1: \documentclass[prb,twocolumn,showpacs,amsmath,amssymb,superscriptaddress,preprintnumbers]{revtex4}
2: \usepackage{graphicx}% Include figure files
3: \usepackage{dcolumn} % Align table columns on decimal point
4: \usepackage{bm}
5: \usepackage{epsfig}
6:
7:
8: \begin{document}
9:
10: \title{Optical properties and magnetic field-induced phase transitions
11: in the ferroelectric state of Ni$_3$V$_2$O$_8$}
12:
13: \author{{R. C. Rai}}
14: \affiliation{Department of Chemistry,
15: University of Tennessee, Knoxville, TN 37996}
16: \author{{J. Cao}}
17: \affiliation{Department of Chemistry, University of Tennessee,
18: Knoxville, TN 37996}
19: \author{{S. Brown}}
20: \affiliation{Department of Chemistry, University of Tennessee,
21: Knoxville, TN 37996}
22: \author{J. L. Musfeldt}
23: \email{musfeldt@utk.edu}\affiliation{Department of Chemistry, University of Tennessee,
24: Knoxville, TN 37996}
25: \author{D. Kasinathan}
26: \affiliation{Department of Physics, University of California Davis,
27: Davis, CA 95616}
28: \author{{D. J. Singh}}
29: \affiliation{Materials Science and Technology Division, Oak Ridge
30: National Laboratory, Oak Ridge,TN 37831-6032}
31: \author{G. Lawes}
32: \affiliation{Department of Physics, Wayne State University, Detroit,
33: MI 48201}
34: \author{N. Rogado}
35: \affiliation{DuPont Central Research and Development, Experimental
36: Station, Wilmington, DE 19880-0328}
37: \author{R. J. Cava}
38: \affiliation{Department of Chemistry and Princeton Materials
39: Institute, Princeton University, Princeton, NJ 08544}
40: \author{{X. Wei}}
41: \affiliation{National High Magnetic Field Laboratory, Florida State
42: University, Tallahassee, Florida 32310}
43:
44: \pacs{75.80.+q, 78.20.Ls, 71.20.Be,75.30.Et}
45:
46:
47: \begin{abstract}
48: We use a combination of optical spectra, first principles
49: calculations, and energy dependent magneto-optical measurements to
50: elucidate the electronic structure and to study the phase diagram of
51: Ni$_3$V$_2$O$_8$. We find a remarkable interplay of magnetic field
52: and optical properties that reveals additional high magnetic field
53: phases and an unexpected electronic structure which we associate
54: with the strong magneto-dielectric couplings in this material over a
55: wide energy range. Specifically, we observed several prominent
56: magneto-dielectric effects that derive from changes in crystal field
57: environment around Ni spine and cross-tie centers. This effect is
58: consistent with a field-induced modification of local structure.
59: Symmetry-breaking effects are also evident with temperature. We find
60: Ni$_3$V$_2$O$_8$ to be an intermediate gap, local moment band
61: insulator. This electronic structure is particularly favorable for
62: magneto-dielectric couplings, because the material is not subject to
63: the spin charge separation characteristic of strongly correlated
64: large gap Mott insulators, while at the same time remaining a
65: magnetic insulator independent of the particular spin order and
66: temperature.
67: \end{abstract}
68:
69: \maketitle
70:
71:
72: \section{INTRODUCTION}
73:
74:
75: Ni$_3$V$_2$O$_8$ is a particularly interesting magnetic material,
76: \cite{sauerbrei73,rogado02,lawes05,lawes04,kenzelmann06} both
77: because of its unusual structure, which provides an example of a
78: spin 1 system on a Kagom$\acute{e}$ staircase lattice, and because
79: of the rich variety of magnetic and structural phases that are
80: stabilized under different conditions. One especially interesting
81: feature is the occurrence of a magnetic, ferroelectric phase as a
82: function of temperature and magnetic field. More generally, coupled
83: magnetic and electric degrees of freedom, flexible lattices, and
84: magnetic frustration in multiferroics can result in cascades of
85: coupled magnetic and dielectric transitions.
86: \cite{cheong04,cruz05,lorenz04,tgoto04,hur04,lawes03,rogado05,saito05}
87: The recent report \cite{hur04} of colossal low frequency (1 kHz)
88: magneto-dielectric effects in inhomogeneously mixed-valent
89: DyMn$_2$O$_5$ is especially important, because it illustrates that
90: sizable dielectric contrast can be achieved by physical tuning
91: through an unusual commensurate-incommensurate magnetic transition
92: and is facilitated by a soft lattice. The 300 K low-frequency
93: magneto-dielectric effect in mixed-valent LuFe$_2$O$_4$ has also
94: attracted attention due to the very low magnetic fields needed to
95: achieve dielectric contrast.\cite{subra06} Ni$_3$V$_2$O$_8$ is
96: another system where the temperature and field dependence of the
97: spontaneous polarization shows a strong coupling between magnetic
98: and ferroelectric order. \cite{lawes05,lawes04,kenzelmann06} This
99: coexistence is unusual and appears only when certain symmetry
100: conditions are fulfilled. \cite{lawes05} That the effect can be
101: controlled with an external magnetic field makes it attractive for
102: device applications. Based upon our previous work with
103: inhomogeneously mixed-valent K$_2$V$_3$O$_8$,\cite{rai06} the
104: significant coupling between spin, lattice, and charge degrees of
105: freedom make Ni$_3$V$_2$O$_8$ an excellent candidate for discovery
106: of higher energy magneto-dielectric effects.
107:
108:
109: \begin{figure}[b]
110: \centerline{
111: \includegraphics[width=3.6in]{Structure.eps}}
112: \vskip -0.2cm \caption{(Color online) (a) Crystal structure of
113: Ni$_3$V$_2$O$_8$. (b) View of Kagom$\acute{e}$ staircase showing
114: only the Ni atoms.}\label{struct}
115: \end{figure}
116:
117:
118:
119: Figure \ref{struct}(a) shows the orthorhombic (Cmca) crystal
120: structure of Ni$_3$V$_2$O$_8$. It consists of Kagom$\acute{e}$
121: layers of edge sharing NiO$_6$ octahedra separated by nonmagnetic
122: VO$_4$ tetrahedra. Ni$_3$V$_2$O$_8$ is considered to be a
123: Kagom$\acute{e}$ staircase compound due to buckling of the lattice
124: perpendicular to the $a$ axis. There are two distinct types of
125: Ni$^{2+}$ ($S$ = 1) centers, which we refer to as ``spine" and
126: ``cross-tie" sites. The ``spine" sites run along the $a$ axis. A
127: view of the Kagom$\acute{e}$ staircase showing only the Ni atoms is
128: displayed in Fig. \ref{struct}(b). Note that the spine and cross-tie
129: sites have very different local symmetries. The spin ordering
130: arrangements in Ni$_3$V$_2$O$_8$ have been extensively investigated
131: by neutron scattering and derive from various local and long range
132: exchange, spin anisotropy, Dzyaloshinskii-Moriya, dipolar, and
133: frustration effects. \cite{kenzelmann06} The $H$-$T$ phase diagram
134: (for $H$$\parallel$$b$) is shown in Fig. \ref{PhaseDia}(b). Two
135: incommensurate phases are observed below the paramagnetic phase
136: (PM). HTI is the (longitudinal) high temperature incommensurate
137: phase, and LTI is the (spiral) low temperature incommensurate phase.
138: The latter displays ferroelectricity. Commensurate canted
139: antiferromagnetism is observed in the C phase.
140:
141: \begin{figure}[h]
142: \centerline{
143: \includegraphics[width=3.0in]{PhaseDia.eps}}
144: \vskip -0.2cm \caption{(Color online) (a) Specific heat of
145: Ni$_3$V$_2$O$_8$ as a function of temperature at $H$ = 8 T. The
146: arrow indicates a proposed phase transition temperature. (b) $H$-$T$
147: phase diagram of Ni$_3$V$_2$O$_8$ for $H$$\parallel$$b$ with
148: additional data points extracted from specific heat
149: (solid triangles) and magneto-optical measurements (solid
150: circles). Solid lines are discussed in the text and the dashed line
151: is a guide to the eye. A new high field (HF) phase is also
152: indicated.}\label{PhaseDia}
153: \end{figure}
154:
155:
156:
157: Here we use a combination of optical spectroscopy, first principles
158: calculations, and energy dependent magneto-optical measurements to
159: elucidate the electronic structure and to study the phase diagram of
160: Ni$_3$V$_2$O$_8$. We find a remarkable interplay of magnetic field
161: and optical properties that reveals additional high magnetic field
162: phases and an unexpected electronic structure which we associate
163: with the strong magneto-dielectric couplings in this material over a
164: wide energy range. Specifically, we observed several prominent
165: magneto-dielectric effects that derive from changes in crystal field
166: environment around Ni spine and cross-tie centers. This effect is
167: consistent with a field-induced modification of local structure.
168: Symmetry-breaking effects are also evident with temperature. Even
169: though both Ni$_3$V$_2$O$_8$ and the prototypical Mott insulator NiO
170: are based on Ni$^{2+}$ ions octahedrally coordinated by O with
171: similar bond lengths, we show that the basic electronic structures
172: of these two materials are very different. NiO has a large band gap
173: and local moments that derive from strong Coulomb interactions and
174: separation of spin and charge degrees of freedom. This separation is
175: theoretically interesting but disadvantageous if one wants to
176: promote coupling effects. The Slater insulator is at the opposite
177: extreme. Here, a small band gap is present in the ground state, but
178: it is a direct result of a specific spin ordering. In contrast, we
179: find Ni$_3$V$_2$O$_8$ to be an intermediate gap, local moment band
180: insulator. This electronic structure is particularly favorable for
181: magneto-dielectric couplings, because the material is not subject to
182: the spin charge separation characteristic of strongly correlated
183: large gap Mott insulators, while at the same time remaining a
184: magnetic insulator independent of the particular spin order and
185: temperature.
186:
187:
188:
189: \section{Methods}
190:
191: \subsection{Crystal Growth}
192:
193: The Ni$_3$V$_2$O$_8$ single crystal samples were prepared from a
194: BaO-V$_2$O$_5$ flux. The crystals used for these measurements were
195: grown as platelets, with their largest faces (a few square
196: millimeters) perpendicular to the crystallographic $b$ axis.
197:
198: \subsection{Spectroscopic Investigations}
199:
200:
201: Near normal $ac$-plane reflectance of Ni$_3$V$_2$O$_8$ was measured
202: over a wide energy range (3.7 meV - 6.5 eV) using several different
203: spectrometers including a Bruker 113 V Fourier transform infrared
204: spectrometer, a Bruker Equinox 55 Fourier transform infrared
205: spectrometer equipped with an infrared microscope, and a Perkin
206: Elmer Lambda 900 grating spectrometer, as described previously.
207: \cite{zhu02} The spectral resolution was 2 cm$^{-1}$ in the far and
208: middle-infrared and 2 nm in the near-infrared, visible, and
209: near-ultraviolet. Optical conductivity was calculated by a
210: Kramers-Kronig analysis of the measured reflectance.\cite{wooten72}
211: An open flow cryostat provided
212: temperature control.
213:
214: The magneto-optical properties of Ni$_3$V$_2$O$_8$ were investigated
215: between 0.75 and 4.1 eV using a 3/4 m grating spectrometer equipped
216: with InGaAs and CCD detectors and a 33 T resistive magnet at the
217: National High Magnetic Field Laboratory (NHMFL), in Tallahassee, FL.
218: 150, 300, and 600 line/mm gratings were used, as appropriate.
219: Experiments were performed with polarized light ($E$$\parallel$$a$
220: and $E$$\parallel$$c$) in the temperature range between 4.2 and 15 K
221: for applied magnetic fields up to 30 T ($H$$\parallel$$b$). With
222: light shining on the sample, we estimate that the base temperature
223: was $\sim$5.0 K. The field-induced changes in the measured
224: reflectance were studied by taking the ratio of the reflectance at
225: each field with the reflectance at zero field, i.e., [R($H$)/R($H$ =
226: 0 T)]. This normalized response is a sensitive way to view the
227: field-induced optical changes.\cite{cao06} Since
228: $\epsilon(E)=\epsilon_1(E) + i\epsilon_2(E)
229: =\epsilon_1(E) + {{4{\pi}i}\over {\omega}}\sigma_1(E)$, it
230: is clear that the field-induced changes in reflectance translate
231: into finite energy magneto-dielectric effects. To obtain the 30 T
232: optical conductivity ($\sigma_1$) and dielectric response
233: ($\epsilon_1$), we renormalized the zero-field absolute reflectance
234: with the high-field reflectance ratios, and recalculated
235: $\sigma$$_1$ and $\epsilon_1$ using Kramers-Kronig
236: techniques.\cite{wooten72,rai06}
237:
238: \subsection{Electronic Structure Calculations}
239:
240:
241: The calculations were done with the experimental crystal structure
242: \cite{sauerbrei73} using the general potential linearized augmented
243: planewave (LAPW) method, with local orbitals, \cite{lapw,lo} as
244: implemented in the WIEN2K program. \cite{wien} The augmented
245: planewave plus local orbital extension was used for the O $s$ and
246: $p$ states, the metal $d$ states and the semi-core levels.
247: \cite{apw} The valence states were treated in a scalar relativistic
248: approximation, while the core states were treated relativistically.
249: The calculations were done self-consistently using well converged
250: basis sets and zone samplings based on 144 {\bf k}-points in the
251: irreducible wedge. A more dense mesh of 1200 {\bf k}-points was used
252: for the calculations of the optical conductivity. The LAPW sphere
253: radii were 1.8, 1.7 and 1.5 $a_0$ for Ni, V and O, respectively.
254: Calculations were done in the local spin density approximation
255: (LSDA) and the LDA+U method. For the LDA+U calculations, we used
256: values of $U$ ranging from 5 eV to 7 eV on the Ni sites. The results
257: shown are for $U$ = 5 eV. \cite{u-note}
258:
259:
260:
261:
262: \section{Results and Discussion}
263:
264: \subsection{Optical and Electronic Properties of Ni$_3$V$_2$O$_8$}
265:
266:
267: Figure \ref{OpConducH0}(a) displays the polarized optical
268: conductivity of Ni$_3$V$_2$O$_8$ in the PM phase at 300 and 12 K.
269: The spectra show directionally dependent vibrational and electronic
270: excitations with an optical energy gap of $\sim$0.35 eV. Based on
271: our electronic structure calculations (discussed in detail below)
272: and comparison with chemically similar Ni-containing
273: compounds,\cite{sawatzky,schnack06,ido91,juza66} we assign the
274: excitations centered near $\sim$0.75 and $\sim$1.35 eV to Ni $d$ to
275: $d$ on-site excitations in the minority spin channel on cross-tie
276: and spine sites, respectively. These excitations are optically
277: allowed due to the modest hybridization between the Ni $d$ and O $p$
278: states. We assign the $\sim$3.0 eV feature in the optical
279: conductivity spectrum to a combination of O 2$p$ to Ni 3$d$ and O
280: 2$p$ to V 3$d$ charge transfer excitations. The $\sim$4.4 eV feature
281: derives from O 2$p$ to V 3$d$ charge transfer excitations.
282:
283: \begin{figure}[t]
284: \centerline{
285: \includegraphics[width=3.2in]{OpConducH0.EPS}}
286: \vskip -0.2cm \caption{(Color online) (a) Polarized optical
287: conductivity of Ni$_3$V$_2$O$_8$ at 300 and 12 K, extracted from
288: reflectance measurements (inset) by a Kramers-Kronig analysis. The
289: energy range of our magneto-optical measurements is indicated by the
290: arrow (0.75 - 4.1 eV). (b) A close-up view of the $E$$\parallel$$a$
291: optical conductivity at 12, 8, and 5 K, corresponding to Ni (spine)
292: $d$ to $d$ on-site excitations. Inset shows a close-up view of Ni
293: (cross-tie) $d$ to $d$ on-site excitations at $T$ = 300, 100, 50,
294: and 12 K, respectively. These curves are offset for clarity.
295: }\label{OpConducH0}
296: \end{figure}
297:
298: Transition metal $d$ to $d$ on-site excitations are sensitive
299: indicators of the local crystal field environment. Figure
300: \ref{OpConducH0}(b) shows a close-up view of the Ni (spine) $d$ to
301: $d$ on-site excitations in the $E$$\parallel$$a$ optical
302: conductivity at 12, 8, and 5 K. This structure is broad and
303: featureless in the PM phase (12 K), but it splits into at least four
304: different components at 8 K (HTI phase). The splitting becomes more
305: pronounced at 5 K (LTI phase). This splitting is due to a local
306: structure distortion around the Ni center with decreasing
307: temperature and demonstrates that the NiO$_6$ environment is
308: different in the PM, HTI, and LTI phases. Although it is at the
309: limit of our detector range, the trailing edge of the $\sim$0.75
310: feature also displays splitting below 12 K (not shown).\cite{note1}
311: Interestingly, the Ni (cross-tie) $d$ to $d$ on-site excitation also
312: shows a notch-like structure that develops between 100 and 50 K
313: (inset of Fig. \ref{OpConducH0}(b)), indicating that a weak
314: structural distortion in the cross-tie direction precedes the
315: cascade of low temperature magnetic phases. Anomalous intermediate
316: temperature phonon shifts have also been reported in the
317: RMn$_2$O$_5$ (R = Bi, Eu, Dy) family of frustrated
318: multiferroics.\cite{flores06}
319:
320:
321: \subsection{Electronic Structure Calculations of Ni$_3$V$_2$O$_8$}
322:
323:
324:
325: As mentioned above, there are two nonequivalent Ni sites in the unit
326: cell, a spine site (4 atoms per cell, denoted Ni2 in the following)
327: and a cross-link site (2 atoms per cell, denoted Ni1). As mentioned,
328: Ni$_3$V$_2$O$_8$ is a local moment magnet with a complex magnetic
329: phase diagram. To assign the peaks in the optical spectrum, we did
330: calculations for two magnetic structures: (1) all spins on each Ni
331: sublattice aligned ferromagnetically, but with the two sublattices
332: opposite to each other (denoted FiM, in the following) and (2) a
333: ferromagnetic ordering (FM). In the LSDA, the FiM ordering had lower
334: energy by 7 meV per formula unit (note that there are two formula
335: units per cell). This shows an antiferromagnetic interaction between
336: the two sublattices.
337:
338: \begin{figure}[h]
339: \epsfig{width=0.85\columnwidth,file=totaldos_lda.eps} \vspace{.2cm}
340: \epsfig{width=0.85\columnwidth,file=projecteddos_lda.eps}
341: \vspace{0.2cm} \caption{(color online) Total (top) and projected DOS
342: (bottom) as obtained with the LSDA for the FiM ordering (see text).
343: The total DOS is on a per formula unit spin basis. The projections
344: are onto the LAPW spheres, and are given per atom. } \label{dos-lda}
345: \end{figure}
346:
347:
348: \begin{figure}[b]
349: \epsfig{width=0.85\columnwidth,file=totaldos_ldau.eps}
350: \vspace{0.2cm}
351: \epsfig{width=0.85\columnwidth,file=projecteddos_ldau.eps}
352: \vspace{0.2cm} \caption{(color online) Total (top) and projected DOS
353: (bottom), as in Fig. \ref{dos-lda}, but with the LDA+U method, $U$ =
354: 5 eV. } \label{dos-ldau}
355: \end{figure}
356:
357:
358: The calculated total and projected densities of states (DOS) for the
359: FiM ordering are shown in Fig. \ref{dos-lda} and Fig.
360: \ref{dos-ldau}, within the LSDA and the LDA+U (U = 5.0 eV)
361: approximation, respectively. In addition to the U = 5 eV DOS shown,
362: we did calculations for U = 6 eV and U = 7 eV (see below).
363: Insulating behavior is found in both approximations. The gaps are
364: $E_g({\rm LSDA})$ = 0.30 eV and $E_g({\rm LDA+U})$ = 1.90 eV. There
365: are only relatively small changes in the O $2p$ bands between these
366: two approximations. The O 2$p$ valence bands extend from $\sim$ -6.5
367: eV to the valence band edge in the LSDA, and from $\sim$ -6.0 eV to
368: the band edge in the LDA+U approximation. Wang and co-workers
369: \cite{wang} investigated Ni$_3$V$_2$O$_8$ as a potential
370: photocatalyst for water splitting using optical spectroscopy from
371: 300 nm to 850 nm, and density functional calculations. However, they
372: did not include spin-polarization in their band structure
373: calculations, and this is needed to obtain the exchange splitting of
374: the Ni $d$ states. As a result, they obtained a metallic electronic
375: structure with a high density of Ni $d$ states at the Fermi energy
376: in disagreement with experiment.
377:
378: The compound is described as Ni$^{2+}_3$V$^{5+}_2$O$^{2-}_8$. As
379: shown in the projected DOS, we find the V $d$ bands well above
380: valence bands in both the LSDA and LDA+U calculations (from $\sim$ 2
381: eV to $\sim$ 5 eV, relative to the valence band edge), in agreement
382: with this ionic model. In all calculations, we find integer spin
383: magnetizations of 2 $\mu_B$ per Ni ion, as expected for Ni$^{2+}$.
384: In the crystal field, the main Ni $d$ splitting is due to the local
385: coordination, which gives a lower 3-fold degenerate $t_{2g}$ and a
386: higher two-fold degenerate $e_g$ manifold per spin. The Ni $d$ bands
387: are already very narrow in the LSDA. This is the reason for the
388: clean gap. In the FM case, the band gap $E_g$ is reduced by only
389: 0.11 eV relative to the FiM case. Presumably, antiferromagnetism
390: within the individual sublattices would narrow the bands further and
391: increase $E_g$ relative to the FiM ordering. Thus, already at the
392: LSDA level, Ni$_3$V$_2$O$_8$ is a local moment insulator. The
393: majority spin Ni $d$ levels are centered at -1.7 eV ($t_{2g}$) and
394: -0.6 eV ($e_g$), for both Ni sites. The corresponding minority spin
395: levels are centered at -0.4 eV ($t_{2g}$) and 0.8 eV ($e_g$).
396:
397:
398: \begin{figure}
399: \epsfig{width=0.85\columnwidth,file=sigma_real_lda.eps}
400: \vspace{0.2cm}
401: \epsfig{width=0.85\columnwidth,file=sigma_real_ldau5.eps}
402: \vspace{0.2cm} \caption{Real part of the optical conductivity within
403: the LSDA (top) and LDA+U, $U$ = 5 eV (bottom), for Ni$_3$V$_2$O$_8$
404: with the FiM ordering. A 0.1 eV broadening was applied. Two
405: definitions of the crystallographic axes are in literature. We use
406: the one where $a$ = 5.936 \AA, $b$ = 11.420 \AA, and $c$ = 8.240
407: \AA.\cite{sauerbrei73}} \label{sigma}
408: \end{figure}
409:
410:
411: The main difference between the LSDA and LDA+U electronic structure
412: is in the position of the Ni $d$ bands. The majority spin Ni
413: manifolds are shifted deep into the valence band. These are centered
414: at $\sim$ -5.5 eV ($t_{2g}$) and $\sim$ -2.5 eV ($e_g$). The
415: minority spin states are split apart by the Coulomb interaction and
416: are now centered at $\sim$ -1 eV ($t_{2g}$) and $\sim$ 2.5 eV
417: ($e_g$). For this value of $U$ the conduction band edge onset
418: derives from both V $d$ and Ni $d$ states. Calculations with $U$ = 6
419: eV and $U$ = 7 eV increase the differences from the LSDA, pushing
420: the Ni $d$ conduction band onset above the V $d$ onset and driving
421: the occupied Ni $d$ manifolds deeper into the O $p$ bands.
422:
423: The calculated LSDA and LDA+U optical spectra for FiM ordering are
424: shown in Fig. \ref{sigma}. Besides the larger LDA+U band gap, the
425: shapes of the spectra are very different. In the LSDA the onset has
426: both Ni $d$ - Ni $d$ and charge transfer character. The first
427: structure just above 1 eV is due to the structure in the Ni $d$ to
428: Ni $d$ excitations. It contains two peaks. The Ni1 (cross-link) site
429: has a lower primary crystal field splitting than the Ni2 (spine)
430: site, as may be seen from the projected DOS. Therefore, the lower
431: energy peak comes more from the Ni1 and the higher peak from the
432: Ni2. The second main structure above 2 eV, which is weaker, is of
433: mainly charge transfer character involving Ni and V, and the onset
434: starting at 3 eV is mainly of charge transfer character into the V
435: $d$ bands. In the LDA+U case the spectrum starts at $\sim$ 2 eV and
436: is dominated throughout by charge transfer excitations.
437:
438: The LSDA spectrum is clearly in better accord with the experimental
439: spectrum than the LDA+U spectrum, both in terms of the gap and in
440: terms of the peak structure. The main differences between LSDA and
441: experiment are in the region of the second peak (above 2 eV) and in
442: the onset of the O $2p$ to V $d$ charge transfer excitations.
443: Considering that V is in a $d^0$ configuration, the underestimate of
444: this gap is not surprising and presumably reflects just the generic
445: LDA band gap underestimation, which is seen in other $d^0$ oxides.
446: The relatively good agreement of the LSDA Ni $d$ position and the
447: fundamental gap with experiment is, however, unexpected and deserves
448: comment.
449:
450:
451: NiO, which like Ni$_3$V$_2$O$_8$, has Ni$^{2+}$, six-fold
452: coordinated by O, is a prototypical Mott insulator. LSDA
453: calculations for antiferromagnetic NiO show a small band gap and
454: features in the valence band dispersions in accord with experiment.
455: \cite{terakura} However, the gap obtained is much smaller than the
456: experimental gap of $\sim$ 4 eV, and furthermore it is dependent on
457: the specific magnetic ordering. Furthermore, the LSDA gap of NiO is
458: of incorrect $d$-$d$ character rather than the charge transfer
459: character observed in experiment. \cite{sawatzky,merlin} This shows
460: an essential role for beyond LSDA Coulomb correlations in describing
461: the electronic structure of NiO. Various methods, including the
462: LDA+U approximation, for incorporating these have been developed and
463: tested using NiO.
464: \cite{norman,anisimov-1,anisimov-2,svanne,arai,kod,aryasetiawan,massidda,li}
465: As mentioned, NiO and Ni$_3$V$_2$O$_8$ both have Ni$^{2+}$ in
466: approximately octahedral O cages. In NiO, the Ni band width is
467: larger than in Ni$_3$V$_2$O$_8$, which explains why the LSDA is able
468: to produce local moment insulating behavior in this compound but not
469: in NiO. The narrower bands in Ni$_3$V$_2$O$_8$ are related to the
470: bonding topology. In NiO, the O are six fold coordinated with
471: 90$^\circ$ and straight 180$^\circ$ bonds. Ni$_3$V$_2$O$_8$ has
472: lower O coordination and bent bonds. In metals, the criterion for
473: dividing strongly correlated materials from band metals involves
474: $U_s/W$, where $W$ is the band width and $U_s$ is the screened
475: Coulomb potential, where the ``s" is simply to distinguish this
476: parameter from the $U$ used in the LDA+U approximation. In general,
477: Mott insulators derived from metals with $U_s/W$ cannot be described
478: in the LSDA and require the use of the LDA+U approximation or more
479: sophisticated approaches, while materials with weaker correlations
480: are often better described in LSDA than in LDA+U calculation. Since
481: the band width of Ni$_3$V$_2$O$_8$ is not larger than that of NiO,
482: the implication of our results is that the effective screened
483: Coulomb interaction $U_s$ on at least one of the Ni sites is
484: sufficiently lower than in NiO to cross-over from strongly
485: correlated Mott insulator to band insulator. We speculate that this
486: could be due to the lower cation coordination of O$^{2-}$ in
487: Ni$_3$V$_2$O$_8$. Specifically, in NiO, six Ni atoms compete for
488: bonding with each O (two per O $p$ orbital). The average Ni-O bond
489: length in Ni$_3$V$_2$O$_8$ is $\sim$ 0.03 \AA$~$ shorter than in
490: NiO, consistent with this from the bond valence point of view.
491: \cite{brown} This is not the case in Ni$_3$V$_2$O$_8$, which may
492: enable the O cage to more effectively screen the bare $U$. As
493: mentioned, we find Ni$_3$V$_2$O$_8$ to be a local moment insulator
494: with weak sensitivity of the band structure to the magnetic order at
495: the LDA level.
496:
497:
498:
499: \begin{figure}[h]
500: \epsfig{width=0.85\columnwidth,file=fm_lda_sigma_real.eps}
501: \vspace{0.2cm} \caption{Real part of the optical conductivity within
502: the LSDA, as in Fig. \ref{sigma}, but with FM ordering.}
503: \label{sigma-fm}
504: \end{figure}
505:
506:
507:
508: Figure \ref{sigma-fm} shows the calculated optical spectra for FM
509: ordering. As may be seen it is qualitatively similar in structure to
510: that of FiM ordering, but shows noticeable quantitative differences.
511: Most notably there are changes in the shape of the low energy Ni $d$
512: - $d$ peak, a suppression of the second peak and changes in the
513: shape of the onset of the higher energy charge transfer onset. Since
514: neither the FiM or the FM ordering are the ground state ordering,
515: these differences should be viewed as indicating the errors on our
516: calculation when comparing with the experimental zero field
517: spectrum. They also support the experimental finding of substantial
518: magnetochromic effects in this material. Interestingly, we find
519: considerable orientation dependence of the magnetochromic effects in
520: this material.
521:
522: Optical excitations are charge excitations. Since Hubbard
523: correlations generally harden charge degrees of freedom leading to
524: spin charge separation one may expect stronger charge spin coupling,
525: and perhaps stronger magnetochromic effects in less correlated
526: materials. This is also relevant to the magnetoelectric coupling
527: observed in Ni$_3$V$_2$O$_8$, since strong Coulomb correlations, if
528: present, would reduce the spin-lattice coupling, so that while
529: ferroelectricity would still be induced by the incommensurate
530: magnetic order, it would be weaker due to reduced coupling. Based on
531: our results, we find that the Ni sites in Ni$_3$V$_2$O$_8$ are not
532: subject to strong correlations, and this underlies the substantial
533: magnetochromic effects seen in this material.
534:
535:
536:
537:
538: \subsection{Magneto-Optical Properties of Ni$_3$V$_2$O$_8$}
539:
540:
541:
542: Figure \ref{ReflecRatio}(a) and (b) shows the reflectance ratio of
543: Ni$_3$V$_2$O$_8$, R($H$)/R($H$=0 T), at 5 K. Rich field-induced
544: changes in reflectance are observed over the full energy range in
545: both polarizations. The most significant modifications are near 0.75
546: and 1.35 eV, providing a preview of the physical origin of the
547: field-induced effects. Based on the aforementioned positions of the
548: magneto-optical features, we attribute these changes to
549: field-induced modifications of Ni (cross-tie) $d$ to $d$ and Ni
550: (spine) $d$ to $d$ on-site excitations. At 30 T, the reflectance
551: deviates from unity by $\pm$15\% near 1.35 eV. On the other hand,
552: reflectance changes in the visible range (centered at $\sim$2.2,
553: 2.6, and 3.3 eV) are much broader and more modest in size (up to
554: $\sim$4\% for 30 T at 5 K), similar in magnitude to ``magnetochromic
555: effects" reported on other complex
556: materials.\cite{rai06,schnack06,woodward05,choi04} These features
557: can be attributed to field-induced changes in O 2$p$ to Ni 3$d$ and
558: O 2$p$ to V 3$d$ charge transfer excitations.
559:
560:
561: \begin{figure}[t]
562: \centerline{
563: \includegraphics[width=3.2in]{ReflecRatioEa.EPS}}
564: \vskip -1.6cm \centerline{
565: \includegraphics[width=3.2in]{ReflecRatioEc.EPS}}
566: \vskip -0.4cm \caption{(Color online) The normalized magneto-optical
567: response, R($H$)/R($H$=0 T), of Ni$_3$V$_2$O$_8$ in an applied
568: magnetic field ($H$$\parallel$$b$) from 0 to 30 T at 5 K (a) for
569: light polarized along the $a$ direction. (b) for light polarized
570: along the $c$ direction. The insets show close-up views of the
571: magneto-optical response near the Ni (cross-tie) and Ni (spine) $d$
572: to $d$ on-site excitations. Data were taken in 2 T steps, but only
573: representative curves are displayed for clarity.}\label{ReflecRatio}
574: \end{figure}
575:
576:
577: \begin{figure}[h]
578: \centerline{
579: \includegraphics[width=3.2in]{OpConducH0T30T.EPS}}
580: \vskip -0.4cm \caption{(Color online) Polarized optical conductivity
581: of Ni$_3$V$_2$O$_8$ at 5 K for $H$ = 0 and 30 T ($H$$\parallel$$b$).
582: $E$$\parallel$$a$ and $E$$\parallel$$c$ polarizations are indicated
583: by dashed and solid lines, respectively. The upper inset shows a
584: close-up view of the energy region near the Ni (spine) $d$ to $d$
585: on-site excitation for $H$ = 0 and 30 T. The lower inset shows a
586: close-up view of the Ni (spine) $d$ to $d$ on-site excitation for
587: $T$ = 12 and 5 K at 30 T.}\label{OpConducH30T}
588: \end{figure}
589:
590:
591:
592:
593: \begin{figure}[t]
594: \centerline{
595: \includegraphics[width=3.2in]{Dielec5K.EPS}}
596: \vskip -0.2cm \caption{ (Color online) (a) Dielectric constant of
597: Ni$_3$V$_2$O$_8$ at 5 K for light polarized along the $a$ direction
598: for $H$ = 0 and 30 T ($H$$\parallel$$b$). The inset shows a close-up
599: view of the energy region near the Ni (spine) $d$ to $d$ on-site
600: excitation. Close-up views of the high-energy dielectric contrast,
601: $\Delta\epsilon_1/\epsilon_1$ = $[{\epsilon_1}(E,H) -
602: {\epsilon_1}(E,0)]/{\epsilon_1}(E,0)$, near (b) the Ni (spine) $d$
603: to $d$ on-site excitation and (c) the charge transfer
604: excitations.}\label{Dielec5K}
605: \end{figure}
606:
607:
608: It is desirable to correlate field-induced reflectance ratio changes
609: with the dispersive and lossy response of the material. Figure
610: \ref{OpConducH30T} displays the polarized optical conductivity of
611: Ni$_3$V$_2$O$_8$ at 5 K at 0 and 30 T ($H$$\parallel$$b$), extracted
612: from a combination of absolute reflectance measurements, the
613: reflectance ratio data of Fig. \ref{ReflecRatio}(a) and (b), and a
614: Kramers-Kronig analysis and calculation of the optical constants.
615: From the optical conductivity, we can immediately confirm that the
616: aforementioned magneto-optical effects correspond to field-induced
617: modifications of the Ni (cross-tie) $d$ to $d$, Ni (spine) $d$ to
618: $d$ on-site excitations, and O 2$p$ to Ni 3$d$ and O 2$p$ to V 3$d$
619: charge transfer excitations. The insets of Fig. \ref{OpConducH30T}
620: show close-up views of the $\sim$1.35 eV Ni on-site excitation
621: splittings that point toward a field-induced local distortion of the
622: NiO$_6$ octahedra.\cite{note1} As mentioned previously, these Ni $d$
623: to $d$ excitations are allowed because oxygen hybridization modifies
624: the matrix elements. The charge transfer excitation centered at
625: $\sim$3.0 eV also broadens and blueshifts with applied field, and a
626: small ``notch" develops at $\sim$ 2.2 eV. The reflectance ratio
627: trends discussed above also capture important dispersive effects.
628: Figure \ref{Dielec5K}(a) shows the high energy magneto-dielectric
629: response of Ni$_3$V$_2$O$_8$ at 5 K for $E$$\parallel$$a$. Although
630: dielectric contrast ($\Delta\epsilon_1/\epsilon_1$) is subtle in the
631: majority of the visible region (Fig. \ref{Dielec5K}(c)), the applied
632: field changes the dielectric constant significantly near the Ni $d$
633: to $d$ on-site excitations (inset, Fig. \ref{Dielec5K}(a), and Fig.
634: \ref{Dielec5K}(b)). At some
635: energies, dielectric contrast is on the order of 15\%
636: (Fig. \ref{Dielec5K}(b)). These results demonstrate that
637: magneto-dielectric effects are not confined to kHz frequencies and
638: that the response may be tunable. Both aspects
639: may be useful for
640: device applications (for instance, magnetically controlled ferroelectric memory). This remarkable interplay between the magnetic
641: field and optical constants of Ni$_3$V$_2$O$_8$ seems to be
642: facilitated by the
643: fact that this material is in the intermediate
644: coupling regime, as discussed below.
645:
646:
647:
648: Comparison of the reflectance ratio data in the upper insets of Fig.
649: \ref{ReflecRatio}(a) and (b) provides insight into possible
650: magnetoeleastic coupling in Ni$_3$V$_2$O$_8$. The structure centered
651: at $\sim$0.83 eV is large in the $c$ (cross-tie) direction, whereas
652: it is much smaller along the $a$ (spine) direction. It grows quickly
653: with applied field. The splitting of this structure ($\sim$12 meV
654: (97 cm$^{-1}$)) corresponds to a vibrational energy scale.
655: Candidates for possible coupling might therefore include the
656: displacement of Ni (cross-tie) atoms or a low-frequency O-Ni-O
657: bending mode. In the presence of a soft lattice, magnetoelastic
658: coupling may lead to a local distortion of the NiO$_6$ octahedra.
659: Note that the PM phase of a frustrated magnetic system often
660: displays significant spin correlations.\cite{flores06} The PM phase
661: in Ni$_3$V$_2$O$_8$ may be similar, consistent with the presence of
662: magnetoelastic coupling. Direct measurements of high field
663: vibrational properties of Ni$_3$V$_2$O$_8$ are clearly desirable. We
664: note that spin-lattice coupling leads to structural distortions in
665: the frustrated magnetic system ZnCr$_2$O$_4$.\cite{sushkov05,lee02}
666: Magnetoelastic coupling also plays an important role in
667: geometrically frustrated magnetic systems such as HoMnO$_3$ and
668: RMn$_2$O$_5$ (R = Tb, Ho, Dy).
669: \cite{cruz05,lorenz04,hur04,flores06,ratclif05,blake05}
670:
671:
672:
673: Spin-charge coupling in Ni$_3$V$_2$O$_8$ can be further investigated
674: by quantifying the subtle changes in the Ni (spine) $d$ to $d$
675: on-site excitation with applied field. Figure \ref{SpinePeakEa5K}
676: shows a close-up view of polarized optical conductivity of
677: Ni$_3$V$_2$O$_8$ at 5 K for $H$ = 0, 10, 20, and 30 T
678: ($H$$\parallel$$b$). The effect of applied field is non-monotonic
679: and correlates with changes in magnetic order. Peak fits, using four
680: model oscillators, were used to elucidate these trends (inset, Fig.
681: \ref{SpinePeakEa5K}). The dependence of peak positions on the
682: applied field clearly demonstrates an interplay between electronic
683: and magnetic properties. For instance, at 5 K, the data shows
684: discontinuities near 7 and 15 T, indicated by the grey shaded
685: regions in the inset. These transition regions move to higher fields
686: with increasing temperature. We note that the low field phase
687: boundaries determined in this manner are in good overall agreement
688: with the previously reported thermodynamic data. This optical
689: properties work also allows us to follow the C/LTI phase boundary to
690: higher fields, where it crosses over both the HTI and PM phases.
691: Finally, these studies also identify a new ``High Field" (HF)
692: region, nestled between C and the cascade of LTI, HTI, and PM phases
693: at higher temperature (Fig. \ref{PhaseDia}(b)). Reexamination of
694: previous specific heat measurements on Ni$_3$V$_2$O$_8$ are
695: consistent with the presence of an additional weak transition for
696: $H$ = 8 T, as shown by the arrow in (Fig. \ref{PhaseDia}(a)). A
697: similar transition was also observed at $H$ = 7 T.
698:
699:
700:
701:
702: \begin{figure}[t]
703: \centerline{
704: \includegraphics[width=3.2in]{SpinePeakEa5K.EPS}}
705: \vskip -0.2cm \caption{(Color online) Polarized optical conductivity
706: of Ni$_3$V$_2$O$_8$ at 5 K for $H$ = 0, 10, 20, and 30 T
707: ($H$$\parallel$$b$). The inset shows the peak position of fitted
708: oscillators as a function of applied magnetic field. The shaded
709: regions represent transition fields, from which the extended $H$-$T$
710: phase diagram in Fig. \ref{PhaseDia}(b) was
711: generated.}\label{SpinePeakEa5K}
712: \end{figure}
713:
714:
715: As discussed in the Introduction, Ni$_3$V$_2$O$_8$ has a complex
716: $H$-$T$ phase diagram with several low field phases arising from the
717: complex spin structures.\cite{lawes05,lawes04} With the new
718: transition fields extracted from the magneto-optical data at 5, 8,
719: and 12 K (solid circles) and specific heat data (solid triangles),
720: we extended the $H$-$T$ phase diagram of Ref.[\onlinecite{lawes04}]
721: for $H$$\parallel$$b$, as shown in Fig. \ref{PhaseDia}(b). We used
722: the combined optical and specific heat data to establish the
723: boundaries for the newly identified high field phase. It has been
724: predicted\cite{lawes04} that the phase boundary for the LTI phase
725: should be quadratic in $H$, so that ($T_c(H)$-$T_c(0)$) $\sim$
726: $H^2$. Rather than this quadratic dependence, we find that this
727: phase boundary can be fit by ($T_c(H)$-$T_c(0)$) $\sim$ $H^{2.5}$.
728: This fit is shown by the lower solid line in Fig. \ref{PhaseDia}(b).
729: The origins for this discrepancy are unclear at this time. The
730: upper solid line in Fig. \ref{PhaseDia}(b) is a guide to the eye for
731: the high field (HF) phase boundary. To the best of our
732: knowledge, there has been no discussion of any possible functional
733: form that might apply to this boundary. Combined magnetization and
734: specific heat measurements in high magnetic fields will be required
735: to precisely determine the magnetic field dependence of both phase
736: boundary lines, which in turn will provide mechanistic information
737: on the underlying magnetic transitions. Addition of the new data
738: points from optical and specific heat studies also demonstrates the
739: existence of a new HF phase that overlays the other low field
740: regions. Neutron experiments are needed to assess magnetic order in
741: the HF phase.
742:
743:
744:
745: In addition to the 5 K (LTI phase) data discussed in detail here, we
746: also collected full data sets in the HTI and PM phases of
747: Ni$_3$V$_2$O$_8$. Strong magneto-optical effects were observed at
748: both 8 and 12 K. As shown in Fig. \ref{OpConduc12K}, the
749: magneto-optical effects in the PM phase are qualitatively similar to
750: those of the LTI and HTI (not shown) phases, although the overall
751: size of the field-induced modification is smaller. This result
752: demonstrates that an applied field induces a local structural
753: distortion around the Ni sites in all of the low temperature phases.
754: In otherwords, the field-induced distortion is not unique to the LTI
755: (ferroelectric) phase with spontaneous polarization. To support this
756: claim, we refer the reader to the inset of Fig.
757: \ref{OpConduc12K}(b)), where the zero-field $\sim$1.35 eV excitation
758: is featureless. An applied field splits this structure into several
759: distinct components, demonstrating the field-induced distortion of
760: the Ni (spine) crystal field environment even in the PM phase.
761:
762:
763: \begin{figure}[h]
764: \centerline{
765: \includegraphics[width=3.2in]{OpConduc12K.EPS}}
766: \vskip -0.2cm \caption{ (Color online) (a) The normalized
767: magneto-optical response, R($H$)/R($H$=0 T), of Ni$_3$V$_2$O$_8$ in
768: an applied magnetic field ($H$$\parallel$$b$) from 0 to 30 T at 12 K
769: for light polarized along the $a$ direction. (b) Polarized optical
770: conductivity of Ni$_3$V$_2$O$_8$ at 12 K for $H$ = 0 and 30 T
771: ($H$$\parallel$$b$). Inset shows a close-up view of the
772: magneto-optical response near the Ni (spine) $d$ to $d$ on-site
773: excitation.}\label{OpConduc12K}
774: \end{figure}
775:
776:
777:
778:
779:
780:
781: \section{Conclusion}
782:
783: We investigated the optical properties, electronic structure, and
784: energy dependent magneto-optical response of Ni$_3$V$_2$O$_8$ to
785: elucidate the electronic structure and to study the phase diagram.
786: The spectra exhibit features centered at $\sim$0.75 and $\sim$1.35
787: eV that we assign as Ni (cross-tie and spine) $d$ to $d$ on-site
788: excitations in the minority spin channel. O 2$p$ to Ni 3$d$ and O
789: 2$p$ to V 3$d$ charge transfer excitations appear at higher energy.
790: Extensive analysis of splitting patterns of the Ni (spine) $d$ to
791: $d$ excitation in the PM, HTI, and LTI phases demonstrates that the
792: local Ni environment is sensitive to magnetic order even at zero
793: field. A splitting of Ni (cross-tie) $d$ to $d$ on-site excitation
794: between 50 and 100 K also points toward a weak local structural
795: distortion that precedes the cascade of low temperature magnetic
796: phases. Although both Ni$_3$V$_2$O$_8$ and the prototypical Mott
797: insulator NiO are based on Ni$^{2+}$ ions octahedrally coordinated
798: by O with similar bond lengths, the basic electronic structures of
799: these two materials are very different. NiO has a large band gap and
800: local moments that derive from strong Coulomb interactions and
801: separation of spin and charge degrees of freedom, while, in
802: contrast, we found Ni$_3$V$_2$O$_8$ to be an intermediate gap, local
803: moment band insulator. This electronic structure is particularly
804: favorable for magneto-dielectric coupling, because the material is
805: not subject to the spin charge separation of the strongly correlated
806: large gap Mott insulator, while at the same time, remaining a
807: magnetic insulator independent of the particular spin order and
808: temperature. The remarkable interplay of magnetic field and optical
809: properties in Ni$_3$V$_2$O$_8$ reveals an additional high field
810: phase and an unexpected electronic structure, which we associate
811: with the strong magnetodielectric couplings in this material. We
812: discovered several prominent magneto-optical effects that derive
813: from changes in crystal field environment around Ni (spine and
814: cross-tie) centers due to a field-induced modification of local
815: structure of NiO$_6$. The magnetoelastic mechanism, responsible for
816: the field-induced distortion of the NiO$_6$ building block unit, is
817: active in the paramagnetic phase as well.
818:
819:
820: \section{ACKNOWLEDGMENTS}
821:
822: Work at the University of Tennessee is supported by the Materials
823: Science Division, Basic Energy Sciences, U.S. Department of Energy
824: (DE-FG02-01ER45885). Research at ORNL is sponsored by the Division
825: of Materials Sciences and Engineering, Office of Basic Energy
826: Sciences, U.S. Department of Energy, under contract
827: DE-AC05-00OR22725 with Oak Ridge National Laboratory, managed and
828: operated by UT-Battelle, LLC. Work at UC Davis is supported by DOE
829: grant DE-FG03-01ER45876. A portion of this research was performed at
830: the NHMFL, which is supported by NSF Cooperation Agreement
831: DMR-0084173 and by the State of Florida. Work at Princeton
832: University is supported by NSF through the MRSEC program (NSF MRSEC
833: grant DMR-9809483). We appreciate useful discussions with W. E.
834: Pickett, and thank Q. Huang and M. Woodward for their assistance in
835: preparing Fig. \ref{struct}.
836:
837:
838:
839: \begin{thebibliography}{100}
840:
841: \bibitem{sauerbrei73}
842: E. E. Sauerbrei, R. Faggiani, and C. Calvo, Acta Cryst. B {\bf 29},
843: 2304 (1973).
844: %Refinement of the Crystal Structures of C03V2Os and Ni3V208
845:
846: \bibitem{rogado02}
847: N. Rogado, G. Lawes, D. A. Huse, A. P. Ramirez, and R. J. Cava,
848: Solid State Commun. {\bf 124}, 229 (2002).
849: %The Kagome´-staircase lattice: magnetic ordering in Ni3V2O8 and Co3V2O8
850:
851: \bibitem{lawes05}
852: G. Lawes, A. B. Harris, T. Kimura, N. Rogado, R. J. Cava, A.
853: Aharony, O. Entin-Wohlman, T. Yildirim, M. Kenzelmann, C. Broholm,
854: and A. P. Ramirez, Phys. Rev. Lett. \textbf{95}, 087205 (2005).
855: %Magnetically Driven Ferroelectric Order in Ni3V2O8
856:
857: \bibitem{lawes04}
858: G. Lawes, M. Kenzelmann, N. Rogado, K. H. Kim, G. A. Jorge, R. J.
859: Cava, A. Aharony, O. Entin-Wohlman, A. B. Harris, T. Yildirim, Q. Z.
860: Huang, S. Park, C. Broholm, and A. P. Ramirez, Phys. Rev. Lett.
861: \textbf{93}, 247201 (2004).
862: %Competing Magnetic Phases on a Kagom$\acute{e}$´ Staircase
863:
864: \bibitem{kenzelmann06}
865: M. Kenzelmann, A. B. Harris, A. Aharony, O. Entin-Wohlman, T.
866: Yildirim, Q. Huang, S. Park, G. Lawes, C. Broholm, N. Rogado, R. J.
867: Cava, K. H. Kim, G. Jorge, and A. P. Ramirez, cond-mat/0510386
868: (2005).
869: %Field dependence of magnetic ordering in Kagom´e-staircase compound
870: %Ni3V2O8
871:
872: \bibitem{cheong04}
873: N. Hur, S. Park, S. Guha, A. Borissov, V. Kiryukhin, and S-W.
874: Cheong, Appl. Phys. Lett. \textbf{87}, 042901 (2005).
875: %Low-field magneto-dielectric effect in terbium iron garnets.
876:
877: \bibitem{cruz05}
878: C. de la Cruz, F. Yen, B. Lorenz, Y. Q. Wang, Y. Y. Sun, M. M.
879: Gospodinov, and C. W. Chu, Phys. Rev. B \textbf{71}, 060407(R)
880: (2005).
881: %Strong spin-lattice coupling in multiferroic HoMnO3: Thermal
882: %expansion anomalies and pressure effect.
883:
884: \bibitem{lorenz04}
885: B. Lorenz, A. P. Litvinchuk, M. M. Gospodinov, and C. W. Chu, Phys.
886: Rev. Lett. \textbf{92}, 087204 (2004).
887: %Field-Induced Reentrant Novel Phase and a Ferroelectric-Magnetic
888: %Order Coupling in HoMnO3.
889:
890: \bibitem{tgoto04}
891: T. Goto, T. Kimura, G. Lawes, A. P. Ramirez, and Y. Tokura, Phys.
892: Rev. Lett. \textbf{92}, 257201 (2004).
893: %Ferroelectricity and Giant Magnetocapacitance in Perovskite
894: %Rare-Earth Manganite.
895:
896: \bibitem{hur04}
897: N. Hur, S. Park, P. A. Sharma, S. Guha, and S-W. Cheong, Phys. Rev.
898: Lett. \textbf{93}, 107207 (2004).
899: %Colossal magneto-dielectric Effects in DyMn2O5.
900:
901: \bibitem{lawes03}
902: G. Lawes, A. P. Ramirez, C. M. Varma, and M. A. Subramanian, Phys.
903: Rev. Lett. \textbf{91}, 257208 (2003).
904: %magneto-dielectric Effects from Spin Fluctuations in Isostructural
905: %Ferromagnetic and Antiferromagnetic Systems.
906:
907: \bibitem{rogado05}
908: N. S. Rogado, J. Li, A. W. Sleight, and M. A. Subramanian, Adv.
909: Mater. \textbf{17}, 2225 (2005).
910: %Magnetocapacitance and Magnetoresistance Near Room Temperature
911: %in a Ferromagnetic Semiconductor: La2NiMnO6
912:
913: \bibitem{saito05}
914: M. Saito, R. Higashinaka, and Y. Maeno, Phys. Rev. B \textbf{72},
915: 144422 (2005)
916: %magneto-dielectric response of the spin-ice Dy2Ti2O7
917:
918: \bibitem{subra06}
919: M. A. Subramanian, T. He, J. Chen, N. S. Rogado, T. G. Calvarese,
920: and A. W. Sleight, Adv. Mater. \textbf{18}, 1737 (2006).
921: %Giant Room-Temperature Magnetodielectric Response in the Electronic
922: %Ferroelectric LuFe2O4.
923:
924: \bibitem{rai06}
925: R. C. Rai, J. Cao, J. L. Musfeldt, D. J. Singh, X. Wei, R. Jin, Z.
926: X. Zhou, B. C. Sales, and D. Mandrus, Phys. Rev. B \textbf{73},
927: 075112 (2006).
928: %Magneto-Dielectric Effect in the $S$ = 1/2 Quasi-Two Dimensional
929: %Antiferromagnet K$_2$V$_3$O$_8$
930:
931: \bibitem{zhu02}
932: Z.-T. Zhu, J. L. Musfeldt, Z. S. Teweldemedhin, and M. Greenblatt,
933: Phys. Rev. B \textbf{65}, 214519 (2002).
934: %Anisotropic ab-lane optical response of the charge-density-wave
935: %superconductor P4W14O50.
936:
937: \bibitem{wooten72}
938: F. Wooten, {\it Optical Properties of Solids} (Academic Press, New
939: York, 1972).
940:
941: \bibitem{cao06}
942: J. Cao, J. T. Haraldsen, R. C. Rai, S. Brown, J. L. Musfeldt, Y. J.
943: Wang, X. Wei, M. Apostu, R. Suryanarayanan, and A. Revcolevschi,
944: Phys. Rev. B \textbf{74}, 045113 (2006).
945: %Magneto-optical investigation of the field-induced spin-glass-insulator
946: %to ferromagnetic-metal transition in the bilayer manganite, La0.4Pr0.6…1.2Sr1.8Mn2O7
947:
948: \bibitem{lapw}
949: D. J. Singh and L. Nordstrom, {\it Planewaves, Pseudopotentials and
950: the LAPW Method, 2nd Ed.} (Springer, Berlin, 2006).
951:
952: \bibitem{lo}
953: D. Singh, Phys. Rev. B {\bf 43}, 6388 (1991).
954:
955: \bibitem{wien}
956: P. Blaha, K. Schwarz, G. K. H. Madsen, D. Kvasnicka, and J. Luitz,
957: {\small WIEN2K}, 2002, {\it an Augmented Plane Wave + Local Orbitals
958: Program for calculating crystal properties} (Karlheinz Schwarz,
959: Technische Universitat Wien, Austria).
960:
961: \bibitem{apw}
962: E. Sjostedt, L. Nordstrom, and D. J. Singh, Solid State Commun. {\bf
963: 114}, 15 (2000).
964:
965: \bibitem{u-note}
966: Here we used an effective $U-J$ = 5 eV, $J$ = 0, though because of
967: the fully spin polarized nature of the Ni the results would not
968: depend on the choice of $J$ provided $U-J$ is held constant.
969:
970: \bibitem{sawatzky}
971: G. A. Sawatzky and J. W. Allen, Phys. Rev. Lett. {\bf 53}, 2339
972: (1984).
973:
974: \bibitem{schnack06}
975: J. Schnack, M. Br\"{u}ger, M. Luban, P. K\"{o}gerler, E. Morosan, R.
976: Fuchs, R. Modler, H. Nojiri, R. C. Rai, J. Cao, and J. L. Musfeldt,
977: Phys. Rev. B, \textbf{73}, 094401 (2006).
978: %Observation of field-dependent magnetic parameters in the magnetic
979: %molecule {Ni4Mo12}
980:
981: \bibitem{ido91}
982: T. Ido, K. Magoshi, H. Eisaki, and S. Uchida, Phys. Rev. B
983: \textbf{44}, 12094 (1991).
984: %Optical study of the La(2-x)Sr(x)NiO(4) system: Effect of hole doping on
985: %the electronic structure of the NiO2 plane.
986:
987: \bibitem{juza66}
988: R. Juza, H. Seidel, and J. Tiedemann, Angew. Chem. Internat. Edit.
989: \textbf{5}, 85 (1966).
990: %Color centers in alkali Borate glasses containing cobal, nickel, or copper.
991:
992: \bibitem{note1}
993: Although magneto-optical measurements below 0.75 eV are limited by
994: the energy range (0.75 - 4.1 eV) of our experimental set-up, we
995: suspect that splitting of Ni(cross-tie) $d$ to $d$ excitation may
996: be similar to the splitting of Ni (spine) $d$ to $d$ excitation.
997:
998: \bibitem{flores06}
999: A. F. Garc\'{i}a-Flores, E. Granado, H. Martinho, R. R. Urbano, C.
1000: Rettori, E. I. Golovenchits, V. A. Sanina, S. B. Oseroff, S. Park,
1001: and S-W. Cheong, Phys. Rev. B \textbf{73}, 104411 (2006).
1002: %Anomalous phonon shifts in the paramagnetic phase of multiferroic
1003: %RMn2O5 (R=Bi, Eu, Dy): Possible manifestations of unconventional
1004: %magnetic correlations.
1005:
1006: \bibitem{wang}
1007: D. Wang, J. Tang, Z. Zou, and J. Ye, Chem. Mater. {\bf 17}, 5177
1008: (2005).
1009:
1010: \bibitem{terakura}
1011: K. Terakura, A. R. Williams, T. Oguchi, and J. K\"{u}bler, Phys.
1012: Rev. Lett. {\bf 52}, 1830 (1984).
1013:
1014: \bibitem{merlin}
1015: R. Merlin, Phys. Rev. Lett. {\bf 54}, 2727 (1985)
1016: % discussion of PE data yielding U large
1017:
1018: \bibitem{norman}
1019: M. R. Norman and A. J. Freeman, Phys. Rev. B {\bf 33}, 8896 (1986).
1020: %excited state calculations yielding large gap and large U
1021:
1022: \bibitem{anisimov-1}
1023: V. I. Anisimov, M. A. Korotin, and E. Z. Kurmaev, J. Phys. Condens.
1024: Matter {\bf 2}, 3973 (1990).
1025:
1026: \bibitem{anisimov-2}
1027: V. I. Anisimov, J. Zaanen, O. K. Andersen, Phys. Rev. B {\bf 44},
1028: 943 (1991).
1029:
1030: \bibitem{svanne}
1031: A. Svanne and O. Gunnarsson, Phys. Rev. Lett. {\bf 65}, 1148 (1990).
1032: %SIC
1033:
1034: \bibitem{arai}
1035: M. Arai and T. Fujiwara, Phys. Rev. B {\bf 51}, 1477 (1995).
1036:
1037: \bibitem{kod}
1038: D. Kodderitzsch, W. Hergert, W. M. Temmerman, Z. Szotek, A. Ernst,
1039: and H. Winter, Phys. Rev. B {\bf 66}, 064434 (2002).
1040: % SIC for NiO
1041:
1042: \bibitem{aryasetiawan}
1043: F. Aryasetiawan and O. Gunnarsson, Phys. Rev. Lett. {\bf 74}, 3221
1044: (1995).
1045: % GW calculations for NiO yield gap 5.5 eV, c.f. expt. 4.0 eV
1046:
1047: \bibitem{massidda}
1048: S. Massidda, A. Continenza, M. Posternak, and A. Baldereschi, Phys.
1049: Rev. B {\bf 55}, 13494 (1997).
1050: % GW NiO
1051:
1052: \bibitem{li}
1053: J. L. Li, G. M. Rignanese, and S. G. Louie, Phys. Rev. B {\bf 71},
1054: 193102 (2005).
1055: %GW calculations for NiO
1056:
1057: \bibitem{brown}
1058: I. D. Brown, {\it Structure and Bonding in Crystals} (Academic
1059: Press, New York, 1981).
1060:
1061: \bibitem{woodward05} J.D. Woodward, J. Choi, J.L. Musfeldt, J.T.
1062: Haraldsen, X. Wei, H.-J. Koo, D. Dai, M.-H. Whangbo, C.P. Landee,
1063: and M.M. Turnbull, Phys. Rev. B. {\bf 71}, 174416 (2005).
1064:
1065: \bibitem{choi04} J. Choi, J.D. Woodward, J.L. Musfeldt, X. Wei, J.
1066: He, R. Jin, and D. Mandrus, Phys. Rev. B. {\bf 70}, 085107 (2004).
1067:
1068: \bibitem{sushkov05}
1069: A. B. Sushkov, O. Tchernyshyov, W. Ratcliff II, S. W. Cheong, and H.
1070: D. Drew, Phys. Rev. Lett. \textbf{94},137202 (2005).
1071: %Probing Spin Correlations with Phonons in the Strongly Frustrated
1072: %Magnet ZnCr2O4
1073:
1074: \bibitem{lee02}
1075: S.-H. Lee, C. Broholm, W. Ratcliff, G. Gasparovic, Q. Huang, T. H.
1076: Kim, and S.-W. Cheong, Nature \textbf{418}, 856 (2002).
1077: %Emergent excitations in a geometrically frustrated magnet.
1078:
1079: \bibitem{ratclif05}
1080: W. Ratcliff II, V. Kiryukhin, M. Kenzelmann, S.-H. Lee, R. Erwin, J.
1081: Schefer, N. Hur, S. Park, and S-W. Cheong, Phys. Rev. B \textbf{72},
1082: 060407(R) (2005).
1083: %Magnetic phase diagram of the colossal magnetoelectric DyMn2O5
1084:
1085: \bibitem{blake05}
1086: G. R. Blake, L. C. Chapon, P. G. Radaelli, S. Park, N. Hur, S-W.
1087: Cheong, and J. Rodr\'{\i}guez-Carvajal, Phys. Rev. B \textbf{71},
1088: 214402 (2005).
1089: %Spin structure and magnetic frustration in multiferroic
1090: %RMn2O5 %(R=Tb,Ho,Dy).
1091:
1092: \end{thebibliography}
1093:
1094: \vfill\eject
1095:
1096: \end{document}
1097: