Mar 5, 2013 - Nee1 temperature down to 22.5 K, the magnetic structure of NiBr, is that of a simple meta- magnet, with sheets of ferromagnetically coupled ...
Home
Search
Collections
Journals
About
Contact us
My IOPscience
Optical and neutron diffraction study of the magnetic phase diagram of NiBr2
This article has been downloaded from IOPscience. Please scroll down to see the full text article. 1976 J. Phys. C: Solid State Phys. 9 2481 (http://iopscience.iop.org/0022-3719/9/13/008) View the table of contents for this issue, or go to the journal homepage for more
Download details: IP Address: 139.143.5.160 The article was downloaded on 03/05/2013 at 15:11
Please note that terms and conditions apply.
J. Phys. C: Solid State Phys., Vol. 9, 1976. Printed in Great Britain. @ 1976
Optical and neutron diffraction study of the magnetic phase diagram of NiBr, P Day, A Dinsdale, E R Krausz? and D J RobbinsJ Oxford University, Inorganic Chemistry Laboratory, South Parks Road, Oxford OX1 3QR, UK
Received 3 March 1976
Abstract. The presence of a magnetic phase transition at 22.5 K in NiBr, (T, = 52 K) has been confirmed by neutron diffraction and by the appearance of new sharp electric-dipole absorption lines in the optical spectrum near 6080A. The latter may be ‘switched off by an external magnetic field parallel to the basal plane equal to 2.75 T at 1.3 K. By measuring the variation with temperature of this critical field, we have established the phase diagram of NiBr, in the HT plane. Powder neutron diffraction measurements confirm that, from the Nee1 temperature down to 22.5 K, the magnetic structure of NiBr, is that of a simple metamagnet, with sheets of ferromagnetically coupled spins in turn coupled together antiferromagnetically, the spins lying in the basal plane. The temperature dependence of the intensity of the (009) magnetic peak in the single-crystal neutron diffraction show that, below 22.5 K, the spins become canted out of the basal plane by about 54”.
1. Introduction
The optical and magneto-optical properties of a number of the binary dihalides of the 3d elements have recently come under scrutiny because such compounds were among the original prototype examples of ‘metamagnetic’ behaviour, that is, antiferromagnets which may nevertheless be ferromagnetically saturated in external magnetic fields. In the Fe and CO dihalides, for example, this type of magnetic behaviour arises from a combination of relatively strong ferromagnetic exchange between near-neighbour ions and much weaker antiferromagnetic exchange between next-nearest neighbours, the result of the CdC1,-type layer structure of the crystals. Recently we showed (Robbins and Day 1976) that the temperature dependence of the visible exciton-magnon absorption bands in this class of compound could be explained if one assumed that excitons corresponding to spin-forbidden (e.g. quintet-to-triplet) ligand field transitions coupled only to thermally populated magnons propagating within the ferromagnetic layers. To a first approximation the compounds therefore behave optically as if they were ferromagnets. In our survey we found that ‘hot-band’ (Shinagawa and Tanabe 1971) excitonmagnon bands indeed dominate the visible absorption in FeX, and COX, (X = C1, Br). f Present dddress: Chemistry Department, University of Virginia, Charlottesville, Virginia, USA
1 Present address: Royal Radar Establishment, St Andrew’s Road, Great Malvern, Worcs, UK. 2481 D13
2482
P Day, A Dinsdale, E R Krausz and D J Robbins
On the other hand, in NiBr, we found that although the major part of the absorption intensity of the triplet-to-singlet transition near 6000A varied within experimental error as T 2 (a result which is explained very straightforwardly by assuming that the compound is behaving as an easy-plane ferromagnet with small single-ion anisotropy), a pair of very sharp electric-dipole lines in the same spectral region actually increased in intensity with decreasing temperature. It appeared to us that these lines might imply that NiBr, had a more complicated magnetic structure than the Fe and CO dihalides. We have therefore examined them more closely as a function of temperature and external magnetic field. At the same time, as no neutron diffraction work has ever been reported on NiBr,, we also report some preliminary observations on the magnetic structure using data obtained from powders and single crystals. The Nee1 temperature of NiBr, has been variously reported as 60K (Tsubokawa 1960) and 54 K (Katsumata and Date 1969, Motimoto and Date 1971) from the susceptibility and 44.4 K from paramagnetic resonance (Katsumata 1971). Antiferromagnetic resonance experiments also confirm that the spins are aligned ferromagnetically in the planes perpendicular to the crystalline c axis, the planes being coupled antiferromagnetically (Katsumata and Date 1969). However, below about 20 K, the antiferromagnetic resonance (AFR) lines broaden and split, indicating a transition to a more complicated magnetic structure. Further evidence that the low-temperature magnetic structure may be more complicated than that of a simple metamagnet comes from the field dependence of the magnetization at 4.2 K (Morimoto and Date 1971). At about 2.7 T, applied in the basal plane, the magnetization increases abruptly by 30 ”/, suggesting that at lower fields the spins may be canted somewhat away from this plane. The 64.87 GHz AFR line, which is found at 2.7T at 4.3 K when the field is applied in the basal plane, also shifts strongly towards higher fields when the field is rotated towards the c axis. Finally, there is some indication that the structural phase relationships of NiBr, themselves may not be entirely straightforward. Ketelaar (1933) found that whilst NiBr, prepared by sublimation had the CdC1, structure, samples obtained by evaporating solutions of NiCO, in aqueous HBr or by recrystallizing from dry ethanol had what he called a ‘wechselstruktur’. The latter, described by Bijvoet and Nieuwenkamp (1933) as the CdBr, structure, is best described as a disordered intergrowth of the CdC1, and CdI, lattices. Layer lattices like that of CdI, itself are notorious for forming polytypes by ordering stacking faults (e.g. Prasad 19741 although the effect of such polytypism on magnetic and optical properties has a apparently never been examined. It has, however, been shown that disordered structures like the ‘wechelstruktur’ can easily be induced by strain, which introduces dislocations (Price and Nadeau 1962). Consequently, it is worth examining the structure of powdered material prepared from solution as well as the melt-grown crystals used for the optical experiments. 2. Experimental
Nickel bromide was 5N anhydrous material from Alfa Inorganics. As supplied, it gave the x-ray powder diffraction pattern of the ‘wechselstruktur’ (see below). It was further purified by sublimation under vacuum and crystals were grown by the Bridgman method in a Metals Research BCG furnace. Being hygroscopic, both powder and crystals were handled exclusively under an atmosphere of dry nitrogen. Neutron diffraction measurements were carried out at AERE, Harwell, the powder scans on the Dido Curran diffractometer at an incident neutron wavelength of 1.368A
Magnetic phase diagram of NiBr,
2483
and the single-crystal experiments on the Mark VI no 2 instrument at a wavelength of 1.0903 8,. In both cases the samples were mounted in pressurized liquid-helium cryostats, the temperature being recorded either by a germanium (4-40K) or platinum (40-300 K) resistance thermometer. The sample temperature could be controlled to about L 0.1 K. For the temperature-dependence measurements, a crystal of about 10 x 3 x l m m 3 was mounted with its a and c axes in the horizontal plane. The high-resolution visible absorption spectra were recorded with the incident light beam propagating along the c axis, using a crystal of approximately 10 x 5 x 1 mm3 dimensions. The measuring equipment consisted of a McPherson RS 10 spectrophotometer, used in its single beam mode, to which had been interfaced a Thor Cryogenics split-coil superconducting magnet. Spectra were conventionally recorded at a spectral bandpass of O . l 0 4 1 5 A , i.e. approximately 0.3 cm-’ at 6OOOA. The temperature of the 1 by a Thor temperature controller and sample was controlled to better than ~ 0 . K a carbon resistance thermometer. Measurements of the variation of the optical spectrum with temperature in the absence of a magnetic field were made using the McPherson spectrophotometer in its double-beam mode, with the sample contained in an Oxford Instruments CF 100 continuous-flow helium cryostat, together with an Oxford Instruments temperature controller. Some preliminary measurements of the magnetic-field dependence of the spectrum were made using a 3.4m spectrograph and Bitter magnet in the Clarendon Laboratory, Oxford, through the courtesy of Dr M J M Leask.
3. Neutron diffraction 3.1. Powder results Preliminary scans at room temperature and liquid-helium temperature on unsublimed powder as received from Alfa Inorganics showed that it had the ‘wechselstruktur’, indexing on a hexagonal cell (a = 2.10 c = 6.12 A) containing molecule per cell. All reflections were very broad. More precise scans were then made over the 28 range from 3” to 128 on material prepared by sublimation and also on powder obtained by crushing melt-grown single crystals. Temperatures of 77 K (above the Neel temperature), 30 K (below the Nkel temperature but above the second magnetic transition) and 6.25 K (below the second magnetic transition) were employed. In all cases the complete set of 30 Bragg peaks were indexable on a CdC1, (Did)structure with the following hexagonal unit cell parameters:
77 K : 30 K : 6.25 K :
a = 364.5 8,
3,645 8, 3.640 8,
c = 18.278, 18.24 8, 18.23 A.
In addition, at 300 K, powder x-ray diffraction gave a = 3.71 k 0.01 8, and c = 18.30 rt 0.04 A. The 30 K profile contained a number of magnetic peaks, indexed as (001) and (101) in a unit cell with twice the volume of the chemical unit cell, formed by doubling the c-axis parameter. Thus it is confirmed that, from the Neel temperature down to the lower phase-transition temperature, the magnetic structure consists of ferromagnetic sheets of spins with successive sheets antiferromagnetically aligned (i.e. the same structure as that of NiCl, (Lindgard et al 1975)). In the 6 5 K scan no further magnetic peaks could be seen which were not present in the 30K scan, although some variations in relative intensities were noticed.
P Day, A Dinsdale, E R Krausz and D J Robbins
2484
3.2. Sing le-crystal resu Its
Detailed measurements were made of the variation of intensity and bandshape of the (009) and (107) magnetic peaks with temperatures from 50 K to 6.5 K. Measurements at a smaller number of selected temperatures with the same range were also made on 605) and (101). Qualitatively they behaved in the same way as (009) and (107). The (300) and (003) nuclear peaks were used to monitor possible variations in unit cell constants, crystal position, orientation and Debye-Waller factors. In fact, no variation either in peak shape or intensity of the nuclear peaks could be detected, and they retained a Gaussian shape, indicating that extinction was negligible.
CO-
O
1
20
30
LO
50
Temperature 1 K I
Figure 1. Intensity of the (009) magnetic peak in the single-crystal neutron diffraction of NiBr,. The full curve is the intensity calculated from the Brillouin function for S = 1 and Tv = 5 2 K .
21 0
10
20
30
LO
Temperature I K I
Figure 2. Temperature variation of the magnetic interaction vector 8 for the (009) magnetic peak in NiBr,.
Figure 1 shows the intensity of the (009) magnetic peak plotted as a function of temperature. Down to 30K the intensity closely follows the square of the sublattice magnetization calculated from a Brillouin function for S = 1. Such a function would be expected to give a good account of the magnetization since the ground state is an orbital singlet and the zero-field splitting will thus be small (for example, in NiCl,, Katsumata and Yamasaka (1973) found a value of 0.40K). Below 30K, on the other hand, the intensity of the magnetic peak falls rapidly, becoming more or less constant from 20 K down. Accompanying the loss of intensity below 30 K is a marked change in the shape, the profile in 26 becoming broader and flat-topped. A similar variation takes place in the (107) peak. The lowering in intensity of the two magnetic peaks below 2&25K can be most simply understood if we assume that at the phase transition which takes place in this temperature range the spins, which were previously constrained to the basal plane, become canted towards the c axis. The canting angle can be found quite simply from the (001) magnetic peaks by calculating the change in the magnetic interaction vector q, since F Z = q2Fiagn. In fact, if a is the angle the magnetization vector makes with the
Magnetic phase diagram of NiBr,
248 5
scattering vector, q2 = sin's. Thus the observed intensity is proportional to sin'tl. In figure 2 we therefore plot q2 = I/lcalcagainst temperature, where I is the observed intensity of the (009) magnetic reflection and Icalcis the intensity calculated from the Brillouin function for J = 1. The limiting value of q2 at zero temperature is 0.34, which defines a value for tl of 35.6". Of course, this angle only has physical significance if all the spins are directed out of the plane by the same angle, and in all other cases would represent the root mean square of the sine of the angle between the magnetiiation vector and the scattering vector. It is also worth noticing that the value derived in this way for q is quite close to the one which would be calculated from the 'step' occurring at 2.75 T in the 4.2 K magnetization curve of NiBr, when the magnetic field is applied in the basal plane. From the published curve we estimate q 0.6, thus lending further support to the idea that the external field induces a phase transition from the canted to a normal metamagnetic structure. We shall see below that the optical determination of the magnetic phase diagram provides final confirmation of this hypothesis. In conclusion, figure 2 tells us that, with no external magnetic field, the phase transition occurs at 22.5 t 0.5 K. 4. Optical properties
4.1. Temperature variation of the absorption spectrum Low-resolution surveys of the ligand-field transitions of NiBr, have been published by Ackerman et a1 (1972) and Kozielski et a1 (1972), and in general agree with results obtained in this laboratory (Adams 1972). The spin-allowed transitions remain rather broad even at low temperatures, and the main effect of temperature on them is simply to decrease their inteiisity somewhat as a result of cooling out absorption due to 'hot' exciton-phonon combinations. In contrast, there is a weak band system near 6000 A
26K 22 5K 18 K
1
6050
5K
ILK
I
6060
I 6070
I
6080
6d90
A Figure 3. Temperature variation of the axial absorption spectrum of NiBr, in the 6000A region.
2486
P Day, A Dinsdale, E R Krausz and D J Robbins
%\
'1
\O\ 0
0
2'0
1'0 Temperature
iK)
Figure 4. Temperature variation of the combined intensity of the 6080.2 and 6082.9A bands in NiBr,.
which shows a wealth of phonon sideband structure and has been assigned to a spinforbidden transition, either to 'A, or a component of 'T,. The axial absorption spectrpm of the zero-phonon portion of this latter band system is shown in figure 3. Its principal features are (i) a relatively broad (halfwidth 15 A) and asymmetric band centred near 6070 A whose intensity increases with temperature, and (ii) a pair of sharp bands (halfwidths about 3 cm- ') at 6080.2 and 6082.9 A which decrease in intensity with increasing temperature. From the g- and n-spectra, it is clear that all these bands are electric-dipole in character. The sharp bands appear only when the electric vector is perpendicular to the c axis. After resolving the curves into components using a Dupont Curve Analyser, it was found that the area of the broad band varied as T2(Robbinsand Day 1976). Thus it should be assigned as a 'hot' exciton-magnon combination. Because the other bands are so narrow, their individual areas cannot be determined so precisely, but their total integrated area is plotted against temperature in figure 4. They finally disappear at about 22 K, the temperature of the phase transition found by neutron diffraction. Another feature which can be seen qualitatively from figure 3 is that the two sharp lines coalesce as their intensity diminishes, though their halfwidths remain unchanged. This
0
10 Temperature I K i
20
Figure 5. Temperature variation of the wavelengths of the two cold bands.
Magnetic phase diagram of NiBr,
2487
3-
4.2. Magnetic-field dependence of the absorption spectrum The effect on the more intense of the two sharp ‘cold‘ lines of applying a magnetic field parallel to the ab plane of the crystal is shown in figure 6 for a temperature of 1.4 K. A most striking phenomenon is observed: at 2.75 T both sharp Iines vanish abruptly. The sharpness of the transition appears limited only by the small inhomogeneity in the field averaged over the sample volume. Since 2.75T is the field at which the ‘step’ occurs in the magnetization, corresponding to the return of the spins from their canted angle of 54“ back into the basal plane, the experiment is sufficient lo demonstrate that the sharp ‘cold’ exciton-magnon lines gain their intensity from an antiferromagnetic component directed along the c axis in the low-temperature phase. Clearly, the disappearance of the two sharp lines is a very sensitive methcd of monitoring the field-induced phase transition. We have therefore used it to determine the magnetic phase diagram of NiBr, in the NTplane. At a fixed temperature, we take the field at which the more intense of the two sharp lines is reduced to half its original height as representing the phase boundary along the H axis, and plot this field as a function of temperature in figure 7. The result is a smooth curve, descending to approximately 22 K in zero field. Consequently, the magnetic phase transition occurring at 22.5K in zero field, which we have established by neutron diffraction, is one and the same as the transition induced at lower temperatures by applying a magnetic field in the basal plane.
2488
P Day, A Dinsdale, E R Krausz and D J Robbins
4.3. Magnetic circular dichroism
Addition of a photoelastic modulator and lock-in amplifier to the split-coil magnet and McPherson monochromator converts the system to a high-resolution magnetic circular dichroism (MCD) or magnetic linear dichroism instrument. MCD spectra were therefore recorded in the 6000 A region at several temperatures below and above 22.5 K at a field of 5 T applied, for this purpose, parallel to the c axis of the crystal. At 4 K, 10 K and 20K only MCD due to the two major sharp lines could be detected (figure 8) but at
6070
6090
6080
Wavelength
[Ai
Figure 8. Magnetic circular dichroism of the cold bands.
higher temperatures the broader, though very much weaker, signal of the 'hot' excitonmagnon band appears. From figure 8 it is evident that the higher-energy and weaker of the two sharp lines has a pseudo A-term (Buckingham and Stephens 1966) while the other shows no sign of a derivative bandshape. However, without more detailed information either on the precise magnetic structure of NiBr, below 22 K or on the assignment of the electronic state responsible for the band system, nothing further can be said at present. 5. Conclusions
The main conclusions from the optical and neutron experiments described in this paper are the following: (i) The crystal structure of sublimed or melt-grown NiBr, is the CdCl, structure, and the Neel temperature is 52 i 1 K. (ii) From the Neel temperature down to 22.5 K, the magnetic structure consists of sheets of ferromagnetically coupled spins lying in the basal plane, successive sheets being coupled together antiferromagnetically. (iii) At 22.5 K a phase transition occurs to a more complicated magnetic structure in which the spins become canted out of the basal plane by a root-mean-square angle of 54.4". There is no significant change in the structural parameters. (iv) Optically, the low-temperature magnetic phase is characterized by the appear-
Magnetic phase diagram of NiBr,
2489
ance of two exceptionally narrow electric-dipole-allowed exciton-magnon combination bands near 6080 A in addition to a broader exciton-magnon band centred near 6070 A whose intensity varies as T 2 . (v) The two sharp lines, which occur exclusively when the electric vector of the incident light is parallel to the basal plane, may be ‘switched off by applying an external magnetic field of 2.75 T at 1.3 K parallel to the same plane. This corresponds to a return to the simple metainagnetic structure found at zero field between 22.5 and 52 K. (vi) By monitoring the field at which the two sharp lines switch off as a function of temperature, the boundary of the low-temperature phase can be established in the HT plane. Further neutron diffraction work is going to be needed to establish the full magnetic structure of the low-temperature phase. There are indications that it may be helical, possibly with a repeat distance along the c axis which varies with temperature. Nor is the precise assignment of the optical transitions yet clear. That the lines appearing in thc low-temperature phase are so sharp argues an extremely small dispersion both of the exciton and magnon states. That they have a ‘cold’ temperature dependence further suggests that they arise from an antiferromagnetic component of coupling within the layers brought about when the spins become canted out of the basal plane. At the same time, even in the low-temperature phase, the dominant contribution to the exchange interaction within the layers must remain ferromagnetic because most of the electricdipole intensity in the spectral region we have been examining resides in the ‘hot’ exciton-magnon band whose variation with temperature is what one would anticipate for an easy-plane ferromagnet with small single-ion anisotropy (Robbins and Day 1976, A K Gregson, P Day, A Okiji and R J Elliott 1976 unpublished). Finally, it does appear however, that in cases such as this, optical spectroscopy may prove a very sensitive tool for establishing phase diagrams of magnetic insulators. Further experiments of this type will be reported in due course. Acknowledgments
We are grateful to Dr M J M Leask for the use of magneto-optical equipment in the Clarendon Laboratory, to Dr A K Gregson for carrying out the preliminary powder neutron diffraction measurements, to the Science Research Council for an equipment grant and a Fellowship (to DJR), and to ICI Ltd for a Research Fellowship (to ERK).
References Ackerman J, Fouassier C, Holt E M and Holt S L 1972 Inorg. Chem. 11 3118 Adams L P 1972 Chemistry Part II Thesis Oxford University Bijvoet R J M and Nieuwenkamp 1933 Z . Kristallogr. 86 466. Buckingham A D and Stephens P J 1966 Ann. Rev. Phys. Chem. Katsumata K 1971 J . Phys. Soc. Japan 30 1498 Katsumata K and Date M 1969 J . Phys. Soc. Japan 27 1360 Katsumata K and Yamasaka K 1973 J . Phys. Soc. Japan 34 346 Ketelaar J A A 1934 Z. Kristallogr. 88 26 Kozielski M, Pollini I and Spinolo G 1972 J . Phys. C : Solid St. Phys. 5 1253 Lindgard P A, Birgeneau R J, Als-Nielsen J and Guggenheim H J 1975 J . Phys. C : Solid St. Phys. 8 1059 Morimoto M and Date M 1971 J . Phys. Soc. Jupan 29 1090
2490
P Day, A Dinsdale, E R Krausz and D J Robbins
Prasad R 1974 2. Kristallogr. 139 136 Price P B and Nadeau J S 1962 J . Appl. Phys. 33 1543 Kobbins D J and Day P 1976 J . Phys. C: Solid Si. Phys. 9 867 Shiriagawa K and Tanabe Y 1971 J . Phys. Soc. Japan 30 1280 Tsubokawa I 1960 J . Phys. Soc. Japan 15 2109