
1306: S_{1,z}(X) &=& [S_z^{(1)}(q) + S_z^{(2)}(q)] e^{iqX} \nonumber \\
1307: && \ + [S_z^{(1)}(q)^* + S_z^{(2)}(q)^*] e^{-iqX} \ , \\
1308: S_{2,x}(X) &=& [S_x^{(1)}(q) - S_x^{(2)}(q)] e^{iqX} \nonumber \\
1309: && \ + [S_x^{(1)}(q)^* - S_x^{(2)}(q)^*] e^{-iqX} \ , \\
1310: S_{2,y}(X) &=& [S_y^{(3)}(q) - S_y^{(4)}(q)] e^{iqX} \nonumber \\
1311: && \ + [S_y^{(3)}(q)^* - S_y^{(4)}(q)^*] e^{-iqX} \ , \\
1312: S_{2,z}(X) &=& [-S_z^{(1)}(q) + S_z^{(2)}(q)] e^{iqX} \nonumber \\
1313: && \ + [-S_z^{(1)}(q)^* + S_z^{(2)}(q)^*] e^{-iqX} \ ,
1314: \label{SPINOPS} \end{eqnarray}
1315: where the superscript labels the irrep as in Table \ref{BASIS}).
1316: Thereby we find the trilinear spin-phonon coupling
1317: (when the spin operators are replaced by their values:
1318: \begin{eqnarray}
1319: V &=& 4N \sin(qa) \sum_p \biggl[ U Q_{x_p} J_{xz}^{x_p}
1320: + Q_{y_p} (V J_{yz}^{y_p} + W D_z^{y_p} ) \nonumber \\
1321: && \ + Q_{z_p} \biggl( X D_y^{z_p} + \sum_{\alpha =x,y,z}
1322: Y_\alpha J_{\alpha \alpha}^{z_p} \biggr) \biggr] \ ,
1323: \end{eqnarray}
1324: where
1325: \begin{eqnarray}
1326: U &=& \Im \left( S_x^{(1)}(q)^* S_z^{(1)}(q) + S_x^{(2)}(q) S_z^{(2)}(q)^* \right)
1327: \end{eqnarray}
1328: \begin{eqnarray}
1329: V &=& \Im \left( S_y^{(4)}(q) S_z^{(2)}(q)^* + S_z^{(1)}(q) S_y^{(3)}(q)^* \right)
1330: \end{eqnarray}
1331: \begin{eqnarray}
1332: W &=& \Im \left( S_x^{(1)}(q) S_y^{(3)}(q)^* + S_x^{(2)}(q)^* S_y^{(4)}(q) \right)
1333: \end{eqnarray}
1334: \begin{eqnarray}
1335: X &=& \Im \left( S_x^{(1)}(q)^* S_z^{(2)}(q) + S_x^{(2)}(q) S_z^{(1)}(q)^* \right)
1336: \end{eqnarray}
1337: \begin{eqnarray}
1338: Y_x &=& \Im \left( S_x^{(1)}(q)^* S_x^{(2)}(q) \right)
1339: \end{eqnarray}
1340: \begin{eqnarray}
1341: Y_y &=& \Im \left( S_y^{(3)}(q)^* S_y^{(4)}(q) \right)
1342: \end{eqnarray}
1343: \begin{eqnarray}
1344: Y_z &=& \Im \left( S_z^{(1)}(q) S_z^{(2)}(q)^* \right)
1345: \end{eqnarray}
1346: The general symmetry arguments indicate that there can not be a polarization
1347: along $\hat i$. We see that $U$ vanishes because all the components within
1348: a single representation have the same phase, so that, for instance,
1349: $S_x^{(1)}(q)S_z^{(1)}(q)^*$ is real. Here we see that, depending on the
1350: spin structure the spontaneous polarization can either be along
1351: $\hat j$ (if either both irreps \#1 and \#3 are active or both irreps
1352: \#2 and \#4 are active) or along $\hat k$ (if either both irreps \#1 and \#2
1353: are active or both irreps \#3 and \#4 are active.) These results are
1354: exactly what the phenomenological analysis would give.
1355:
1356: \section{Conclusions}
1357: %TYTYTY BEGIN
1358: In this paper we present neutron scattering measurements of phonons in NVO,
1359: the first-principles computation of the zone-center phonons and their symmetry
1360: analysis. We identified two particularly interesting phonons among the twelve
1361: B$_{2u}$ modes which have the right symmetry to induce the experimentally
1362: observed dipole moment along the ${\bf b}$-axis in NVO. Using the calculated atomic
1363: charges and the eigenvectors we conclude that the required distortion to
1364: induce the observed dipole moment is small (~0.001 \AA)
1365: and would be difficult to observe directly by neutron powder diffraction.
1366: Finally, we present
1367: %TYTYTY END
1368: a symmetry analysis to characterize the microscopic
1369: magnetoelectric coupling in Ni$_3$V$_2$O$_8$. The method can easily
1370: be applied to similar systems such as TbMnO$_3$. The result of
1371: this analysis is that we can specify those strain derivatives of
1372: the exchange tensor which should now be the targets of more
1373: fundamental quantum calculations, perhaps based on the LDA\cite{LDA}
1374: or similar schemes. (These results are given in Eqs. (\ref{G1EQ})-(\ref{G3EQ}),
1375: where the superscript indicates the component of the gradient.)
1376: In Subsec. \ref{RESULTS} we show how these gradients lead to
1377: an evaluation of the spontaneous polarization from first principles.
1378: In Sec. \ref{TOYSEC} we also studied some structurally simpler
1379: toy models. A general conclusion is that the local site symmetry
1380: in NVO is low enough that almost all strain derivatives of the
1381: exchange tensor are involved. It is true that the Dzyaloshinskii-Moriya
1382: interaction must be present (and, of course, it is allowed in such
1383: low symmetry systems). In the presence of such interactions ferroelectricity
1384: results from the strain dependence of all the components of the
1385: exchange tensor, even that of the isotropic exchange interaction.
1386:
1387:
1388: \begin{acknowledgments}
1389: This work was supported in part by the Israel US Binational Science
1390: Foundation under grant number 2000073.
1391: %TYTYTY BEGIN
1392: We thank C. Broholm for providing us
1393: the NVO powder samples used in this study.
1394: %TYTYTY END
1395: \end{acknowledgments}
1396:
1397: \begin{appendix}
1398: \section{Next-nearest neighbor spine interactions}
1399:
1400: In this section we consider next-nearest neighbor (nnn) interactions
1401: between spins on a given spine. Since only the gradients with respect
1402: to $Q_{y_p}$ are needed, we only consider those here. We set
1403: \begin{eqnarray}
1404: {\partial J_{\alpha \beta} (1,1') \over \partial Q_{y_p}}
1405: &=& V_{\alpha \beta}^{y_p} \ .
1406: \end{eqnarray}
1407: The operation $2_y$ takes this bond into itself with reversed indices.
1408: So, taking account of the transformation properties of $Q_{y_p}$,
1409: we require that
1410: \begin{eqnarray} {\bf 2}_y {\bf V}^{y_p} &=& \tilde {\bf V}^{y_p} \ ,
1411: \end{eqnarray}
1412: where ${\bf 2}_y$ is the two-fold rotation operator. In terms of matrices,
1413: this relation is
1414: \begin{eqnarray}
1415: 2_y {\bf V}^{y_p} 2_y &=& \tilde {\bf V}^{y_p} \ ,
1416: \end{eqnarray}
1417: where $2_y = \sigmav_x \sigmav_z$. Thus
1418: \begin{eqnarray}
1419: {\bf V}^{y_p} &=& \left[ \begin{array} {c c c}
1420: K_{xx}^y & E_z^y & K_{xz}^y \\ - E_z^y & K_{yy}^y & E_x^y \\
1421: K_{xz}^y & - E_x^y & K_{zz}^y \\ \end{array} \right] \ .
1422: \end{eqnarray}
1423: We have that
1424: \begin{eqnarray}
1425: {\partial J_{\alpha \beta} (2,2') \over \partial Q_{y_p}} &=&
1426: - 2_x {\bf V}^{y_p} 2_x \ ,
1427: \end{eqnarray}
1428: so that
1429: \begin{eqnarray}
1430: {\partial J_{\alpha \beta} (2,2') \over \partial Q_{y_p}} &=&
1431: \left[ \begin{array} {c c c}
1432: -K_{xx}^y & E_z^y & K_{xz}^y \\ - E_z^y & -K_{yy}^y & -E_x^y \\
1433: K_{xz} & E_x^y & -K_{zz}^y \\
1434: \end{array} \right] \ .
1435: \end{eqnarray}
1436: We have that
1437: \begin{eqnarray}
1438: {\partial J_{\alpha \beta} (3',3) \over \partial Q_{y_p}} &=&
1439: - \sigmav_y {\bf V}^{\gamma_p} \sigmav_y \ ,
1440: \end{eqnarray}
1441: so that
1442: \begin{eqnarray}
1443: {\partial J_{\alpha \beta} (3',3) \over \partial Q_{y_p}} &=&
1444: \left[ \begin{array} {c c c}
1445: -K_{xx}^y & E_z^y & - K_{xz}^y \\ -E_z^y & -K_{yy}^y & E_x^y \\ -K_{xz}^y & -E_x^y & -K_{zz}^y \\
1446: \end{array} \right] \ .
1447: \end{eqnarray}
1448: We have that
1449: \begin{eqnarray}
1450: {\partial J_{\alpha \beta} (4',4) \over \partial Q_{y_p}} &=&
1451: \sigmav_x \tilde {\bf V}^{\gamma_p} \sigmav_x \ ,
1452: \end{eqnarray}
1453: so that
1454: \begin{eqnarray}
1455: {\partial J_{\alpha \beta} (4',4) \over \partial Q_{y_p}} &=&
1456: \left[ \begin{array} {c c c}
1457: K_{xx}^y & E_z^y & - K_{xz}^y \\ - E_z^y & K_{yy}^y & - E_x^y \\
1458: -K_{xz}^y & E_x^y & K_{zz}^y \\
1459: \end{array} \right] \ .
