Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides

The Raman optical activity (ROA) of magnons in rutile-structure antiferromagnetic FeF₂ (TN = 78 K) has been studied as a function of temperature and applied magnetic field. For exciting light incident along the c axis, ROA is observed for magnons but not for phonons. In zero field, a small splitt...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Datum:2005
Hauptverfasser: Hoffman, K.R., Lockwood, D.J., Yen, W.M.
Format: Artikel
Sprache:English
Veröffentlicht: Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України 2005
Schriftenreihe:Физика низких температур
Schlagworte:
Online Zugang:https://nasplib.isofts.kiev.ua/handle/123456789/121698
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Zitieren:Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides / K.R. Hoffman, D.J. Lockwood, W.M. Yen // Физика низких температур. — 2005. — Т. 31, № 8-9. — С. 1032-1041. — Бібліогр.: 32 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-121698
record_format dspace
spelling nasplib_isofts_kiev_ua-123456789-1216982025-06-03T16:29:05Z Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides Hoffman, K.R. Lockwood, D.J. Yen, W.M. К семидесятилетию антиферромагнетизма The Raman optical activity (ROA) of magnons in rutile-structure antiferromagnetic FeF₂ (TN = 78 K) has been studied as a function of temperature and applied magnetic field. For exciting light incident along the c axis, ROA is observed for magnons but not for phonons. In zero field, a small splitting (0.09 cm⁻¹) of the two acoustic–magnon branches is observed for the first time by inelastic light scattering. The splitting in applied magnetic field is found to reduce with increasing temperature in accordance with theory. No ROA was detected for two-magnon excitations. In optical absorption measurements performed over thirty years ago, a very small circular dichroism (CD) was observed in the magnon sidebands of other simple rutile antiferromagnetic fluorides (MnF₂ and CoF₂). The origin of this CD was not understood at the time. The Raman studies of the one-magnon Raman scattering in FeF₂ have demonstrated that in zero field the degeneracy of the antiferromagnetic magnon branches is lifted by a weak magnetic dipole–dipole interaction, as predicted by Pincus and Loudon and by White four decades ago. The source of the observed CD in the magnon sidebands can now be traced to this same magnetic–dipole induced splitting. The precise experimental work of Professor F.L. Scarpace and Dr. Y.H. (Russ) Wong from thirty years ago is noted with appreciation. We thank Professors R. M. White and M. G. Cottam for useful discussions and the National Science Foundation for continued support spanning these three decades. 2005 Article Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides / K.R. Hoffman, D.J. Lockwood, W.M. Yen // Физика низких температур. — 2005. — Т. 31, № 8-9. — С. 1032-1041. — Бібліогр.: 32 назв. — англ. 0132-6414 PACS: 75.30.Ds, 75.30.Gw, 75.50.Ee, 78.20.Ls, 78.30.Hv https://nasplib.isofts.kiev.ua/handle/123456789/121698 en Физика низких температур application/pdf Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
language English
topic К семидесятилетию антиферромагнетизма
К семидесятилетию антиферромагнетизма
spellingShingle К семидесятилетию антиферромагнетизма
К семидесятилетию антиферромагнетизма
Hoffman, K.R.
Lockwood, D.J.
Yen, W.M.
Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides
Физика низких температур
description The Raman optical activity (ROA) of magnons in rutile-structure antiferromagnetic FeF₂ (TN = 78 K) has been studied as a function of temperature and applied magnetic field. For exciting light incident along the c axis, ROA is observed for magnons but not for phonons. In zero field, a small splitting (0.09 cm⁻¹) of the two acoustic–magnon branches is observed for the first time by inelastic light scattering. The splitting in applied magnetic field is found to reduce with increasing temperature in accordance with theory. No ROA was detected for two-magnon excitations. In optical absorption measurements performed over thirty years ago, a very small circular dichroism (CD) was observed in the magnon sidebands of other simple rutile antiferromagnetic fluorides (MnF₂ and CoF₂). The origin of this CD was not understood at the time. The Raman studies of the one-magnon Raman scattering in FeF₂ have demonstrated that in zero field the degeneracy of the antiferromagnetic magnon branches is lifted by a weak magnetic dipole–dipole interaction, as predicted by Pincus and Loudon and by White four decades ago. The source of the observed CD in the magnon sidebands can now be traced to this same magnetic–dipole induced splitting.
format Article
author Hoffman, K.R.
Lockwood, D.J.
Yen, W.M.
author_facet Hoffman, K.R.
Lockwood, D.J.
Yen, W.M.
author_sort Hoffman, K.R.
title Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides
title_short Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides
title_full Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides
title_fullStr Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides
title_full_unstemmed Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides
title_sort circular dichroism and raman optical activity in antiferromagnetic transition metal fluorides
publisher Фізико-технічний інститут низьких температур ім. Б.І. Вєркіна НАН України
publishDate 2005
topic_facet К семидесятилетию антиферромагнетизма
url https://nasplib.isofts.kiev.ua/handle/123456789/121698
citation_txt Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides / K.R. Hoffman, D.J. Lockwood, W.M. Yen // Физика низких температур. — 2005. — Т. 31, № 8-9. — С. 1032-1041. — Бібліогр.: 32 назв. — англ.
series Физика низких температур
work_keys_str_mv AT hoffmankr circulardichroismandramanopticalactivityinantiferromagnetictransitionmetalfluorides
AT lockwooddj circulardichroismandramanopticalactivityinantiferromagnetictransitionmetalfluorides
AT yenwm circulardichroismandramanopticalactivityinantiferromagnetictransitionmetalfluorides
first_indexed 2025-11-24T02:25:34Z
last_indexed 2025-11-24T02:25:34Z
_version_ 1849636824624922624
fulltext Fizika Nizkikh Temperatur, 2005, v. 31, Nos. 8/9, p. 1032–1041 Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides K.R. Hoffman Department of Physics, Whitman College, Walla Walla, WA 99362, USA D.J. Lockwood Institute for Microstructural Sciences, National Research Council, Ottawa, ON K1A 0R6, Canada E-mail: david.lockwood@nrc.ca W.M. Yen Department of Physics and Astronomy, University of Georgia, Athens, GA 30602, USA Received February 17, 2005 The Raman optical activity (ROA) of magnons in rutile-structure antiferromagnetic FeF2 (TN = 78 K) has been studied as a function of temperature and applied magnetic field. For exciting light incident along the c axis, ROA is observed for magnons but not for phonons. In zero field, a small splitting (0.09 cm–1) of the two acoustic–magnon branches is observed for the first time by inelastic light scattering. The splitting in applied magnetic field is found to reduce with increasing temperature in accordance with theory. No ROA was detected for two-magnon excitations. In opti- cal absorption measurements performed over thirty years ago, a very small circular dichroism (CD) was observed in the magnon sidebands of other simple rutile antiferromagnetic fluorides (MnF2 and CoF2). The origin of this CD was not understood at the time. The Raman studies of the one-magnon Raman scattering in FeF2 have demonstrated that in zero field the degeneracy of the antiferromagnetic magnon branches is lifted by a weak magnetic dipole–dipole interaction, as pre- dicted by Pincus and Loudon and by White four decades ago. The source of the observed CD in the magnon sidebands can now be traced to this same magnetic–dipole induced splitting. PACS: 75.30.Ds, 75.30.Gw, 75.50.Ee, 78.20.Ls, 78.30.Hv 1. Introduction In 1965, the discovery of the one-magnon sideband in the visible absorption spectrum of MnF2 [1] and the two-magnon absorption of the far-infrared spec- trum of FeF2 [2] spurred many investigations into the optical properties of antiferromagnetic insulators. The theoretical framework regarding these two processes was developed by Tanabe et al. [3] and Allen et al. [4] In these materials, coupling between the various in- trinsic excitations of the lattice, i.e., excitons, mag- nons and phonons, leads to the formation of coopera- tive transitions which can account for much of the spectral structure. An article discussing this work ap- pears elsewhere in this special issue of Low Tempera- ture Physics. In addition to absorption and lumines- cence measurements, magnon lines were also observed in Raman scattering measurements [5,6]. Raman scat- tering measurements permit the physical properties of magnons to be studied without the additional contri- butions due to electronic states associated with optical emission or absorption. A comprehensive review of Raman theory and experimental results for antifer- romagnets is presented in the book by Cottam and Lockwood [7]. The magnetic character of the magnon transitions led to the utilization of circularly polarized light to explore the dichroic properties of these materials. In the early 1970’s, a series of papers reported on their magnetic circular dichroism (MCD) in an applied magnetic field [8–10]. The authors reported at that time the observation of circular dichroism (CD) in © K.R. Hoffman, D.J. Lockwood, and W.M. Yen, 2005 certain magnon sidebands of MnF2 [8] and CoF2 [10] but not in FeF2 [9]; CD occurs in the absence of an applied magnetic field and corresponds to zero field MCD though the origins of the dichroism may arise from different sources. The authors were able to elimi- nate experimental artifacts as the source of the ob- served CD, but were not able at that time to identify the physical basis for the unexpected observation. Re- cently, we revisited this problem and showed that the source of the CD is to be found in the dipolar interac- tions that exist between the antiferromagnetic sub- lattices, which render them inequivalent [11]. Raman optical activity (ROA) characterizes the difference in the Raman line intensity when excited with right and left circularly polarized light. The ear- liest experiments to combine Raman spectroscopy with chirality in solids were performed in the 1970’s on CdS under resonance excitation conditions [12]. Attempts were also made to observe a similar effect in antiferromagnetic fluoride materials but the experi- mental sensitivity was insufficient. At about the same time ROA was also measured in a chiral molecule [13]. The chiral structure of some molecules makes them naturally suited to couple more strongly to ei- ther right or left circularly polarized light. The theo- retical [14] and experimental developments in this field have been reviewed in the literature [15,16]. A series of Raman papers in the 1980’s on the magnetic properties of dilute magnetic semiconductors employed the use of circular polarization to isolate spin flip tran- sitions of Mn ions [17,18]. The improvement in detec- tor sensitivity over a period of twenty years afforded us the opportunity in the 1990’s to revisit the question of ROA in antiferromagnetic insulators. We were able to observe ROA in the one-magnon line of FeF2 [19–21] and several phonon lines in rutile structure materials [22,23]. In addition, we demonstrated for the first time that Raman scattering could be used to measure a zero field splitting of the magnon branches in antifer- romagnetic FeF2 [24]. These results were then utilized to reinterpret the CD measurements completed in the early 1970’s that were alluded to above [11]. In the following Sections, we present our current understanding of the dichroic properties of antifer- romagnetic insulators possessing the rutile structure. The focus of this review will be on ROA measurements to explore the magnetic properties of these materials. We start with a summary of the zero field dichroism measurements in MnF2, CoF2, and FeF2 because the CD studies prompted one of us (WMY) to explore the use of ROA in antiferromagnetic fluorides. We then present a more thorough review of the ROA measure- ments, techniques, and analysis in FeF2. Here we mention first the work on phonons before focusing on the one- and two-magnon scattering results, including a revised assessment of the results from the temperature dependence and zero magnetic field studies of FeF2. 2. Optical dichroism in rutile antiferromagnets Dichroism in a material is defined as the difference in the absorption coefficient of a medium for two re- lated polarizations; thus CD arises from the difference in the absorption for right- and left-circularly polar- ized radiation. When these differences are induced by the application of an external magnetic field, the ef- fect is termed MCD. In both CD and MCD, a unique direction is necessary in order to define the sense of circulation in the circularly polarized light; this is provided by some unique crystalline axis or by the di- rection of the applied magnetic field. Materials show- ing CD are normally called «optically active» and such crystals are relatively rare, whereas MCD is quite common in paramagnetic systems possessing a unique axis. MCD is conjugated to Faraday rotation by the Kramers–Kronig relations and constitutes the resonant rather than the dispersive phenomena in the transformation. The compounds that we wish to discuss here are the common difluorides, MnF2, CoF2, and FeF2, all of which possess the rutile crystal structure. Below the Neel temperature, TN, these materials are describable as simple two sublattice antiferromagnets and in prin- ciple they are ideal systems on which to study magnet- ically dichroic phenomena. The ground states of ions in each sublattice as well as the collective magnetic excitations (magnons) are identical to each other in every respect except their polarization. The sole re- maining energy degeneracy, which will respond to an applied magnetic field, is that of the two sublattices; hence, the origin of magnetic dichroism in these sys- tems is directly related to the lifting of this energy equivalence both in the electronic (exciton), the spin (magnon) or the combined (magnon sideband) systems. For simple antiferromagnets in their ordered state, the ground state of each sublattice is non-degenerate and is identical to the other except for their sense of magnetization with respect to the unique crystalline axis. If the state for the «up» or A sublattice is given by |L,M� then the corresponding «down» or B sublattice is represented by |L,–M�. For pure elec- tronic transitions, the appropriate dipolar transition operators for circular polarized radiation transform as L±+ 2S± where the (+) and (–) subscripts refer to right and left circular polarizations, respectively. If ions in one of the sublattices have a non-vanishing transition probability between two states for one of these operators, then the ions in the opposite sub- lattice will have an identical element for the corre- Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides Fizika Nizkikh Temperatur, 2005, v. 31, Nos. 8/9 1033 sponding conjugated operator. In the absence of a magnetic field, the sublattices are degenerate in en- ergy and no CD should be observed. However, weak interactions between the sublattices, such as magnetic dipole–dipole interactions, can produce a small aniso- tropy between the two sublattices that results in CD. At low temperatures, a very small zero-field CD signal was observed in two of the three compounds investi- gated and it is the origin of these signals that will be addressed in this paper. 3. Experimental results for circular dichroism measurements A finite CD was observed at low temperatures in two of the three difluorides studied. Signals were observed in the magnon sidebands of the 6A1g ground state to 4T1g and ( 4 1A g, 4Eg) absorptions of MnF2 and of the 4T1 to 2A, 2T1 (2P) of CoF2; no CD signals were observed in FeF2. As noted earlier, the origins of CD and MCD are different; as a consequence, CD can be subtracted from MCD in order to determine the magnetic field induced splittings. When this is done the, the residual or pure MCD signal can be fitted accurately with a differential line shape; this indicates that in our cases the MCD arise simply from the lifting of the sublattice degeneracy, as per our expectation. Figure 1 shows the absorption of a magnon side- band of the (2A, 2T1) transition of CoF2 along with the observed CD and MCD in a 2 T applied magnetic field. The CD is also purely differential in shape and is estimated to be an additive intrinsic splitting of the order of �� = 0.015 cm–1. The CD in CoF2 becomes unobservable in the vicinity of 10 K. Figure 2 shows the absorption spectrum of the so-called �1 and �2 sidebands accompanying the ground state to 4T1g transition in MnF2. These transi- tions are �-polarization active. A small CD signal is observed that is clearly related to this transition. The zero field splittings for the sidebands were determined to be ��(�1) = 0.07 cm–1 and ��(�2) = –0.05 cm–1, respectively. The negative sign in ��(�2) signifies that the sense of the splitting in this sideband is out of phase with that of �1. Such a change in the sense of splitting is also observed when uniaxial stress is applied to MnF2 along the [110] direction in the basal plane. The behavior of the CD was studied as a function of basal plane stress allowing the intrinsic strain in the basal plane of or- dered MnF2 to be estimated: � �� � � �d/d 12 10 5. . This result highlights the advantages of chiral tech- niques for exploring interactions in antiferromagnets. FeF2 was studied in the 21500–28500 cm–1 spectral region. The observed transitions without exception exhibited MCD signals consistent with a simple sublattice splitting and with a linear dependence on magnetic field. No CD signals were encountered in any of the �-active transitions investigated. The sensitivity of our apparatus places an upper limit of 10–3 cm–1 on any splitting that might occur in this compound at 4 K. 1034 Fizika Nizkikh Temperatur, 2005, v. 31, Nos. 8/9 K.R. Hoffman, D.J. Lockwood, and W.M. Yen 22800 22850 22900 –2 –1 0 1 2 �, cm–1 0,0 0,5 1,0 1,5 T=4K Absorption CD H=0 CoF2 ( ) L( ) R ( ) ( V ) Fig. 1. The CD spectrum of the one-magnon sideband near 22,800 cm–1 of CoF2 at 4 K and the corresponding absorp- tion spectrum of the sideband. 18 470 18 480 18 49018 460 ( ) �1 �2 0 C ir cu la r d ic hr oi sm + ( ) ( ) �, cm–1 – – Fig. 2. The absorption and CD spectra of the so-called �1 and �2 sidebands of MnF2 at 2 K; these traces connect the observed CD to the magnon sidebands. 4. Raman optical activity Raman optical activity is observed experimentally in the difference in a Raman line intensity when ex- cited by right and left circularly polarized light, re- spectively. The circular intensity difference (CID) is normally used to quantify ROA [14]: � � � � � � � I I I I R L R L , (1) where I R � and I L � represent the scattered light inten- sity with � polarization due to incident right (R) or left (L) circularly polarized light. The normalization term permits a simple comparison of data from differ- ent spectroscopic systems. One major difficulty of ROA measurements is the sensitivity of the technique to intrinsic or extrinsic depolarization effects. Care must be taken to account for any birefringent effects in the steering and focusing optics. Additionally, the crystal itself can contribute to significant depolariza- tion effects that appear as CID in spectra that nor- mally would not be expected to produce any differ- ence. We will discuss this effect in more detail in the next Section of the paper. Fundamentally, the source of Raman scattering is the modulation of the induced polarizability of a me- dium by fundamental excitations in the medium. To couple the incident light to the medium, its suscepti- bility is expanded in terms of phonon or magnon ope- rators. This results in a modulation of the polarizability at the frequency of the fundamental excitation. In the case of magnons, the inclusion of spin operators re- sults in a change of the overall angular momentum component along the z axis: �mj = ±1. Figure 3 illus- trates the important mechanism for magnon light scat- tering from antiferromagnetic fluorides. The energy levels shown in the picture belong to the transition metal ions on the two sublattices. These transitions obey the same selection rules that contribute to CD or MCD signals. The circularly polarized light couples to different sublattices in the Raman process as well, thus providing a way of distinguishing between the two sublattices. The figure shows energy levels for both sublattices in the medium and illustrates that left and right circular polarizations create magnons on dif- ferent sublattices. Symmetry arguments can be applied to determine selection rules for the various scattering processes and are summarized in the form of Raman scattering ten- sors. The scattering tensors for the tetragonal antifer- romagnetic fluorides are presented in Table 1. The tensors have been transformed so that they are appropriate for [R, L, z] incident polarizations and [x, y, z] scattered polarizations. The Raman tensors describe the scattered electric field amplitudes and thus the scattered light intensity in a specific polariza- tion is calculated through multiplication of the tensor by its conjugate value. To describe the expected CID spectrum, we need to calculate the difference between two intensity terms. For example, the A1g phonon mode should not generate a CID signal for scattered light polarized along the x axis: � x g x R x L x R x L A I I I I a a a a ( ) .1 2 2 2 2 0� � � (2) Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides Fizika Nizkikh Temperatur, 2005, v. 31, Nos. 8/9 1035 J=S J=S –1 S–1 S J= –S J=S+1 – S+1 –S s_ z s+ z Fig. 3. The electronic levels of the magnetic ions in antiferromagnetically ordered rutile materials responsible for the observed ROA signal. The two sets of lines corre- spond to the two antiferromagnetic sublattices. In both cases a magnon is generated during the scattering process. Table 1. The Raman tensors for circularly polarized incident light that is analyzed for linearly polarized light scattered at 90 degrees. A g a a ia ia b 1 1 2 0 0 0 0 2 � � � � � � � � � B g c c ic ic1 1 2 0 0 0 0 0 � � � � � � � � � B g ie ie e e2 1 2 0 0 0 0 0 � �� � � � � � � 3 4 1 1 1 2 0 0 2 0 0 2 0 � � � � � � � � � � � � � � � � � � � � � � � � � i ( ) ( ) � � � � � Eg d id id ( )1 1 2 0 0 0 0 0 2 0 � � � � � � � � � Eg d d d ( )2 1 2 0 0 2 0 0 0 0 � � � � � � � � In the case of the magnon scattering we can calcu- late the CID for each sublattice separately. In the case of the �3 + band, we obtain the following CID expres- sion: � �z ( ) ( * * * ) ( * * * ) ( * * 3 2 2 2 2 2 � � � � � � � � � � � � � � � � � * ) ( * * * ) * ( * )� � � � � � � � � � � � � 2 2 2 2 2 2 . (3) Likewise for the �4 + magnon band we get a similar result: � �z ( ) * ( * )4 2 2 2 � � � � � (4) The negative sign reflects the coupling of left circu- lar polarization to this particular sublattice. Notice that if the energies of the two magnon bands are equal and the line profiles equivalent, the total CID is the sum of these terms and will be zero. The application of a magnetic field lifts the degeneracy of the magnon branches. As a result, the CID of the two branches will no longer completely cancel. This result is analo- gous to the differences observed between MCD and CD spectra. The important distinction is that the Raman spectra are determined only by the energies of the magnon bands with no interference from the ex- cited states of the ion. 5. Raman optical activity experiments The experimental arrangement, shown in Fig. 4, used the standard 90° Raman scattering geometry. The incident beam was the 488-nm line of a continu- ous-wave Spectra Physics 166 argon–ion laser. The focused laser power density at the sample was 150 kW/cm2, well below the damage threshold of the materials used. A Conoptics 370 electro-optic modula- tor (EOM) was used to alternate between right and left circular polarization. The scattered light was dis- persed by a Spex 1401, 0.85-m double monochromator and detected using a cooled RCA 31034-A02 photomultiplier. The signal was measured using a Stanford SR-400 dual-gated photon counter and stored on a computer. The signals counted on the two gates, corresponding to the two incident polarizations, were stored separately, permitting both IR – IL and IR+ + IL to be easily calculated. Low temperatures were achieved using a Cryosystems closed-cycle refrig- erator. A permanent magnet assembly located inside the refrigerator applied a constant magnetic field par- allel to the crystal c axis. The magnets were drilled through so that the incident laser beam could be fo- cused in the crystal co-linear with the applied field and allowed to pass on through the refrigerator. The crystals used in this study were the rutile struc- ture fluorides FeF2 and MgF2. These materials are uniaxial, with FeF2 undergoing an antiferromagnetic phase transition at TN = 78 K. The MgF2 crystal ori- entation was determined from x-ray Laue-diffraction measurements before the cuboid sample was cut (to within a 1° accuracy) and polished. The c-axis orienta- tion for the FeF2 sample was found using a polariscope. In addition, a polished crown glass cube was used as a reference material to check for instru- mental artifacts. Rayleigh and Raman CID in MgF2 and FeF2 Our measurements of the magnon and phonon CID in FeF2 generated several signals that were difficult to understand initially. Figure 5 shows the difference spectrum IR – IL as well as the separate IR and IL spectra from FeF2 with no applied magnetic field. What is curious about these measurements is that the magnon exhibits no apparent differential scattering but the B1g phonon mode does. When the spectrum is enlarged to encompass all of the phonon modes and the Rayleigh line, we observe a similar CID spectrum for the A1g and B2g modes and the Rayleigh line but the sign of the difference signal is opposite to that of the B1g mode. Referring back to the Raman tensors, no CID is expected from any of the phonon modes nor should Rayleigh scattering exhibit CID. MgF2 is also a rutile structure fluoride, but it does not exhibit any magnetic ordering. It showed a similar differential scat- 1036 Fizika Nizkikh Temperatur, 2005, v. 31, Nos. 8/9 K.R. Hoffman, D.J. Lockwood, and W.M. Yen Fig. 4. The experimental setup for the ROA measure- ments. The laser light was directed vertically through the sample to align the beam path with the entrance slit of the spectrometer. tering pattern. An isotropic glass sample was used to check for instrumental artifacts and none were found. To explain the CID spectra of the phonons we looked carefully at the effects of birefringence in the crystal on the scattered intensity. Light propagating along the c-axis will experience an isotropic index of refraction so that no change in the polarization occurs as it moves through the sample. A slight misalignment results in distinct indices of refraction parallel and perpendicular to the plane containing the c axis. As a result, circularly polarized light becomes elliptically polarized as it moves through the sample. The changes in polarization affect the relative amplitude of the x and y components of the incident light. The crucial point is that changes of Ix and Iy are different for left and right circular polarization so that any scattering that depends on only one of these terms will exhibit a CID spectrum. This result permits a very sensitive tool for aligning the samples for ROA measurements so that intrinsic birefringence effects can be eliminated. Raman optical activity of Magnons in FeF2 Figure 6 presents the conventional Raman and ROA spectrum of the one-magnon line of FeF2 in an applied magnetic field. The modest applied field, B < 0.6 T, is too small to observe the splitting of the magnon branches within the magnon line width using conven- tional (linearly polarized light) Raman techniques. However, the magnon line exhibits a clear CID spec- trum. To determine the energy difference between the two branches we fit the data using offset Gaussian line shapes to describe the one magnon scattering peak. In this manner we can calculate the energy difference as a function of temperature. Figure 7 shows the tempera- ture dependence of the frequency splitting between the two magnon branches in FeF2. As the temperature increases, the energy splitting between the branches decreases, while the linewidth increases resulting in a larger uncertainty for the energy splitting between the branches. Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides Fizika Nizkikh Temperatur, 2005, v. 31, Nos. 8/9 1037 IR IL 40 50 60 70 80 Energy, cm–1 3000 2000 1000 0 –1000 In te n si ty (c o n ts ) IR IL– FeF2 magnon B1g phonon Fig. 5. The ROA spectrum of FeF2 at 4 K. The presence of a CID spectrum in the B1g phonon mode indicates a mis- alignment of the incident laser and the crystalline c axis. 40 45 50 55 60 65 –400 0 400 800 1200 1600 2000 FeF2 28 K B = 0.59 T Frequency shift, cm –1 Conventional Raman CID R am an in te n si ty ,a rb . u n its Fig. 6. The low temperature ROA spectrum of FeF2 in the presence of an applied magnetic field. The upper trace shows the conventional Raman spectrum. The lower trace is the measured CID spectrum. 0 20 40 60 80 0.0 0.4 0.8 1.2 1.6 FeF 2 B = 0.59/0.55 T Temperature, K Theory Expt.M ag n o n sp lit tin g , c m – 1 Fig. 7. The temperature dependence of the magnon branch splitting obtained from ROA measurements in FeF2. The theoretical points are calculated using applied fields of 0.59 (filled diamonds) and 0.55 T (filled circles). The er- ror bars on the theoretical points derive largely from the uncertainty in the applied magnetic field strength. Theoretically, [28] the energy splitting between the two branches is given by �� �( ) [ ]| |T g J HB� 2 (5) where H is the applied magnetic field along the crys- tal c-axis, �|| is the parallel susceptibility per spin site, µB is the Bohr magneton, J is the dominant near- est neighbor exchange, and g is the Lande g factor. At low temperatures, �|| is nearly zero so: �� ( )T g HB� �0 2 (6) whereas at temperatures near TN, �|| = g B/2J so: �� ( )T T g HN B� � . (7) This theory predicts that the splitting between the branches should change by a factor of two when go- ing from low temperature to the phase transition tem- perature. Using parameters obtained from the experi- ment or the literature the theoretical curve shown in Fig. 7 was obtained. The theoretical curve accurately describes the temperature variation of the energy splitting between the magnon bands, but overall there is a general offset between the measured data and the theoretical curve. The primary uncertainty for this measurement arises from determining the ac- tual value of the magnetic field inside the refrigera- tor. The magnetic field of 0.59 T for the permanent magnet gap was measured at room temperature. The magnet will be at a much lower temperature during the Raman measurement, but the affects of the lower temperature on the gap field strength is not known. These measurements highlight the sensitivity of this technique in measuring magnon branch splittings. The line widths of the two peaks at 62 K were greater than 10 cm–1, but the separation between the peaks was only 0.4 cm–1. For a simple antiferromagnetic, we would not ex- pect to observe any CID spectrum in zero applied field ROA measurements. However, careful measurements revealed a small but repeatable CID spectrum in FeF2. Figure 8 shows the CID spectrum and the separate I R and I L Raman signals of FeF2. Care was taken to elim- inate all of the depolarization effects by using the Ray- leigh scattering spectrum to ensure alignment of the crystal c-axis with the incident laser light. The solid line fit determined the zero field splitting between the two magnon branches to be �� = (0.09 ± 0.02) cm–1. To our knowledge this result is the first measurement of a zero field splitting of the magnon branches in antiferromagnets using Raman scattering. Again the high sensitivity and resolution of this technique is highlighted in obtaining this result. Finally, ROA measurements of the two-magnon scattering peak in FeF2 showed a null result. The lack of any CID signal confirms that the dominant mecha- nism for two-magnon scattering involves exchange coupled magnons on opposite sublattices. Other possi- ble sources of two-magnon scattering are higher order processes that would be expected to be much weaker than the one-magnon scattering. Multiple spin-flip transitions have been observed in dilute magnetic semiconductors in the paramagnetic phase under reso- nance excitation conditions. These transitions were observed in a forward scattering geometry where the scattered light was analyzed for the circular polariza- tion as well. The intensities of the two spin flip signals are much lower than the single spin flip signals consis- tent with a higher order interaction. 6. Magnetic dipole–dipole interactions In 1963 Loudon and Pincus [25] examined the ef- fect of the classical dipole–dipole interaction between magnetic moments on the spin wave spectrum of a sim- ple uniaxial antiferromagnet. In the absence of an ap- plied magnetic field, as discussed above, the spin wave branches are degenerate in these systems. The magnon dispersion relation is given by � ��� H H H H b kA A E E /2 2 2 2 1 22 2 ] (8) where � = gµB is the gyromagnetic ratio, HA is the uniaxial single-ion internal anisotropy field and HE is the effective exchange field; k is the wave vector and b = az–1/2 with a the nearest neighbor distance and z the number of nearest neighbors. On inclusion of the 1038 Fizika Nizkikh Temperatur, 2005, v. 31, Nos. 8/9 K.R. Hoffman, D.J. Lockwood, and W.M. Yen 44 48 52 56 60 0 2 4 6 IR IL CID Frequency shift, cm –1 FeF2 26 K B = 0 R am an in te n si ty ,a rb . u n its Fig. 8. The zero-field ROA spectrum of the one magnon line in FeF2. The solid line is a fit modeling the small CID apparent in this spectrum. dipolar fields, Eq. (8) splits into two modes with their dispersions given by: � �1 2 2 2 2 1 22 2� [ ]H H H H b kA E A E / , (9) � � � �2 2 2 2 2 1 22 4� [ ( )( sin )]H H H b k H MA A E E s k / (10) where Ms is the sublattice magnetization and �k is the angle between the easy axis (z) of magnetization and the direction of propagation of the spin wave, k. Only the frequency of the second mode is affected by the dipolar interaction. For spin waves with k = 0, Eqs. (9) and (10) reduce to �� �1 1 22� [ )]H H HA A E / (11a) � � � �2 2 0 1 22 8� [ ( sin )]H H H MA A E s / . (11b) In the case of longitudinal spin waves, �0 = 0, and there is no dipolar effect. For the transverse spin waves, however, �0 = �/2, and the dipolar effects are at a maximum. Subsequently, Harris [26] has shown that Eq. (10) is an approximate result and has derived the full expres- sion for the spin wave dispersion in the presence of a dipolar field. However, the difference between Eq. (10) and the complete result only becomes important when 4�Ms is comparable to HE, which is not the case here. The Loudon and Pincus calculation is based on a molecular field approach that neglects the affect of the Lorentz field of one sublattice upon the other sublattice. When this effect is included, White [27] showed that the resonant frequencies for k = 0 are given by: �( ) ,k � �0 12 � � � � � � � � � � � � � � � �H H M H H M NA A s A E s x / 8 3 2 4 1 2 ( ) , (12a) �( ) ,k � �0 3 4 � � � � � � � � � � � � � � � �H H M H H M NA A s A E s y / 8 3 2 4 1 2 ( ) (12b) and for k ! 0 by � � �( ) ( ),k H H H HA A E E k12 2 2 22 1� � � � � � � 8 3 1 1 2 � �M H Hs A E k / [ ( )] , (13a) � � �( ) ( ),k H H H HA A E E k3 4 2 2 22 1� � � � � � � � � � � � � 8 1 3 12 1 2 � � �M H Hs k A E k / sin [ ( )] (13b) where N is the demagnetization dyadic [27], and �k is defined as (1/z) �� exp(ik•�), where the �’s are the vectors to the z nearest neighbors. Notice that for the uniform spin precession (k = 0) in a sphere (Nx = Ny = Nz = 1/3), the Lorentz field just cancels the surface demagnetizing field. There- fore, there is no shift in the resonant frequency as com- pared to non-dipolar cases. The symmetric occurrence of Nx and Ny in Eq. (12) is due to the fact that the normal modes labeled 1, 2 and 3, 4 are characterized by net transverse magnetizations that are linearly po- larized in the x and y directions, respectively. Notice also that although the 1, 2 spin wave modes are inde- pendent of � they are shifted by the Lorentz field. Since the samples used in the optical studies had rectangular cross sections in the basal plane, there will be dipolar splittings of the magnons. Thus, from Eq. (12), �� � � � k s A x y A E A M H N N H H H � � � 0 1 3 2 4 2 ( ) . (14) The demagnetizing factors are given by integrals over the surface [27]. In order to obtain estimates of these splittings we approximate Nx and Ny by 1/2 and 1/4, respectively, based on the sample dimen- sions. The parameter values used in the calculations and the resulting splittings are listed for MnF2, CoF2, and FeF2 in Table 2. 7. Comparison of theory with experiment Optical circular dichroism The calculated splittings given in Table 2 are quite small compared with the magnon energies in these compounds. Given the nature of the natural line widths, splittings of this magnitude are difficult to measure and they have been resolved only recently using Raman spectroscopy [11,15]. The calculated values for �� in MnF2 and CoF2 (see Table 2) are of similar magnitude to those found experimentally for the magnon sidebands in the visi- Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides Fizika Nizkikh Temperatur, 2005, v. 31, Nos. 8/9 1039 ble region: 0.02 cm–1 for CoF2 and 0.07 and 0.05 cm–1 for MnF2. Thus the origin of the observed splittings reported in the CD measurements is well encompassed by the effects of the magnetic dipole–dipole interac- tion in these simple rutile antiferromagnets. The size of the splittings detected by spontaneous magnetic CD cannot be compared directly with those calculated in Table 2, as the processes involved in the formation of magnon sidebands are not as simple and direct as those encountered in Raman scattering. For MnF2, for example, the sideband involves a combination of an exciton on one sublattice with essentially zero disper- sion [25] with a magnon on the other with nearly 50 cm–1 dispersion; k conservation requires the creation of exciton–magnon pairs with equal and opposite wave vectors from throughout the Brillouin zone. Al- though the splitting is mostly uniform across the Brillouin zone, the CD is, in effect, sampling the de- rivative of the joint density of states. It follows that an accurate calculation of the CD requires the use of Eq. (13), but the theory needs to be extended to in- clude specific symmetry points in the Brillouin zone and an appropriate summation over the entire zone. Conceivably, another possible cause of CD in these compounds could be a slight tilting of the spins away from their ordered positions along the c-axis. How- ever, at low temperatures and in zero applied magnetic field, all experimental results to date point to collin- ear spin alignment in all of these compounds [29]. Raman circular intensity difference In the Raman equivalent of the CD measurements discussed above, a �� of (0.09 ± 0.02) cm–1 was ob- served in FeF2 at low temperatures using a 90° scatter- ing geometry. From Eq. (11), the theoretical splitting for magnons propagating at approximately 45° to the c (z) axis is �� = 0.15 cm–1, which considering the ex- pected large uncertainty in Ms is in good agreement with the experimental results. Including the Lorenz field correction, Eq. (14) gives �� = 0.065 cm–1 (see Table 2), which is in even closer agreement with experiment. The zero-field splitting of the magnon branches observed in FeF2 is thus attributed to the magnetic dipole–dipole interactions that can be im- portant for long wavelength magnons [27]. Another way to quantitatively describe the interac- tion is to define it in terms of an effective anisotropy field, H", that is treated in the same manner as an exter- nal magnetic field. In this approach the energy of the magnons shows up in much the same form as presented in Eq. (6) except the splitting is now due to H". �� �( ) ' .T H� � � �0 2 0 09 1cm (15) Table 2. Magnetic parameters at low temperature for vari- ous transition metal fluorides possessing the rutile struc- ture. The calculated dipole–dipole induced splittings, ��k=0 and ��k = �/2, are also given. Parameter MnF 2 [28–30] FeF 2 [29,31] CoF 2 [29,32] Effective spin S 5/2 2 1/2 H E (T) 51.5 53.313 37.6 H A (T) 0.84 19.745 9.54 g 2.00 2.25 2.80 � (cm–1/T) 0.934 1.05 1.31 M S (T) 0.060 0.056 �0.06 �� k = 0 (cm–1) 0.016 0.065 0.083 �� k = �/2 (cm–1) 0.02 0.08 0.12 Using the parameter value � = 1.05 cm–1/T (Table 2) we calculate the anisotropy field to be: H" = (0.04 ± 0.01) T. By comparing Figs. 8 and 6 we see that the CID spectra have opposite signs for the case of H = H" and H = Happlied + H", respectively. Since the energy split- ting of the two magnetic sublattices switches sign with the external field, the direction of the applied field must be opposite to the direction of the effective anisotropy field. Going back to the temperature de- pendence of the magnon branch splitting we can now correct for the presence of the effective anisotropy field in the material. Equation (5) becomes �� �( ) [ ]( ')| |T g J H HB� �2 applied � 2 0 55[ ]( . ).| |g JB � T (16) The second theoretical curve shown on Fig. 7 uses this new value for the magnetic field and it gives better agreement with the measured data. 8. Conclusions The magnon sideband CD observed in rutile struc- ture antiferromagnets can be quite readily explained as being the result of magnetic dipole–dipole interac- tions breaking the degeneracy of the two spin wave branches. These optical measurements performed three decades ago are seen now to represent the first obser- vation of such splittings, as first predicted by Loudon and Pincus in 1963. It would be informative to per- form further experiments to confirm the sin2� depend- ence of the splitting by varying the angle �0 from 0 to �/2, i.e., the incident light direction is varied from along to perpendicular to the easy axis, and also the sample shape effect on the splitting. 1040 Fizika Nizkikh Temperatur, 2005, v. 31, Nos. 8/9 K.R. Hoffman, D.J. Lockwood, and W.M. Yen The ability to measure an effective anisotropy field in FeF2 highlights the sensitivity of ROA measure- ments for determining small splittings between the magnon branches in antiferromagnets. The measure- ment of such a splitting due to magnetic dipole–dipole interactions in FeF2 has permitted a final resolution of questions concerning the origins of CD in optical mea- surements completed over thirty years ago. The high sensitivity of ROA also permits the study of anti- ferromagnetic branch splittings for low applied mag- netic fields. The precise experimental work of Professor F.L. Scarpace and Dr. Y.H. (Russ) Wong from thirty years ago is noted with appreciation. We thank Professors R. M. White and M. G. Cottam for useful discussions and the National Science Foundation for continued support spanning these three decades. 1. R.L. Greene, D.D. Sell, W.M. Yen, A.L. Schawlow, and R.M. White, Phys. Rev. Lett. 15, 656 (1965). 2. J.W. Halley and I. Silvera, Phys. Rev. Lett. 15, 654 (1965). 3. Y. Tanabe, T. Moriya, and, S. Sugano, Phys. Rev. Lett. 15, 1023 (1965). 4. S.J. Allen, R. Loudon, and P.L. Richards, Phys Rev. Lett. 16, 643 (1966). 5. P.A. Fleury, S.P.S. Porto, L.E. Cheeseman, and H.J. Guggenheim, Phys. Rev. Lett. 17, 84 (1966). 6. P.A. Fleury and R. Loudon, Phys. Rev. 166, 514 (1968). 7. M.G. Cottam and D.J. Lockwood, Light Scattering in Magnetic Solids, Wiley, New York (1986). 8. F.L Scarpace, M.Y. Chen, and W.M Yen, J. Appl. Phys. 42, 1655 (1971). 9. M.Y. Chen, F.L. Scarpace, M.W. Passow, and W.M. Yen, Phys. Rev. B4, 1332 (1971). 10. Y.H. Wong, C.D. Pfeifer, and W.M Yen, in: Magne- tism and Magnetic Materials (1972) C.D. Graham, Jr. and J.J. Rhyne (eds.) AIP Conference Proc. No. 10 AIP, New York (1973) p. 1960. 11. D.J. Lockwood, R.M White, and W.M. Yen, Phys. Rev. B69, 174413 (2004). 12. W.M. Yen and Y. Oka, Bussei Kenkyu 13, 507 (1972). 13. L.D. Barron, M.P. Bogaard, and A.D. Buckingham, J. Am. Chem. Soc. 95, 603 (1973). 14. L.D. Barron and A.D. Buckingham, Mol. Phys. 20, 1111 (1971). 15. L.D. Barron, Molecular Light Scattering and Optical Activity, Cambridge University Press, Cambridge (1982). 16. L.D. Barron, L. Hecht, In: J. Nakanishi, N. Berova, and R.W. Woody (eds.) Circular Dichroism Princi- ples and Applications, VCH Publishers, New York (1994). 17. Petrou, D.L. Peterson, S. Venugopalan, R.R. Galazka, A.K. Ramdas, and S. Rodriguez, Phys. Rev. B27, 3471 (1983). 18. D.L. Peterson, D.U. Barthololmew, A.K. Ramdas, and S. Rodriguez, Phys. Rev. B31, 7932 (1985). 19. K.R. Hoffman, W. Jia, andW.M. Yen, Opt. Lett. 15, 332 (1990). 20. K.R. Hoffman, W. Jia, and W.M. Yen, in: J.R. Durig and J.F. Sullivan (eds.), Proceedings of the 12th In- ternational Conference on Raman Spectroscopy, Wiley, New York (1990), p. 864. 21. K.R. Hoffman, W.M. Yen, and D.J. Lockwood in: N.T. Yu and X.Y. Li (eds.), Proceedings of the 14th International Conference on Raman Spectroscopy, Wiley, Chichester (1994), p. 1078. 22. K.R. Hoffman, W.M. Yen, and D.J. Lockwood in: W. Kiefer et. al. (eds.), Proceedings of the 13th Interna- tional Conference on Raman Spectroscopy, Wiley, Chichester (1992), p. 766. 23. K.R. Hoffman, W.M. Yen, D.J. Lockwood, and P.E. Sulewski, Phys. Rev. B49, 182 (1994). 24. D.J. Lockwood, K.R. Hoffman, and W.M. Yen, J. Lumin. 100, 147 (2002). 25. R. Loudon and P. Pincus, Phys. Rev. 132, 673 (1963). 26. A.B. Harris, Phys. Rev. 143, 353 (1966). 27. M.G. Cottam and D.J. Lockwood, Light Scattering in Magnetic Solids, Wiley, New York, (1986) Chapter 5. 28. D.D. Sell, R.L. Greene, and R.M. White, Phys. Rev. 158, 149 (1967). 29. R.A. Cowley and W.J.L. Buyers, Rev. Mod. Phys. 44, 406 (1972). 30. J.P. Kotthaus and V. Jaccarino, Phys. Rev. Lett. 28, 1649 (1972). 31. T. Dumelow, R.E. Camley, K. Abraha, and D.R. Til- ley, Phys. Rev. B58, 897 (1998). 32.P.L. Richards, J. Appl. Phys. 35, 850 (1964). Circular dichroism and Raman optical activity in antiferromagnetic transition metal fluorides Fizika Nizkikh Temperatur, 2005, v. 31, Nos. 8/9 1041