1460: \end{eqnarray}
1461:
1462: \section{Spine cross-tie interactions}
1463:
1464: In this section we analyze the spine-cross tie interactions,
1465: in Fig. \ref{SPCR}, again keeping only derivative with respect to
1466: $Q_{y_p}$. We set the interaction of type A to be
1467: \begin{eqnarray}
1468: {\partial J_{\alpha \beta} (c,1) \over dQ_{y_p}} &=&
1469: W_{\alpha \beta}^{y_p} \ ,
1470: \end{eqnarray}
1471: so that (when the index $p$ is suppressed)
1472: \begin{eqnarray}
1473: {\bf W}^y &=& \left[ \begin{array} {c c c}
1474: L_{xx}^y & L_{xy}^y + F_z^y & L_{xz}^y - F_y^y \\
1475: L_{xy}^y - F_z^y & L_{yy}^y & L_{yz}^y + F_x^y \\
1476: L_{xz}^y + F_y^y & L_{yz}^y - F_x^y & L_{zz}^y \\
1477: \end{array} \right] \ .
1478: \end{eqnarray}
1479: \begin{figure}
1480: \includegraphics[height=5.5cm]{SPCR.eps}
1481: \caption{\label{SPCR} As Fig. \protect{\ref{SPSP}}. Here we show
1482: the eight different nearest-neighbor spine-cross tie interactions,
1483: which we label A ... H. The arrow points from the first site index
1484: to the second site index. We also give the symmetry operation one
1485: has to apply to interaction A to get each of the other interactions.}
1486: \end{figure}
1487:
1488: \noindent
1489: We have that
1490: \begin{eqnarray}
1491: {\partial {\bf J} (c,4) \over \partial Q_{y_p}} &=&
1492: \sigmav_x {\bf W}^{\gamma_p} \sigmav_x \ ,
1493: \end{eqnarray}
1494: so that
1495: \begin{eqnarray}
1496: {\partial {\bf J} (c,4) \over \partial Q_{y_p}} &=&
1497: \left[ \begin{array} {c c c}
1498: L_{xx}^y & - L_{xy}^y - F_z^y & - L_{xz}^y + F_y^y \\
1499: - L_{xy}^y + F_z^y & L_{yy}^y & L_{yz}^y + F_x^y \\
1500: - L_{xz}^y - F_y^y & L_{yz}^y - F_x^y & L_{zz}^y \\
1501: \end{array} \right] \ . \nonumber \\
1502: \end{eqnarray}
1503: We have that
1504: \begin{eqnarray}
1505: {\partial {\bf J} (c,2) \over \partial Q_{y_p}} &=&
1506: - 2_x {\bf W}^{y_p} 2_x \ ,
1507: \end{eqnarray}
1508: so that
1509: \begin{eqnarray}
1510: {\partial {\bf J} (c,2) \over \partial Q_{y_p}} &=&
1511: \left[ \begin{array} {c c c}
1512: - L_{xx}^y & L_{xy}^y + F_z^y & L_{xz}^y - F_y^u \\
1513: L_{xy}^y - F_z^y & - L_{yy}^y & - L_{yz}^y - F_x^y \\
1514: L_{xz}^y + F_y^y & -L_{yz}^y + F_x^y & - L_{zz}^y \\
1515: \end{array} \right] \ . \nonumber \\
1516: \end{eqnarray}
1517: We have that
1518: \begin{eqnarray}
1519: {\partial {\bf J} (c,3) \over \partial Q_{y_p}} &=&
1520: - {\cal I} {\bf W}^{\gamma_p} {\cal I} \ ,
1521: \end{eqnarray}
1522: so that
1523: \begin{eqnarray}
1524: {\partial {\bf J} (c,3) \over \partial Q_{y_p}} &=&
1525: \left[ \begin{array} {c c c}
1526: - L_{xx}^y & -L_{xy}^y - F_z^y & -L_{xz}^y + F_y^y \\
1527: - L_{xy}^y + F_z^y & - L_{yy}^y & - L_{yz}^y - F_x^y \\
1528: - L_{xz}^y - F_y^y & - L_{yz}^y + F_x^y & - L_{zz}^y \\
1529: \end{array} \right] \ . \nonumber \\
1530: \end{eqnarray}
1531: We have that
1532: \begin{eqnarray}
1533: {\partial {\bf J} (c',1) \over \partial Q_{y_p}} &=&
1534: 2_y {\bf W} 2_y \ ,
1535: \end{eqnarray}
1536: so that
1537: \begin{eqnarray}
1538: {\partial {\bf J} (c',1) \over \partial Q_{y_p}} &=&
1539: \left[ \begin{array} {c c c}
1540: L_{xx}^y & -L_{xy}^y - F_z^y & L_{xz}^y - F_y^y \\
1541: -L_{xy}^y + F_z^y & L_{yy}^y & -L_{yz}^y - F_x^y \\
1542: L_{xz}^y + F_y^y & -L_{yz}^y + F_x^y & L_{zz}^y \\
1543: \end{array} \right] \ . \nonumber \\
1544: \end{eqnarray}
1545: We have that
1546: \begin{eqnarray}
1547: {\partial {\bf J} (c',4) \over \partial Q_{y_p}} &=&
1548: \sigmav_z {\bf W}^{\gamma_p} \sigmav_z \ ,
1549: \end{eqnarray}
1550: so that
1551: \begin{eqnarray}
1552: {\partial {\bf J} (c',4) \over \partial Q_{y_p}} &=&
1553: \left[ \begin{array} {c c c}
1554: L_{xx}^y & L_{xy}^y + F_z^y & - L_{xz}^y+ F_y^y \\
1555: L_{xy}^y - F_z^y & L_{yy}^y & - L_{yz}^y - F_x^y \\
1556: -L_{xz}^y - F_y^y & - L_{yz}^y + F_x^y & L_{zz}^y \\
1557: \end{array} \right] \ . \nonumber \\
1558: \end{eqnarray}
1559: We have that
1560: \begin{eqnarray}
1561: {\partial {\bf J} (c',2) \over \partial Q_{y_p}} &=&
1562: - 2_z {\bf W}^{\gamma_p} 2_z \ ,
1563: \end{eqnarray}
1564: so that
1565: \begin{eqnarray}
1566: {\partial {\bf J} (c',2) \over \partial Q_{y_p}} &=&
1567: \left[ \begin{array} {c c c}
1568: - L_{xx}^y & - L_{xy}^y - F_z^y & L_{xz}^y - F_y^y \\
1569: - L_{xy}^y + F_z^y & - L_{yy}^y & L_{yz}^y + F_x^y \\
1570: L_{xz}^y + F_y^y & L_{yz}^y - F_x^y & - L_{zz}^y \\
1571: \end{array} \right] \ . \nonumber \\
1572: \end{eqnarray}
1573: We have that
1574: \begin{eqnarray}
1575: {\partial {\bf J} (c',3) \over \partial Q_{y_p}} &=&
1576: - \sigmav_y {\bf W}^{\gamma_p} \sigmav_y \ ,
1577: \end{eqnarray}
1578: so that
1579: \begin{eqnarray}
1580: {\partial {\bf J} (c',3) \over \partial Q_{y_p}} &=&
1581: \left[ \begin{array} {c c c}
1582: - L_{xx}^y & L_{xy}^y + F_z^y & - L_{xz}^y + F_y^y \\
1583: L_{xy}^y - F_z^y & - L_{yy}^y & L_{yz}^y + F_x^y \\
1584: - L_{xz}^y - F_y^y & L_{yz}^y - F_x^y & - L_{zz}^y \\
1585: \end{array} \right] \ . \nonumber \\
1586: \end{eqnarray}
1587:
1588: \section{Spin-Phonon Interaction}
1589:
1590: \subsection{nn spine interactions}
1591:
1592: Now we collect terms involving $J_{xx}^y$:
1593: \begin{eqnarray}
1594: V_y(J_{xx}^y) &=& Q_yJ_{xx}^y \sum_{uc} \biggl[
1595: S_x(1) S_x(4) - S_x(2) S_x(3) \nonumber \\ && \
1596: + S_x(4) S_x(1') - S_x(3) S_x(2') \biggr]
1597: \nonumber \\ &=& Q_y J_{xx}^y \sum_{uc} e^{iqx_{14}} \biggl[
1598: (a_x+ib_x)(-a_x^*+ib_x^*) \nonumber \\ && \
1599: - (-a_x+ib_x)(a_x^*+ib_x^*) \nonumber \\ && \
1600: + (-a_x-ib_x)(a_x^*-ib_x^*) \nonumber \\ && \
1601: - (a_x-ib_x)(-a_x^*-ib_x^*) \biggr] + {\rm c. \ c.}
1602: \nonumber \\ &=& 4Q_y J_{xx}^y N_{uc} \left( e^{iqx_{14}} [
1603: ia_xb_x^* -i b_xa_x^*] + {\rm c. \ c.} \right)
1604: \nonumber \\ &=&
1605: - 16 N_{uc} Q_y J_{xx}^y \cos(qa/2) \Im [ a_x b_x^* ] \ .
1606: \end{eqnarray}
1607:
1608: Now we collect terms involving $J_{yy}^y$:
1609: \begin{eqnarray}
1610: V_y(J_{yy}^y) &=& Q_yJ_{yy}^y \sum_{uc} \biggl[
1611: S_y(1) S_y(4) - S_y(2) S_y(3) \nonumber \\ && \
1612: + S_y(4) S_y(1') - S_y(3) S_y(2') \biggr]
1613: \nonumber \\ &=& Q_y J_{yy}^y \sum_{uc} e^{iqx_{14}} \biggl[
1614: (ia_y+b_y)(ia_y^*-b_y^*) \nonumber \\ && \
1615: - (ia_y-b_y)(ia_y^*+b_y^*) \nonumber \\ && \
1616: + (-ia_y-b_y)(-ia_y^*+b_y^*) \nonumber \\ && \
1617: - (-ia_y+b_y)(-ia_y^*-b_y^*) \biggr]
1618: + {\rm c. \ c.} \nonumber \\ &=& 4Q_y J_{yy}^y N_{uc} \left( e^{iqx_{14}} [
1619: -ia_yb_y^* + i b_ya_y^*] + {\rm c. \ c.} \right) \nonumber
1620: \\ &=& 16 N_{uc} Q_y J_{yy}^y \cos(qa/2) \Im [a_yb_y^*] \ .
1621: \end{eqnarray}
1622:
1623: Now we collect terms involving $J_{zz}^y$:
1624: \begin{eqnarray}
1625: V_y (J_{zz}^y) &=& Q_yJ_{zz}^y \sum_{uc} \biggl[
1626: S_z(1) S_z(4) - S_z(2) S_z(3) \nonumber \\ && \
1627: + S_z(4) S_z(1') - S_z(3) S_z(2') \biggr]
1628: \nonumber \\ &=& Q_y J_{zz}^y \sum_{uc} e^{iqx_{14}} \biggl[
1629: (a_z+ib_z)(a_z^*-ib_z^*) \nonumber \\ && \
1630: - (a_z-ib_z)(a_z^*+ib_z^*) \nonumber \\ && \
1631: + (a_z+ib_z)(a_z^*-ib_z^*) \nonumber \\ && \
1632: - (a_z-ib_z)(a_z^*+ib_z^*) \biggr] + {\rm c. \ c.}
1633: \nonumber \\ &=& 4Q_y J_{zz}^y N_{uc} \left( e^{iqx_{14}} [
1634: -ia_zb_z^* + i b_za_z^*] + {\rm c. \ c.} \right) \nonumber
1635: \\ &=& 16 N_{uc} Q_y J_{zz}^y \cos (qa/2) \Im [a_z b_z^* ] .
1636: \end{eqnarray}
1637:
1638: Now we collect terms involving $J_{yz}^y$:
1639: \begin{eqnarray}
1640: V_y(J_{yz}^y) &=& Q_yJ_{yz}^y \sum_{uc} \biggl[
1641: S_y(1) S_z(4) + S_z(1) S_y(4) \nonumber \\ && \
1642: - S_z(2) S_y(3) - S_y(2) S_z(3) \nonumber \\
1643: && \ - S_y(4) S_z(1') - S_z(4) S_y(1') \nonumber \\ && \
1644: + S_y(3) S_z(2') + S_z(3) S_y(2')
1645: \biggr] \nonumber \\ &=& Q_y J_{yz}^y \sum_{uc} e^{iqx_{14}} \biggl[
1646: (ia_y+b_y)(a_z^*-ib_z^*) \nonumber \\ && \
1647: + (a_z+ib_z)(ia_y^*-b_y^*) \nonumber \\ &&
1648: - (a_z-ib_z)(ia_y^*+b_y^*) \nonumber \\ && \
1649: - (ia_y-b_y)(a_z^*+ib_z^*) \nonumber \\ && \
1650: - (-ia_y -b_y)(a_z^*-ib_z^*) \nonumber \\ && \
1651: - (a_z+ib_z)(-ia_y^*+b_y^*) \nonumber \\ && \
1652: + (-ia_y+b_y)(a_z^*+ib_z^*) \nonumber \\ && \
1653: + (a_z-ib_z)(-ia_y^*-b_y^*) \biggr] + {\rm c. \ c.}
1654: \nonumber \\ &=& 4Q_y J_{yz}^y N_{uc} \left( e^{iqx_{14}} [
1655: a_yb_z^* + b_ya_z^* - a_zb_y^* - b_za_y^* ] \right.
1656: \nonumber \\ && \ \left.
1657: + {\rm c. \ c.} \right) \nonumber \\
1658: &=& -16 N_{uc} Q_y \sin(qa/2) J_{yz}^y
1659: \Im [a_y^*b_z+b_y^*a_z] \ .
1660: \end{eqnarray}
1661:
1662: Now we collect terms proportional to $D_z^y$:
1663: \begin{eqnarray}
1664: V_y(D_z^y) &=& Q_y D_z^y \sum_{uc} \biggl[
1665: S_x(1)S_y(4)-S_y(1)S_x(4) \nonumber \\ && \
1666: + S_x(2)S_y(3) - S_y(2)S_x(3) \nonumber \\ &&
1667: + S_x(4)S_y(1') - S_y(4)S_x(1') \nonumber \\ && \
1668: + S_x(3)S_y(2') - S_y(3)S_x(2') \biggr]
1669: \nonumber \\ &=& Q_yD_z^y \sum_{uc} e^{iqx_{14}} \biggl[
1670: (a_x+ib_x)(ia_y^*-b_y^*) \nonumber \\ && \
1671: - (ia_y+b_y)(-a_x^*+ib_x^*) \nonumber \\ && \
1672: + (-a_x+ib_x)(ia_y^*+b_y^*) \nonumber \\ && \
1673: - (ia_y-b_y)(a_x^*+ib_x^*) \nonumber \\ && \
1674: + (-a_x-ib_x)(-ia_y^*+b_y^*) \nonumber \\ && \
1675: - (-ia_y-b_y)(a_x^*-ib_x^*) \nonumber \\ && \
1676: + (a_x-ib_x)(-ia_y^*-b_y^*) \nonumber \\ && \
1677: - (-ia_y+b_y)(-a_x^*-ib_x^*) \biggr] +
1678: {\rm c. \ c.} \nonumber \\ &=& \
1679: 4Q_y D_z^y N_{uc} \left( e^{iqx_{14}} [-a_xb_y^*-b_xa_y^*+a_yb_x^*+b_ya_x^*]
1680: \right. \nonumber \\ && \left. \ + {\rm c. \ c.} \right) \nonumber \\ &=&
1681: -16 N_{uc} Q_y \sin(qa/2) D_z^y \Im [a_y^*b_x+b_y^*a_x] \ .
1682: \end{eqnarray}
1683:
1684: Now we collect terms proportional to $D_y^y$:
1685: \begin{eqnarray}
1686: V_y(D_y^y) &=& Q_y D_y^y \sum_{uc} \biggl[
1687: - S_x(1)S_z(4) + S_z(1)S_x(4) \nonumber \\ && \
1688: - S_x(2)S_z(3) + S_z(2)S_x(3) \nonumber \\ &&
1689: + S_x(4)S_z(1') - S_z(4)S_x(1') \nonumber \\ && \
1690: + S_x(3)S_z(2') - S_z(3)S_x(2') \biggr]
1691: \nonumber \\ &=& Q_yD_y^y \sum_{uc} e^{iqx_{14}} \biggl[
1692: -(a_x+ib_x)(a_z^*-ib_z^*) \nonumber \\ && \
1693: + (a_z+ib_z)(-a_x^*+ib_x^*) \nonumber \\ && \
1694: - (-a_x+ib_x)(a_z^*+ib_z^*) \nonumber \\ && \
1695: + (a_z-ib_z)(a_x^*+ib_x^*) \nonumber \\ && \
1696: + (-a_x-ib_x)(a_z^*-ib_z^*) \nonumber \\ && \
1697: - (a_z+ib_z)(a_x^*-ib_x^*) \nonumber \\ && \
1698: + (a_x-ib_x)(a_z^*+ib_z^*) \nonumber \\ && \
1699: - (a_z-ib_z)(-a_x^*-ib_x^*) \biggr] \nonumber
1700: \\ && \ + {\rm c. \ c.} \nonumber \\ &=& \
1701: 4Q_y D_y^y N_{uc} \left( e^{iqx_{14}} i[a_xb_z^* - b_x a_z^* + a_zb_x^* -b_za_x^*]
1702: \right. \nonumber \\&& \left. \ + {\rm c. \ c.} \right) \nonumber \\ &=&
1703: -16 N_{uc} Q_y \cos(qa/2) D_y^y \Im [a_xb_z^*+b_x^*a_z] \ .
1704: \end{eqnarray}
1705:
1706: \subsection{nnn spine interactions}
1707:
1708: First we consider terms proportional to $K_{xx}^y$ (where we omit
1709: the $p$ index):
1710: \begin{eqnarray}
1711: V_y(K_{xx}^y) &=& Q_y K_{xx}^y \sum_{uc} [ S_x(1) S_x(1') - S_x(2) S_x(2')
1712: \nonumber \\ && \ - S_x(3) S_x(3') + S_x(4) S_x(4')] \nonumber \\
1713: &=& N_{uc}Q_y K_{xx}^y [ t_{xx} + t_{xx}^*] \ ,
1714: \end{eqnarray}
1715: where
1716: \begin{eqnarray}
1717: t_{xx} &=& e^{-iqa} \bigl[
1718: (a_{s,x} + i b_{s,x}) (a_{s,x}^*-ib_{s,x}^*)
1719: \nonumber \\ && \
1720: - (-a_{s,x} + i b_{s,x}) (-a_{s,x}^*-ib_{s,x}^*) \nonumber \\ && \
1721: -(a_{s,x} - i b_{s,x}) (a_{s,x}^* + ib_{s,x}^*)
1722: \nonumber \\ && \
1723: + (-a_{s,x} - i b_{s,x}) (-a_{s,x}^* +ib_{s,x}^*) \bigr]
1724: \nonumber \\ &=&
1725: e^{-iqa} \bigl[ -4i a_{s,x} b_{s,x}^* + 4i a_{s,x}^* b_{s,x} \bigr]
1726: \nonumber \\ &=& 8 e^{-iqa} \Im [a_{s,x} b_{s,x}^* ] \ ,
1727: \end{eqnarray}
1728: so that
1729: \begin{eqnarray}
1730: V_y(K_{xx}^y) &=& 16 N_{uc} Q_y K_{xx}^y \cos (qa) \Im [a_{s,x} b_{s,x}^*] \ .
1731: \end{eqnarray}
1732:
1733: Next, we consider terms proportional to $K_{yy}^y$:
1734: \begin{eqnarray}
1735: V_y(K_{yy}^y) &=& Q_y K_{yy}^y \sum_{uc} [ S_y(1) S_y(1') - S_y(2) S_y(2')
1736: \nonumber \\ && \ - S_y(3) S_y(3') + S_y(4) S_y(4') ] \nonumber \\
1737: &=& N_{uc}Q_y K_{yy}^y [ t_{yy} + t_{yy}^*] \ ,
1738: \end{eqnarray}
1739: where
1740: \begin{eqnarray}
1741: t_{yy} &=& e^{-iqa} \bigl[
1742: (ia_{s,y} + b_{s,y}) (-ia_{s,y}^* +b_{s,y}^*)
1743: \nonumber \\ && \
1744: - (ia_{s,y} - b_{s,y}) (-ia_{s,y}^*-b_{s,y}^*) \nonumber \\ && \
1745: -(-ia_{s,y} + b_{s,y}) (ia_{s,y}^* + b_{s,y}^*)
1746: \nonumber \\ && \
1747: + (-ia_{s,y} - b_{s,y}) (ia_{s,y}^* -b_{s,y}^*) \bigr]
1748: \nonumber \\ &=&
1749: e^{-iqa} \bigl[ 4i a_{s,y} b_{s,y}^* - 4i a_{s,y}^* b_{s,y} \bigr]
1750: \nonumber \\ &=& 8 e^{-iqa} \Im [a_{s,y}^* b_{s,y} ] \ ,
1751: \end{eqnarray}
1752: so that
1753: \begin{eqnarray}
1754: V_y(K_{yy}^y) &=& 16 N_{uc} Q_y K_{yy}^y \cos (qa) \Im [a_{s,y}^* b_{s,y}] \ .
1755: \end{eqnarray}
1756:
1757: Next, we consider terms proportional to $K_{zz}^y$:
1758: \begin{eqnarray}
1759: V_y(K_{zz}^y) &=& Q_y K_{zz}^y \sum_{uc} \bigl( S_z(1) S_z(1') - S_z(2) S_z(2')
1760: \nonumber \\ && \ - S_z(3) S_z(3') + S_z(4) S_z(4') \bigr) \nonumber \\
1761: &=& N_{uc}Q_y K_{zz}^y [ t_{zz} + t_{zz}^*] \ ,
1762: \end{eqnarray}
1763: where
1764: \begin{eqnarray}
1765: t_{zz} &=& e^{-iqa} \bigl[
1766: (a_{s,z} + ib_{s,z}) (a_{s,z}^* - ib_{s,z}^*)
1767: \nonumber \\ && \
1768: - (a_{s,z} - ib_{s,z}) (a_{s,z}^* +ib_{s,z}^*) \nonumber \\ && \
1769: -(a_{s,z} -i b_{s,z}) (a_{s,z}^* + ib_{s,z}^*) \nonumber \\ && \
1770: + (a_{s,z} +i b_{s,z}) (a_{s,z}^* - ib_{s,z}^*) \bigr]
1771: \nonumber \\ &=&
1772: e^{-iqa} \bigl[ -4i a_{s,z} b_{s,z}^* + 4i a_{s,z}^* b_{s,z} \bigr]
1773: \nonumber \\ &=& 8 e^{-iqa} \Im [a_{s,z} b_{s,z}^* ] \ ,
1774: \end{eqnarray}
1775: so that
1776: \begin{eqnarray}
1777: V_y(K_{zz}^y) &=& 16 N_{uc} Q_y K_{zz}^y \cos (qa) \Im [a_{s,z} b_{s,z}^*] \ .
1778: \end{eqnarray}
1779:
1780: Now we consider terms proportional to $E_z^y$:
1781: \begin{eqnarray}
1782: V_y(E_z^y) &=& Q_y E_z^y \sum_{uc} \bigl[ S_x(1) S_y(1') - S_y(1) S_x(1')
1783: \nonumber \\ && \ + S_x(2) S_y(2') - S_y(2) S_x(2') \nonumber \\ && \
1784: + S_x(3') S_y(3) - S_y(3') S_x(3) \nonumber \\ && \
1785: + S_x(4') S_y (4) - S_y(4') S_x(4) \bigr] \nonumber \\
1786: &=& N_{uc}Q_y E_z^y [ t_{zy} + t_{zy}^*] \ ,
1787: \end{eqnarray}
1788: where
1789: \begin{eqnarray}
1790: t_{zy} &=& e^{-iqa} \bigl[
1791: (a_{s,x} + ib_{s,x}) (-ia_{s,y}^* + b_{s,y}^*) \nonumber \\ && \
1792: - (ia_{s,y} + b_{s,y}) (a_{s,x}^* - ib_{s,x}^*) \nonumber \\ && \
1793: + (-a_{s,x} +i b_{s,x}) (-ia_{s,y}^* - b_{s,y}^*) \nonumber \\ && \
1794: - (ia_{s,y} - b_{s,y}) (-a_{s,x}^* - ib_{s,x}^*) \nonumber \\ && \
1795: + (a_{s,x} - ib_{s,x}) (ia_{s,y}^* + b_{s,y}^*) \nonumber \\ && \
1796: - (-ia_{s,y} + b_{s,y}) (a_{s,x}^* + ib_{s,x}^*) \nonumber \\ && \
1797: + (-a_{s,x} - i b_{s,x}) (ia_{s,y}^* - b_{s,y}^*) \nonumber \\ && \
1798: - (-ia_{s,y} - b_{s,y}) (-a_{s,x}^* + ib_{s,x}^*) \bigr] \nonumber \\ &=&
1799: 4e^{-iqa} \bigl[ a_{s,x} b_{s,y}^* + b_{s,x} a_{s,y}^*
1800: - a_{s,y} b_{s,x}^* - b_{s,y}a_{s,x}^* \bigr]
1801: \nonumber \\ &=& 8 i e^{-iqa} \Im [a_{s,x} b_{s,y}^*+b_{s,x} a_{s,y}^* ] \ ,
1802: \end{eqnarray}
1803: so that
1804: \begin{eqnarray}
1805: V_y(E_z^y) &=& 16 N_{uc} Q_y E_z^y \sin (qa) \Im [a_{s,x} b_{s,y}^*
1806: + b_{s,x}a_{s,y}^* ] \ .
1807: \end{eqnarray}
1808:
1809: Now we consider terms proportional to $E_x^y$:
1810: \begin{eqnarray}
1811: V_y(E_x^y) &=& Q_y E_x^y \sum_{uc} \bigl[ S_y(1) S_z(1') - S_z(1) S_y(1')
1812: \nonumber \\ && \
1813: - S_y(2) S_z(2') + S_z(2) S_y(2') \nonumber \\ && \
1814: + S_y(3') S_z(3) - S_z(3') S_y(3) \nonumber \\ && \
1815: - S_y(4') S_z (4) + S_z(4') S_y(4) \bigr] \nonumber \\
1816: &=& N_{uc}Q_y E_x^y [ t_{xy} + t_{xy}^*] \ ,
1817: \end{eqnarray}
1818: where
1819: \begin{eqnarray}
1820: t_{xy} &=& e^{-iqa} \bigl[
1821: (ia_{s,y} + b_{s,y}) (a_{s,z}^* -i b_{s,z}^*) \nonumber \\ && \
1822: - (a_{s,z} + ib_{s,z}) (-ia_{s,y}^* + b_{s,y}^*) \nonumber \\ && \
1823: - (ia_{s,y} - b_{s,y}) (a_{s,z}^* +i b_{s,z}^*) \nonumber \\ && \
1824: + (a_{s,z} - ib_{s,z}) (-ia_{s,y}^* - b_{s,y}^*) \nonumber \\ &&
1825: + (-ia_{s,y} + b_{s,y}) (a_{s,z}^* + ib_{s,z}^*) \nonumber \\ && \
1826: - (a_{s,z} -i b_{s,z}) (ia_{s,y}^* + b_{s,y}^*) \nonumber \\ && \
1827: - (-ia_{s,y} - b_{s,y}) (a_{s,z}^* -i b_{s,z}^*) \nonumber \\ && \
1828: + (a_{s,z} +i b_{s,z}) (ia_{s,y}^* - b_{s,y}^*) \bigr] \nonumber \\ &=&
1829: 4e^{-iqa} \bigl[ a_{s,y} b_{s,z}^* + b_{s,y} a_{s,z}^* \nonumber \\ && \
1830: - a_{s,z} b_{s,y}^* - b_{s,z}a_{s,y}^* \bigr]
1831: \nonumber \\ &=& 8 i e^{-iqa} \Im [a_{s,y} b_{s,z}^*+b_{s,y} a_{s,z}^* ] \ ,
1832: \end{eqnarray}
1833: so that
1834: \begin{eqnarray}
1835: V_y(E_x^y) &=& 16 N_{uc} Q_y E_x^y \sin (qa) \Im [a_{s,y} b_{s,z}^*
1836: + b_{s,y}a_{s,z}^* ] \ .
1837: \end{eqnarray}
1838:
1839: Now we consider terms proportional to $K_{xz}^y$:
1840: \begin{eqnarray}
1841: V_y(K_{xz}^y) &=& Q_y K_{xz}^y \sum_{uc} \bigl[ S_x(1) S_z(1') + S_z(1) S_x(1')
1842: \nonumber \\ && \
1843: + S_x(2) S_z(2') + S_z(2) S_x(2') \nonumber \\ && \
1844: - S_x(3) S_z(3') - S_z(3) S_x(3')\nonumber \\ && \
1845: - S_x(4) S_z (4') - S_z(4) S_x(4')
1846: \bigr] \nonumber \\
1847: &=& N_{uc}Q_y K_{xz}^y [ t_{xz} + t_{xz}^*] \ ,
1848: \end{eqnarray}
1849: where
1850: \begin{eqnarray}
1851: t_{xz} &=& e^{-iqa} \bigl[
1852: (a_{s,x} + ib_{s,x}) (a_{s,z}^* -i b_{s,z}^*) \nonumber \\ && \
1853: + (a_{s,z} + ib_{s,z}) (a_{s,x}^* -i b_{s,x}^*) \nonumber \\ && \
1854: + (-a_{s,x} +i b_{s,x}) (a_{s,z}^* +i b_{s,z}^*) \nonumber \\ && \
1855: + (a_{s,z} - ib_{s,z}) (-a_{s,x}^* - ib_{s,x}^*) \nonumber \\ &&
1856: - (a_{s,x} -i b_{s,x}) (a_{s,z}^* + ib_{s,z}^*) \nonumber \\ && \
1857: - (a_{s,z} -i b_{s,z}) (a_{s,x}^* + ib_{s,x}^*) \nonumber \\ && \
1858: - (-a_{s,x} - ib_{s,x}) (a_{s,z}^* -i b_{s,z}^*) \nonumber \\ && \
1859: - (a_{s,z} +i b_{s,z}) (-a_{s,y}^* +i b_{s,y}^*) \bigr] \nonumber \\ &=&
1860: 4ie^{-iqa} \bigl[ -a_{s,x} b_{s,z}^* + b_{s,x} a_{s,z}^*
1861: - a_{s,z} b_{s,x}^* + b_{s,z}a_{s,x}^* \bigr]
1862: \nonumber \\ &=& 8 e^{-iqa} \Im [a_{s,x} b_{s,z}^*+a_{s,z} b_{s,x}^* ] \ ,
1863: \end{eqnarray}
1864: so that
1865: \begin{eqnarray}
1866: V_y(K_{xz}^y) &=& 16 N_{uc} Q_y K_{xz}^y \cos (qa) \nonumber \\ && \
1867: \times \Im [a_{s,x} b_{s,z}^* + a_{s,z} b_{s,x}^* ] \ .
1868: \end{eqnarray}
1869:
1870: \subsection{Spine cross-tie interactions}
1871:
1872: Here we analyze the spin-cross tie interactions.
1873:
1874: We consider the terms proportional to $L_{xx}^y$:
1875: \begin{eqnarray}
1876: V_y(L_{xx}^y) &=& Q_y L_{xx}^y [ S_x(5) + S_x(6)]
1877: \nonumber \\ && \ \times
1878: [S_x(1) - S_x(2) - S_x(3) + S_x(4)] \nonumber \\
1879: &\equiv& N_{uc} Q_y L_{xx}^y [ u_{xx} + u_{xx}^*] \ ,
1880: \end{eqnarray}
1881: where
1882: \begin{eqnarray}
1883: u_{xx} &=& b_{cx} [ (a_{s,x}^*-ib_{s,x}^*)e^{-iqa/4} \nonumber \\ && \
1884: - (-a_{s,x}^*-ib_{s,x}^*)e^{-iqa/4}
1885: - (a_{s,x}^*+ib_{s,x}^*)e^{iqa/4} \nonumber \\ && \
1886: + (-a_{s,x}^*+ib_{s,x}^*)e^{iqa/4} \nonumber \\ && \
1887: - (a_{s,x}^*-ib_{s,x}^*)e^{iqa/4}
1888: + (-a_{s,x}^*-ib_{s,x}^*)e^{iqa/4} \nonumber \\ && \
1889: + (a_{s,x}^*+ib_{s,x}^*)e^{-iqa/4}
1890: - (-a_{s,x}^*+ib_{s,x}^*)e^{-iqa/4} ] \nonumber \\ &=&
1891: 4 b_{cx} a_{s,x}^* [e^{-iqa/4} - e^{iqa/4}]
1892: \nonumber \\ &=& -8i \sin(qa/4) b_{cx} a_{s,x}^*
1893: \end{eqnarray}
1894: so that
1895: \begin{eqnarray}
1896: V_y(L_{xx}^y) &=& 16 N_{uc} Q_y L_{xx}^y \sin (qa/4) \Im [b_{cx} a_{s,x}^*] \ .
1897: \end{eqnarray}
1898:
1899: We consider the terms proportional to $L_{yy}^y$:
1900: \begin{eqnarray}
1901: V_y(L_{yy}^y) &=& Q_y L_{yy}^y [ S_y(5) + S_y(6)] \nonumber \\ && \
1902: [S_y(1) - S_y(2) - S_y(3) + S_y(4)] \nonumber \\ &\equiv&
1903: N_{uc} Q_y L_{yy}^y [ u_{yy} + u_{yy}^*] \ ,
1904: \end{eqnarray}
1905: where
1906: \begin{eqnarray}
1907: u_{yy} &=& a_{cy} [ (-ia_{s,y}^*+b_{s,y}^*)e^{-iqa/4}
1908: - (-ia_{s,y}^*-b_{s,y}^*)e^{-iqa/4} \nonumber \\ && \
1909: - (ia_{s,y}^*+b_{s,y}^*)e^{iqa/4}
1910: + (ia_{s,y}^*-b_{s,y}^*)e^{iqa/4} \nonumber \\ && \
1911: - (-ia_{s,y}^*+b_{s,y}^*)e^{iqa/4}
1912: + (-ia_{s,y}^*-b_{s,y}^*)e^{iqa/4} \nonumber \\ && \
1913: + (ia_{s,y}^*+b_{s,y}^*)e^{-iqa/4}
1914: - (ia_{s,y}^*-b_{s,y}^*)e^{-iqa/4} ] \nonumber \\ &=&
1915: 4 a_{cy} b_{s,y}^* [e^{-iqa/4} - e^{iqa/4}]
1916: \nonumber \\ &=& -8i \sin(qa/4) a_{cy} b_{s,y}^*
1917: \end{eqnarray}
1918: so that
1919: \begin{eqnarray}
1920: V_y(L_{yy}^y) &=& 16 N_{uc} Q_y L_{yy}^y \sin (qa/4) \Im [a_{cy} b_{s,y}^*] \ .
1921: \end{eqnarray}
1922:
1923: We consider the terms proportional to $L_{zz}^y$:
1924: \begin{eqnarray}
1925: V_y(L_{zz}^y) &=& Q_y L_{zz}^y [ S_z(5) + S_z(6)] \nonumber \\ &&
1926: \ \times [S_z(1) - S_z(2) - S_z(3) + S_z(4)] \nonumber \\
1927: &\equiv& N_{uc} Q_y L_{zz}^y [ u_{zz} + u_{zz}^*] \ ,
1928: \end{eqnarray}
1929: where
1930: \begin{eqnarray}
1931: u_{zz} &=& a_{cz} [ (a_{s,z}^*-ib_{s,z}^*)e^{-iqa/4}
1932: - (a_{s,z}^*+ib_{s,z}^*)e^{-iqa/4} \nonumber \\ && \
1933: - (a_{s,z}^*+ib_{s,z}^*)e^{iqa/4}
1934: + (a_{s,z}^*-ib_{s,z}^*)e^{iqa/4} \nonumber \\ && \
1935: + (a_{s,z}^*-ib_{s,z}^*)e^{iqa/4}
1936: - (a_{s,z}^*+ib_{s,z}^*)e^{iqa/4} \nonumber \\ && \
1937: - (a_{s,z}^*+ib_{s,z}^*)e^{-iqa/4}
1938: + (a_{s,z}^*-ib_{s,z}^*)e^{-iqa/4} ] \nonumber \\
1939: &=& 4 a_{cz} b_{s,z}^* i \left[ - e^{-iqa/4} - e^{iqa/4} \right]
1940: \nonumber \\ &=& -8i a_{cz} b_{s,z}^* \cos(aq/4) \ ,
1941: \end{eqnarray}
1942: so that
1943: \begin{eqnarray}
1944: V_y(L_{zz}^y) &=& 16 N_{uc} Q_y L_{zz}^y \cos(qa/4) \Im [ a_{cz} b_{s,z}^* ] \ .
1945: \end{eqnarray}
1946:
1947: We consider the terms proportional to $L_{xy}^y$:
1948: \begin{eqnarray}
1949: V_y(L_{xy}^y) &=& Q_y L_{xy}^y \biggl( [ S_x(5) - S_x(6)]
1950: \nonumber \\ && \ \times [S_y(1) + S_y(2) - S_y(3) - S_y(4)]
1951: \nonumber \\ && \ + [ S_y(5) - S_y(6)] \nonumber \\
1952: && \ \times [S_x(1) + S_x(2) - S_x(3) - S_x(4)]
1953: \biggr) \nonumber \\ & \equiv&
1954: N_{uc}Q_y L_{xy}^y [u_{xy} + u_{xy}^*] \ ,
1955: \end{eqnarray}
1956: where
1957: \begin{eqnarray}
1958: u_{xy} &=& b_{cx} \left[ (-ia_{s,y}^*+b_{s,y}^*)e^{-iqa/4}
1959: + (-ia_{s,y}^*-b_{s,y}^*)e^{-iqa/4} \right. \nonumber \\ && \
1960: - (ia_{s,y}^*+b_{s,y}^*)e^{iqa/4}
1961: - (ia_{s,y}^*-b_{s,y}^*)e^{iqa/4} \nonumber \\ && \
1962: + (-ia_{s,y}^*+b_{s,y}^*)e^{iqa/4}
1963: + (-ia_{s,y}^*-b_{s,y}^*)e^{iqa/4} \nonumber \\ && \
1964: \left. - (ia_{s,y}^*+b_{s,y}^*)e^{-iqa/4}
1965: - (ia_{s,y}^*-b_{s,y}^*)e^{-iqa/4} \right] \nonumber \\ &&
1966: + a_{cy} \left[
1967: (a_{s,x}^* - i b_{s,x}^*) e^{-iqa/4} \right. \nonumber \\ && \
1968: + (-a_{s,x}^* - i b_{s,x}^*) e^{-iqa/4}
1969: - (a_{s,x}^* + i b_{s,x}^*) e^{iqa/4} \nonumber \\ && \
1970: - (-a_{s,x}^* + i b_{s,x}^*) e^{iqa/4} \nonumber \\ && \
1971: + (a_{s,x}^* - i b_{s,x}^*) e^{iqa/4}
1972: + (-a_{s,x}^* - i b_{s,x}^*) e^{iqa/4} \nonumber \\ && \
1973: \left. - (a_{s,x}^* + i b_{s,x}^*) e^{-iqa/4}
1974: - (-a_{s,x}^* + i b_{s,x}^*) e^{-iqa/4} \right] \nonumber \\ &=&
1975: 4 i b_{cx} \left[ -a_{s,y}^* e^{-iqa/4} - a_{s,y}^* e^{iqa/4} \right]
1976: \nonumber \\ && \
1977: + 4 i a_{cy} \left[ -b_{s,x}^*e^{-iqa/4} -b_{s,x}^*e^{iqa/4} \right]
1978: \nonumber \\ &=& -8i[b_{cx} a_{s,y}^* +a_{cy} b_{s,x}^*] \cos (qa/4) \ ,
1979: \end{eqnarray}
1980: so that
1981: \begin{eqnarray}
1982: V_y(L_{xy}^y) &=& 16 N_{uc} Q_y L_{xy}^y \cos(qa/4) \nonumber \\ &&
1983: \ \times \Im \left[ b_{cx}a_{s,y}^* + a_{cy} b_{s,x}^* \right] \ .
1984: \end{eqnarray}
1985:
1986: We consider the terms proportional to $L_{xz}^y$:
1987: \begin{eqnarray}
1988: V_y(L_{xz}^y) &=& Q_y L_{xz}^y \Biggl( [ S_x(5) + S_x(6)] \nonumber \\
1989: && \ \times [S_z(1) + S_z(2) - S_z(3) - S_z(4)]
1990: \nonumber \\ && \
1991: + [ S_z(5) + S_z(6)] \nonumber \\ && \ \times
1992: [S_x(1) + S_x(2) - S_x(3) - S_x(4)]
1993: \Biggr) \nonumber \\ & \equiv&
1994: N_{uc}Q_y L_{xz}^y [u_{xz} + u_{xz}^*] \ ,
1995: \end{eqnarray}
1996: where
1997: \begin{eqnarray}
1998: u_{xz} &=& b_{cx} \left[ (a_{s,z}^*-ib_{s,z}^*)e^{-iqa/4}
1999: + (a_{s,z}^*+ib_{s,z}^*)e^{-iqa/4} \right. \nonumber \\ && \
2000: - (a_{s,z}^*+ib_{s,z}^*)e^{iqa/4}
2001: - (a_{s,z}^*-ib_{s,z}^*)e^{iqa/4} \nonumber \\ && \
2002: - (a_{s,z}^*-ib_{s,z}^*)e^{iqa/4}
2003: - (a_{s,z}^*+ib_{s,z}^*)e^{iqa/4} \nonumber \\ && \
2004: \left. + (a_{s,z}^*+ib_{s,z}^*)e^{-iqa/4}
2005: + (a_{s,z}^*-ib_{s,z}^*)e^{-iqa/4} \right] \nonumber \\ && \
2006: + a_{cz} \left[ (a_{s,x}^* -i b_{s,x}^*) e^{-iqa/4}
2007: \right. \nonumber \\ && \
2008: + (-a_{s,x}^* -i b_{s,x}^*) e^{-iqa/4} \nonumber \\ && \
2009: - (a_{s,x}^* +i b_{s,x}^*) e^{iqa/4}
2010: - (-a_{s,x}^* +i b_{s,x}^*) e^{iqa/4} \nonumber \\ && \
2011: + (a_{s,x}^* -i b_{s,x}^*) e^{iqa/4}
2012: + (-a_{s,x}^* -i b_{s,x}^*) e^{iqa/4} \nonumber \\ && \
2013: \left. - (a_{s,x}^* +i b_{s,x}^*) e^{-iqa/4}
2014: - (-a_{s,x}^* +i b_{s,x}^*) e^{-iqa/4} \right] \nonumber \\ &=&
2015: 4 b_{cx} a_{s,z}^* [e^{-iqa/4} - e^{-iqa/4} ] \nonumber \\ && \
2016: + 4 a_{cz} b_{s,x}^* [-ie^{-iqa/4} - i e^{iqa/4} ]
2017: \nonumber \\ &=&
2018: -8ib_{cx} a_{s,z}^* \sin (qa/4) -8i a_{cz} b_{s,x}^* \cos (qa/4) \ ,
2019: \end{eqnarray}
2020: so that
2021: \begin{eqnarray}
2022: V_y (L_{xz}^y) &=& 16 N_{uc} Q_y L_{xz}^y \Biggl(
2023: \sin(qa/4) \Im[ b_{cx}a_{s,z}^*] \nonumber \\ && \
2024: + \cos(qa/4) \Im[ a_{cz} b_{s,x}^*] \Biggr) \ .
2025: \end{eqnarray}
2026:
2027: We consider the terms proportional to $L_{yz}^y$:
2028: \begin{eqnarray}
2029: V_y(L_{yz}^y) &=& Q_y L_{yz}^y \Biggl( [ S_y(5) - S_y(6)] \nonumber \\
2030: && \ \times [S_z(1) - S_z(2) - S_z(3) + S_z(4)]
2031: \nonumber \\ && \ + [ S_z(5) - S_z(6)] \nonumber \\ && \
2032: \times [S_y(1) - S_y(2) - S_y(3) + S_y(4)] \Biggr)
2033: \nonumber \\ &\equiv& N_{uc}Q_y L_{yz}^y [u_{yz} + u_{yz}^*] \ ,
2034: \end{eqnarray}
2035: where
2036: \begin{eqnarray}
2037: u_{yz} &=& a_{cy} \left[ (a_{s,z}^*-ib_{s,z}^*)e^{-iqa/4}
2038: - (a_{s,z}^*+ib_{s,z}^*)e^{-iqa/4} \right. \nonumber \\ && \
2039: - (a_{s,z}^*+ib_{s,z}^*)e^{iqa/4}
2040: + (a_{s,z}^*-ib_{s,z}^*)e^{iqa/4} \nonumber \\ && \
2041: + (a_{s,z}^*-ib_{s,z}^*)e^{iqa/4}
2042: - (a_{s,z}^*+ib_{s,z}^*)e^{iqa/4} \nonumber \\ && \
2043: \left. - (a_{s,z}^*+ib_{s,z}^*)e^{-iqa/4}
2044: + (a_{s,z}^*-ib_{s,z}^*)e^{-iqa/4} \right] \nonumber \\ && \
2045: + a_{c,z} \left[ (-ia_{s,y}^*+b_{s,y}^*)e^{-iqa/4} \right.
2046: \nonumber \\ && \ - (-ia_{s,y}^*-b_{s,y}^*)e^{-iqa/4}
2047: \nonumber \\ && \ - (ia_{s,y}^*+b_{s,y}^*)e^{iqa/4}
2048: + (ia_{s,y}^*-b_{s,y}^*)e^{iqa/4} \nonumber \\ && \
2049: - (-ia_{s,y}^*+b_{s,z}^*)e^{iqa/4}
2050: + (-ia_{s,y}^*-b_{s,y}^*)e^{iqa/4} \nonumber \\ && \
2051: \left. + (ia_{s,y}^*+b_{s,y}^*)e^{-iqa/4}
2052: - (ia_{s,y}^*-b_{s,y}^*)e^{-iqa/4} \right] \nonumber \\ &=&
2053: -4i a_{cy} b_{sz}^* \left[ e^{-iqa/4} + e^{iqa/4} \right]
2054: \nonumber \\ && \ +
2055: 4 a_{cz} b_{s,y}^* \left[ e^{-iqa/4} - e^{iqa/4} \right] \nonumber \\
2056: &=& -8i a_{cy} b_{s,z}^* \cos (aq/4) -8ia_{cz} b_{s,y}^*
2057: \sin (qa/4) \ ,
2058: \end{eqnarray}
2059: so that
2060: \begin{eqnarray}
2061: V_y(L_{yz}^y) &=& 16 N_{uc} Q_y L_{yz}^y \left( \cos(qa/4) \Im[ a_{cy}b_{s,z}^*]
2062: \right. \nonumber \\ && \left. \ + \sin(qa/4) \Im[a_{cz} b_{s,y}^* ] \right) \ .
2063: \end{eqnarray}
2064:
2065: We consider the terms proportional to $F_x^y$:
2066: \begin{eqnarray}
2067: V_y(F_x^y) &=& Q_y F_x^y \Biggl( [ S_y(5) - S_y(6)] \nonumber \\ &&
2068: \ \times [S_z(1) - S_z(2) - S_z(3) + S_z(4)]
2069: \nonumber \\ && \ - [ S_z(5) - S_z(6)] \nonumber \\ &&
2070: \ \times [S_y(1) - S_y(2) - S_y(3) + S_y(4)] \Biggr)
2071: \nonumber \\ &\equiv& N_{uc}Q_y F_x^y [u_x + u_x^*] \ ,
2072: \end{eqnarray}
2073: where
2074: \begin{eqnarray}
2075: u_x &=& a_{cy} \left[ (a_{s,z}^*-ib_{s,z}^*)e^{-iqa/4}
2076: - (a_{s,z}^*+ib_{s,z}^*)e^{-iqa/4} \right. \nonumber \\ && \
2077: - (a_{s,z}^*+ib_{s,z}^*)e^{iqa/4}
2078: + (a_{s,z}^*-ib_{s,z}^*)e^{iqa/4} \nonumber \\ && \
2079: + (a_{s,z}^*-ib_{s,z}^*)e^{iqa/4}
2080: - (a_{s,z}^*+ib_{s,z}^*)e^{iqa/4} \nonumber \\ && \
2081: \left. - (a_{s,z}^*+ib_{s,z}^*)e^{-iqa/4}
2082: + (a_{s,z}^*-ib_{s,z}^*)e^{-iqa/4} \right] \nonumber \\ && \
2083: + a_{c,z} \left[ -(-ia_{s,y}^*+b_{s,y}^*)e^{-iqa/4}
2084: \right. \nonumber \\ && \
2085: + (-ia_{s,y}^*-b_{s,y}^*)e^{-iqa/4} \nonumber \\ && \
2086: + (ia_{s,y}^*+b_{s,y}^*)e^{iqa/4}
2087: - (ia_{s,y}^*-b_{s,y}^*)e^{iqa/4} \nonumber \\ && \
2088: + (-ia_{s,y}^*+b_{s,y}^*)e^{iqa/4}
2089: - (-ia_{s,y}^*-b_{s,y}^*)e^{iqa/4} \nonumber \\ && \
2090: \left. - (ia_{s,y}^*+b_{s,y}^*)e^{-iqa/4}
2091: + (ia_{s,y}^*-b_{s,y}^*)e^{-iqa/4} \right] \nonumber \\ &=& \
2092: -4i a_{cy} b_{s,z}^* \left[ e^{-iqa/4} + e^{iqa/4} \right]
2093: \nonumber \\ && \ +
2094: 4 a_{cz} b_{s,y}^* \left[ e^{iqa/4} - e^{-iqa/4} \right] \nonumber \\
2095: &=& -8ia_{cy} b_{s,z}^* \cos (qa/4) + 8i a_{cz} b_{s,y}^* \sin (qa/4) \ ,
2096: \end{eqnarray}
2097: so that
2098: \begin{eqnarray}
2099: V_y(F_x^y) &=& 16 N_{uc} Q_y F_x^y \Biggl(
2100: \cos(qa/4) \Im[ a_{cy}b_{s,z}^*] \nonumber \\ && \
2101: + \sin (qa/4) \Im [a_{cz}^* b_{s,y} ] \Biggr) \ .
2102: \end{eqnarray}
2103:
2104: We consider the terms proportional to $F_y^y$:
2105: \begin{eqnarray}
2106: V_y(F_y^y) &=& Q_y F_y^y \Biggl( -[ S_x(5) + S_x(6)] \nonumber \\ &&
2107: \ \times [S_z(1) + S_z(2) - S_z(3) - S_z(4)]
2108: \nonumber \\ && \
2109: + [ S_z(5) + S_z(6)] \nonumber \\ && \
2110: \times [S_x(1) + S_x(2) - S_x(3) - S_x(4)] \Biggr) \nonumber
2111: \\ &\equiv& N_{uc}Q_y L_{yz}^y [u_y + u_y^*] \ ,
2112: \end{eqnarray}
2113: where
2114: \begin{eqnarray}
2115: u_y &=& b_{cx} \left[ -(a_{s,z}^*-ib_{s,z}^*)e^{-iqa/4}
2116: - (a_{s,z}^*+ib_{s,z}^*)e^{-iqa/4} \right. \nonumber \\ && \
2117: + (a_{s,z}^*+ib_{s,z}^*)e^{iqa/4}
2118: + (a_{s,z}^*-ib_{s,z}^*)e^{iqa/4} \nonumber \\ && \
2119: + (a_{s,z}^*-ib_{s,z}^*)e^{iqa/4}
2120: + (a_{s,z}^*+ib_{s,z}^*)e^{iqa/4} \nonumber \\ && \
2121: \left. - (a_{s,z}^*+ib_{s,z}^*)e^{-iqa/4}
2122: - (a_{s,z}^*-ib_{s,z}^*)e^{-iqa/4} \right] \nonumber \\ && \
2123: + a_{c,z} \left[ (a_{s,x}^* -ib_{s,x}^*)e^{-iqa/4}
2124: \right. \nonumber \\ && \
2125: + (-a_{s,x}^*-ib_{s,x}^*)e^{-iqa/4} \nonumber \\ && \
2126: - (a_{s,x}^*+ib_{s,x}^*)e^{iqa/4}
2127: - (-a_{s,x}^*+ib_{s,x}^*)e^{iqa/4} \nonumber \\ && \
2128: (a_{s,x}^*-ib_{s,x}^*)e^{iqa/4}
2129: + (-a_{s,x}^*-ib_{s,x}^*)e^{iqa/4} \nonumber \\ && \
2130: \left. - (a_{s,x}^*+ib_{s,x}^*)e^{-iqa/4}
2131: - (-a_{s,x}^*+ib_{s,x}^*)e^{-iqa/4} \right] \nonumber \\ &=& \
2132: 4b_{cx} a_{s,z}^* \left[ -e^{-iqa/4} + e^{iqa/4} \right] \nonumber
2133: \\ && \ + 4 a_{cz} b_{s,x}^* \left[ -i e^{-iqa/4} -i e^{iqa/4} \right] \nonumber \\
2134: &=& 8i b_{cx} a_{s,z}^* \sin (qa/4) -8i a_{cz} b_{s,x}^* \cos (aq/4) \ ,
2135: \end{eqnarray}
2136: so that
2137: \begin{eqnarray}
2138: V_y(F_y^y) &=& 16 N_{uc} Q_y F_y^y \Biggl( \sin(qa/4) \Im[ b_{cx}^* a_{s,z}]
2139: \nonumber \\ && \ + \cos(aq/4) \Im [ a_{cz} b_{s,x}^*] \Biggr) \ .
2140: \end{eqnarray}
2141:
2142: Finally, we consider the terms proportional to $F_z^y$:
2143: \begin{eqnarray}
2144: V_y(F_z^y) &=& Q_y F_z^y \Biggl( [ S_x(5) - S_x(6)] \nonumber \\ && \
2145: \times [S_y(1) + S_y(2) - S_y(3) - S_y(4)] \nonumber \\ && \
2146: - [ S_y(5) - S_y(6)] \nonumber \\ && \
2147: \times [S_x(1) + S_x(2) - S_x(3) - S_x(4)] \Biggr)
2148: \nonumber \\ &\equiv& N_{uc}Q_y F_z^y [u_z + u_z^*] \ ,
2149: \end{eqnarray}
2150: where
2151: \begin{eqnarray}
2152: u_z &=& b_{cx} \left[ (-ia_{s,y}^*+b_{s,y}^*)e^{-iqa/4}
2153: \right. \nonumber \\ && \
2154: + (-ia_{s,y}^*-b_{s,y}^*)e^{-iqa/4} \nonumber \\ && \
2155: - (ia_{s,y}^*+b_{s,y}^*)e^{iqa/4}
2156: - (ia_{s,y}^*-b_{s,y}^*)e^{iqa/4} \nonumber \\ && \
2157: + (-ia_{s,y}^*+b_{s,y}^*)e^{iqa/4}
2158: + (-ia_{s,y}^*-b_{s,y}^*)e^{iqa/4} \nonumber \\ && \
2159: \left. - (ia_{s,y}^*+b_{s,y}^*)e^{-iqa/4}
2160: - (ia_{s,y}^*-b_{s,y}^*)e^{-iqa/4} \right] \nonumber \\ && \
2161: + a_{c,y} \left[ -(a_{s,x}^*-ib_{s,x}^*)e^{-iqa/4}
2162: \right. \nonumber \\ && \
2163: - (-a_{s,x}^*-ib_{s,x}^*)e^{-iqa/4} \nonumber \\ && \
2164: + (a_{s,x}^*+ib_{s,x}^*)e^{iqa/4}
2165: + (-a_{s,x}^*+ib_{s,x}^*)e^{iqa/4} \nonumber \\ && \
2166: - (a_{s,x}^*-ib_{s,x}^*)e^{iqa/4}
2167: - (-a_{s,x}^*-ib_{s,x}^*)e^{iqa/4} \nonumber \\ && \
2168: \left. + (a_{s,x}^*+ib_{s,x}^*)e^{-iqa/4}
2169: + (-a_{s,x}^*+ib_{s,x}^*)e^{-iqa/4}\right] \nonumber \\ &=& \
2170: 4ib_{cx} a_{s,y}^* \left[ -e^{-iqa/4} - e^{iqa/4} \right] \nonumber \\ && \
2171: +4 i a_{cy} b_{s,x}^* \left[ e^{-iqa/4} + e^{iqa/4} \right] \nonumber \\
2172: &=& -8i b_{cx} a_{s,y}^* \cos(qa/4) + 8 i a_{cy} b_{s,x}^* \cos(qa/4) \ ,
2173: \end{eqnarray}
2174: so that
2175: \begin{eqnarray}
2176: V_y(F_z^y) &=& 16 N_{uc} Q_y F_z^y \cos(qa/4) \nonumber \\ &&
2177: \ \times \Im[ b_{cx}a_{s,y}^* + a_{cy}^* b_{s,x} ] \ .
2178: \end{eqnarray}
2179: \end{appendix}
2180:
2181: \begin{thebibliography}{10}
2182: \bibitem{Rogado} %1
2183: N. Rogado, G. Lawes, D. A. Huse, A. P. Ramirez, and R. J. Cava,
2184: Solid State Comm. {\bf 124}, 229 (2002).
2185:
2186: \bibitem{LawesKenzelmann} %2
2187: G. Lawes, M. Kenzelmann, N. Rogado, K. H. Kim, G. A. Jorge, R. J. Cava,
2188: A. Aharony, O. Entin-Wohlman, A. B. Harris, T. Yildirim, Q. A. Huang,
2189: S. Park, C. Broholm, and A. P. Ramirez, Phys. Rev. Lett. {\bf 93},
2190: 247201 (2004).
2191:
2192: \bibitem{FERRO} %3
2193: G. Lawes, A. B. Harris, T. Kimura, N. Rogado, R. J. Cava, A. Aharony,
2194: O. Entin-Wohlman, T. Yildirim, M. Kenzelmann, C. Broholm, and
2195: A. P. Ramirez, Phys. Rev. Lett. {\bf 95}, 087205 (2005).
2196:
2197: \bibitem{EXPT} %4
2198: M. Kenzelmann, A. B. Harris, A. Aharony, O. Entin-Wohlman, T. Yildirim,
2199: Q. Huang, S. Park, G. Lawes, C. Broholm, N. Rogado, R. J. Cava,
2200: K. H. Kim, G. Jorge, and A. P. Ramirez, cond-mat/0509xxx.
2201:
2202: \bibitem{TMO1} %5
2203: N. Hur, S. Park, P. A. Sharma, J. S. Ahn, S. Guha, and S.-W. Cheong,
2204: Nature {\bf 429}, 392 (2004).
2205:
2206: \bibitem{TMO2} %6
2207: M. Kenzelmann, A. B. Harris, S. Jonas, C. Broholm, J. Schefer, S. B. Kim,
2208: C. L. Zhang, S.-W. Cheong, O. P. Vajk, and J. W. Lynn, Phys. Rev. Lett.
2209: {\bf 95}, 087206 (2005).
2210:
2211: \bibitem{REV} %7
2212: A. B. Harris and G. Lawes, {\it Ferroelectricity in Incommensurate
2213: Magnets}, in {\it The Handbook of Magnetism and Advanced Magnetic Materials},
2214: (J. Wiley, London, 2006); cond-mat/0508617; A. B. Harris, cond-mat/0508730.
2215:
2216: \bibitem{DM1} %8
2217: I. Dzyaloshinskii, J. Phys. Chem. Solids {\bf 4}, 241 (1958).
2218:
2219: \bibitem{DM2} %9
2220: T. Moriya, Phys. Rev. {\bf 120}, 91 (1960).
2221:
2222: \bibitem{DAGOTTO}
2223: I. A. Sergienko and E. Dagotto, cond-mat/0508075.
2224:
2225: \bibitem{LDA} %10
2226: P. C. Hohenberg and W. Kohn, Phys. Rev. {\bf 136}, B864 (1964);
2227: W. Kohn and L. J. Sham, Phys. Rev. {\bf 140}, A1133 (1965).
2228:
2229: \bibitem{LDAU} %11
2230: V. I. Anisimov, F. Aryasetiawan, and A. I. Lichtenstein,
2231: J. Phys. Cond. Matter {\bf 9}, 767 (1997).
2232:
2233: \bibitem{HAHN} %12
2234: A.~J.~C. Wilson, {\em International Tables For Crystallography} (Kluwer
2235: Academic Publishers, Dordrecht, 1995), Vol.~A.
2236:
2237: \bibitem{SAUER} %13
2238: E. E. Sauerbrei, F. Faggiani, and C. Calvo, Acta Crystallogr. B {\bf 29},
2239: 2304 (1973).
2240:
2241: \bibitem{PWA} %14
2242: P. W. Anderson, Phys. Rev. {\bf 115}, 2 (1959).
2243:
2244: \bibitem{TY} %14
2245: T. Yildirim, A. B. Harris, A. Aharony, and O. Entin-Wohlman, Phys. Rev.
2246: B {\bf 52}, 10239 (1996).
2247:
2248: \bibitem{OE} %15
2249: O. Entin-Wohlman, A. B. Harris, and A. Aharony, Phys. Rev. B {\bf 53},
2250: 11661 (1996).
2251:
2252: \bibitem{RS1} %16
2253: R. Schmitz, O. Entin-Wohlman, A. Aharony, A. B. Harris, and E.
2254: M\"uller-Hartmann, Phys. Rev. B {\bf 71}, 144412 (2005).
2255:
2256: \bibitem{RS2} %17
2257: R. Schmitz, O. Entin-Wohlman, A. Aharony, A. B. Harris, and
2258: E. M\"uller-Hartmann, Phys. Rev. B {\bf 71}, 214438 (2005).
2259:
2260: \bibitem{fans} %18
2261: J. R. D. Copley, D. A. Neumann, W. A. Kamitakahara,
2262: Can. J. Phys. {\bf 73}, 763 (1995).
2263:
2264: \bibitem{pwscf} %19
2265: S. Baroni, A. Dal Corso, S. de Gironcoli, and P. Giannozzi,
2266: http://www.pwscf.org
2267:
2268: \bibitem{taner_phonon} %20
2269: T. Yildirim, Chem. Phys. {\bf 261}, 205 (2000).
2270:
2271: \bibitem{taner_web} %21
2272: Animations of the B$_{2u}$ phonons can be found at
2273: http://www.ncnr.nist.gov/staff/taner/nvo
2274:
2275: \bibitem{GROUP} %22
2276: E. P. Wigner, {\it Group Theory}. (Academic Press, New York, 1959).
2277: M. Tinkham, {\it Group Theory
2278: and Quantum Mechanics}, (McGraw-Hill, New York, 1964).
2279:
2280: \bibitem{THEORY} %23
2281: A.~B. Harris {\it et al.}, unpublished.
2282:
2283: \end{thebibliography}
2284: \end{document}
2285:
2286: \bibitem{BroholmAeppli} %3
2287: C. Broholm, G. Aeppli, G.~P. Espinosa, and A.~S. Cooper, Phys. Rev. Lett.
2288: {\bf 65}, 3173 (1990).
2289:
2290: \bibitem{RamirezEspinosa} %4
2291: A.~P. Ramirez, G.~P. Espinosa, and A.~S. Cooper, Phys. Rev. Lett. {\bf 64},
2292: 2070 (1990).
2293:
2294: \bibitem{LeeBroholm} %5
2295: S.-H. Lee, C. Broholm, M. F. Collins, L. Heller, A. P. Ramirez, Ch. Kloc,
2296: E. Bucheri, R. W. Erwin, and N. Lacevic, Phys. Rev. B {\bf 56}, 8091 (1997).
2297:
2298: \bibitem{YSLee} %6
2299: D. Grohol, Q. Huang, B. H. Toby, J. W. Lynn, Y. S. Lee,
2300: and D. G. Nocera, Phys. Rev. B {\bf 68}, 094404 (2003).
2301:
2302: \bibitem{Sauerbrei} %12
2303: E.~E. Sauerbrei, F. Faggiani, and C. Calvo, Acta Crystallogr. B {\bf 29},
2304: 2304 (1973).
2305:
2306: \bibitem{GSAS} %13
2307: A. Larson and R.~V. Dreele, Los Alamos National Laboratory Report No.
2308: LAUR086-748, 1990.
2309:
2310: %\bibitem{Fullprof}
2311: %J. Rodriguez-Carvajal, Physica B {\bf 192}, 55 (1993).
2312:
2313: \bibitem{ChouAharony} %15
2314: F.~C. Chou, A. Aharony, R. J. Birgeneau, O. Entin-Wohlman, M. Greven,
2315: A. B. Harris, M. A. Kastner, Y. J. Kim, D. S. Kleinberg, Y. S. Lee, and
2316: Q. Zhu, Phys. Rev. Lett. {\bf 78}, 535 (1997).
2317:
2318: \bibitem{KiryukhinBirgeneau} %16
2319: V. Kiryukhin, Y. J. Kim, K. J. Thomas, F. C. Chou, R. W. Erwin, Q. Huang,
2320: M. A. Kastner, and R. J. Birgeneau, Phys. Rev. B {\bf 63}, 144418 (2001).
2321:
2322: \bibitem{Tornow} %17
2323: S. Tornow, O. Entin-Wohlman, and A. Aharony, Phys. Rev. B {\bf 60}, 10206
2324: (1999).
2325:
2326: \bibitem{CONV} %18
2327: Our convention for wavevectors (in r. l. u.) is that the $x$-component is
2328: measured in units of $2\pi /a$, the $y$-component in units of
2329: $2 \pi/b$ and the $z$ component in units of $2\pi/c$.
2330: Lower case Roman letters (except $q$) denote integer components,
2331: whereas $q$ and $Q_x$ denote general noninteger values.
2332: In expressions such as $\exp(i {\bf Q} \cdot {\bf r})$,
2333: ${\bf Q}$ and ${\bf r}$ must be taken in compatible units
2334:
2335: \bibitem{Heine} %19
2336: V. Heine, in {\em Group Theory in Quantum Mechanics} (Dover Publications, New
2337: York, 1993), pp.\ 119, 288.
2338:
2339: \bibitem{IED} %20
2340: I. E. Dzyaloshinskii, Sov. Phys. JETP {\bf 5}, 1259 (1957).
2341:
2342: \bibitem{LL} %21
2343: L. D. Landau and E. M. Lifshitz, {\it Statistical Physics}
2344: (Pergamon, London, 1958).
2345:
2346: \bibitem{Nagamiya} %22
2347: T. Nagamiya, in {\em Solid State Physics}, edited by F. Seitz and D. Turnbull
2348: (Academic, New York, 1967), Vol.~29, p.\ 346.
2349:
2350: \bibitem{Kaplan} %23
2351: T.~A. Kaplan, Phys. Rev. {\bf 124}, 329 (1961).
2352:
2353: \bibitem{Aubry} %24
2354: S. Aubry, in {\em Solitons and Condensed Matter Physics}, edited by A. Bishop
2355: and T. Schneider (Springer Verlag, Berlin, 1978), p.\ 264.
2356:
2357: \bibitem{ANNNI} %25
2358: M.~E. Fisher and W. Selke, Phys. Rev. Lett. {\bf 44}, 1502 (1980).
2359:
2360: \bibitem{TCP} %28
2361: A. D. Bruce and A. Aharony, Phys. Rev. B {\bf 11}, 478 (1975).
2362:
2363: \bibitem{YildirimHarris} %29
2364: T. Yildirim, A.~B. Harris, O. Entin-Wohlman, and A. Aharony, Phys. Rev. Lett.
2365: {\bf 72}, 3710 (1994).
2366:
2367: \bibitem{SachidanandamYildirim} %30
2368: R. Sachidanandam, T. Yildirim, A. B. Harris, A. Aharony, and O. Entin-Wohlman,
2369: Phys. Rev. B {\bf 56}, 260 (1997).
2370:
2371: \bibitem{kastner} %31
2372: M. A. Kastner, A. Aharony, R. J. Birgeneau, F. C. Chou, O. Entin-Wohlman,
2373: M. Greven, A. B. Harris, Y. J. Kim, Y. S. Lee, M. E. Parks, and
2374: Q. Zhu, Phys. Rev. B {\bf 59}, 14702 (1999).
2375:
2376: %\bibitem{EFS}
2377: %E. F. Shender, Sov. Phys. JETP {\bf 56}, 178 (1982).
2378: %
2379: %\bibitem{Kittel}
2380: %C. Kittel, {\em Introduction to Solid State Physics} (John Wiley \& Sons, Inc.,
2381: % New York, 1986).
2382:
2383: \bibitem{BLEAN} %32
2384: % A. Abragam and B. Bleaney, {\it Electron Paramagnetic Resonance of
2385: % Transition Ions}, (Oxford U. P., Oxford, 1970). Reprinted New York:
2386: % Dover, 1986.
2387: K. D. Bowers and J. Owen, Rep. Prg. Phys. {\bf 18}, 304 (1955).
2388:
2389: \bibitem{Gavin} %33
2390: G. Lawes, unpublished.
2391:
2392: \bibitem{Lovesey_book} %34
2393: S.~W. Lovesey, {\em Theory of Neutron Scattering from Condensed Matter}
2394: (Clarendon Press, Oxford, 1984).
2395:
2396: \bibitem{IntTables} %35
2397: A.~J.~C. Wilson, {\em International Tables For Crystallography} (Kluwer
2398: Academic Publishers, Dordrecht, 1995), Vol.~C.
2399:
2400: \bibitem{Cooper_Nathans} %36
2401: M.~J. Cooper and R. Nathans, Acta Crys. {\bf 23}, 357 (1967).
2402:
2403: \bibitem{Squires} %37
2404: G.~L. Squires, {\em Thermal Neutron Scattering} (Cambridge University Press,
2405: Cambridge, 1978).
2406:
2407: \bibitem{PRL}
2408: Our PRL.
2409:
2410: \bibitem{Nagamiya}
2411: Nagamiya.
2412:
2413: \bibitem{SECOND}
2414: Second harmonic generation.
2415:
2416:
2417: \end{thebibliography}
2418:
2419:
2420: \end{document}
2421:
2422:
2423: Table IV:
2424: "are also given." The acoustic B_{2u}$ mode is
2425:
2426:
2427:
2428: