An EPR study of a new low-temperature phase transition in KLiSO4:MoO43-

Article

Reference

An EPR study of a new low-temperature phase transition in KLiSO4:MoO43-

BILL, Hans, RAVI SEKHAR, Y., LOVY, Dominique

Abstract

A new phase transition is observed in the KLiSO4 crystals below 77 K using MoO43- as a paramagnetic probe. The EPR spectra of this molecular ion, which substitutes for the host SO42- ion, show triclinic symmetry at 77 K due to the joint action of a trigonal crystal field and a trigonal Jahn-Teller effect of E(X)e type. Above 77 K, motional averaging in the a-b plane occurs; thereby an axial spectrum is observed. Below 64 K, the two sulphate sites are no longer equivalent resulting in the appearance of two distinct EPR spectra of the molybdate ion. The possible space groups of the crystal in the vicinity of the phase transition are discussed. The authors' results indicate that one needs to distinguish carefully between the local dynamics and the structural changes in this crystal.

BILL, Hans, RAVI SEKHAR, Y., LOVY, Dominique. An EPR study of a new low-temperature phase transition in KLiSO4:MoO43-. Journal of Physics. C, Solid State Physics , 1988, vol.

21, no. 15, p. 2795-2804

DOI : 10.1088/0022-3719/21/15/012

Available at:

http://archive-ouverte.unige.ch/unige:3084

Disclaimer: layout of this document may differ from the published version.

1 / 1

J . Phys. C: Solid State Phys. 21 (1988) 2795-2804. Printed in the UK

An

^EPR

study of a new low-temperature phase transition in KL~SO,:MOO,~-

H Bill, Y Ravi Sekhar and D Lovy

Groupe Physico-Chimie d u Solide, Sciences 11, UniversitC d e Genkve, 30 quai E Ansermet. CH-1211 Geneva 4. Switzerland

Received 28 October 1987. in final form 7 December 1987

Abstract. A new phase transition is observed in the KLiSOJ crystals below 77 K using MOO,'- as a paramagnetic probe. The EPR spectra of this molecular ion, which substitutes for the host S 0 4 2 - ion, show a triclinic symmetry at 77 K due to the joint action of a trigonal crystal field and a trigonal Jahn-Teller effect of E 8 e type, Above 77 K , motional averaging in the a-b plane occurs; thereby an axial spectrum is observed. Below 64 K , the two sulphate sites are no longer equivalent resulting in the appearance of two distinct EPR spectra of the molybdate ion. T h e possible space groups of the crystal in the vicinity of the phase transition are discussed. O u r results indicate that one needs to distinguish carefully between the local dynamics and the structural changes in this crystal.

1. Introduction

Lithium potassium sulphate (LPS) is known to undergo a variety of structural phase transitions, both above and below room temperature. Their order and even their sequence is still highly controversial (e.g. Hoiuz and Drozdowski 1981, Balagurov et a1 1986). Most publications agree, however, that the crystal at room temperature (RT)

belongs to the hexagonal space group P63 ( C i ) (Karppinen eta1 1983) with two molecules per unit cell. Raman measurements below RT have shown the occurrence of a phase transition at around 201 K (while cooling) involving a change in the space group from P63 to P31c (Bausal et a1 1980). This transition was interpreted as being associated with a cooperative reorientation of sulphate ions. Recently published optical birefringence results (Sorge and Hempei 1986) on this host are in agreement with this assignment but find a transition temperature of 190 K. In addition, they indicate a phase transition to monoclinic symmetry at 83 K. On the other hand EPR experiments (Holuz and Drozdowski 1981, Fonseca et a1 1983, Shibata er a1 1986) using the SO4- radical as a probe have indicated the presence of two phase transitions at around 250 and 181 K.

There is further some controversy about the existence of an incommensurate phase between 226 and 181 K. While Fonseca and co-workers have observed asymmetric EPR

lineshapes characteristic of an incommensurate system at 181 K , no clear-cut change of the EPR spectrum has been found by Shibata and co-workers.

In addition to the phase transitions at temperatures above 80 K at least two more have been detected at about 65 and 38 K with the aid of heat capacity and DSC analysis experiments (Abello et a1 1985). There is, however, no spectroscopic evidence so far.

0022-3719/88/152795

+

10 $02.50

0

1988 IOP Publishing Ltd 2795

C15

2796 H Bill et a1

The first aim of this paper was to present spectroscopic results regarding the last- mentioned transitions below 77 K, but the following matter added supplementary jus- tification to this.

A more recent EPR study (Sastry et a1 1987) points out that the diverging inter- pretations of the phase transitions between RT and 80 K arise in part from the fact that the EPR probe used (SO,-) shows local dynamics. Depending on the timescale of the

EPR experiment compared with the local reorientation time one observes either dynamic or static behaviour-which is further a function of temperature. Unfortunately, the published EPR papers do not give the principal values and the orientation (the six components) of the g tensor. They give the values along the c , a and b axes ( b is perpendicular to the two crystallographic axes c and a ) . The angular dependence in one or two planes is further given. This is not enough to determine the orientation of the g tensor where g = (gtrg)li2. This point is of importance for a full investigation of the possibility whether the local motional effects of the SO4- explain all or part of the discrepancies mentioned.

Note that the LPS host is well suited for the study of the anionic and the cationic sublattices as it is possible to introduce the isovalent ions like Cu+ and Ag+ into the former and Moo4’- into the latter one. We preseat here EPR results or irradiated LiKSOj : MOO,’- crystals at low temperatures and identify the centres observed in these samples. Unlike the SO4 ^- probe, this transition metal oxyanion has larger g- anisotropy making it a more sensitive probe to study the structural changes. It is important, however, to keep in mind the possibility that impurities may have important effects on the vibrational dynamics of a host.

2. Experimental details

Single crystals of LPS doped with molybdate ions were grown by slow evaporation of aqueous solutions containing Li2S04 and K 2 S 0 4 in the stoichiometric ratio. The as- grown crystals were found to develop two shapes. One type formed hexagonal platelets which showed no anisotropy of the index of refraction while rotated in this plane under a microscope with crossed polarisers. The other variety showed some resemblance with quartz crystal fingers but often of the form of rectangular prisms. These samples exhibited under the microscope with crossed polarisers anisotropic behaviour of the index of refraction. We selected the hexagonal platelets in all of the experiments (but did not check for possible twinning related to the merohedry in the room temperature phase).

The crystals were x-rayed at 77 K (Philips fluorescence tube with W anode, typically 40 kV, 30 mA. 20 h) and transferred into the EPR Dewar without any warm-up. EPR

spectra were recorded at and below this temperature on a Varian spectrometer.

3. Results

3.1. Experiments at 78 K

Figure 1 shows the EPR spectra when the magnetic field H is oriented along the c and b axes of the crystal, respectively. The lines with g around 2.0090 in figure l(a) are due to host lattice radicals (Holuz and Drozdowski 1981, Fonseca et a1 1983). In the

Low-temperature phase transition in K L i S 0 4 2797

I a' 9 . 2 0094

Figure 1. ( a ) X band EPR spectrum of LiKSO, : MOO,'- crystals after x-irradiation at 77 K . H is parallel to the c axis and T = 77 K. The low-field lines around g = 2.0093 arise from the host matrix. ( b ) The EPR spectrum when H i s parallel to the b axis of the crystals.

Only the high-field region is shown on an expanded scale.

high-field region one observes, in addition to the intense line, six weak components centred approximately on this former one. They can also be seen to accompany the line at g = 1.9604 in figure l ( b ) . Their splitting is about 22 G (intensity approximately 4 % ) when H is parallel to c and is found to be 12 G with H parallel to b. They clearly arise as a result of the hyperfine interaction with one molybdenum nucleus (MO isotopes: 75.5% with I = 0; "MO : I =

3,

g, = -0.3656. 15.9%; "MO: I = % ^{- 9} g n =

-0.3734, 9.6%). As the satellites were difficult to observe with H pointing along a general direction no attempt was made to determine the complete hyperfine structure parameters.

Figure 2 shows the angular dependence of the main EPR lines in the a-b and b-c planes. The pattern in the a-b plane exhibits 60" repetition. Further, all the lines coalesce when Hllc, the symmetry axis. From the number of components (6) (and the orientation of the g tensor, see below) observed for a general orientation of H it is

2798 H

Bill

et a1

e ldegl

Figure 2. T h e angular variation of the high-field lines in the a-b and b-c planes at 77 K.

clear that the magnetic centre has triclinic symmetry. When the P63 or P31c crystal symmetry group proposed in the literature is taken into account, it is necessary to determine absolutely the orientation of the extrema in the angular plot of the a-b plane, with respect to the a axes of the macroscopic crystal. This was performed in the following way.

A well developed hexagonal LPS : Moo4*- crystal was glued on a flat Plexiglass holder in the cavity in such a way that its a axis was parallel to a fourfold axis of a CaFz crystal previously oriented by x-rays and ground to exact shape and orientation.

Then, the whole was x-rayed at 7 8 K and transferred into the EPR Dewar without warming it up.

The EPR spectrum observed on this system consisted of the lines due to the and the spectrum of the

Vk

centre in CaF2. This latter spectrum was used to align the H field very precisely parallel to the fourfold crystal axis-and thus to the a axis of the LPS crystal. This experiment showed that 0" of the a-b plane angular plot (figure 2) means that

H

is parallel to a macroscopic a axis of the LPS crystal (error 2 0.2"). The principal axes and the eigenvalues of the g tensor have been obtained from these measurements with the aid of a conventional triclinic Zeeman spin-Hamiltonian. The results are listed in table 1. They are related to the a and c crystal axes together with the b axis (orthogonal to a and c ) . The continuous lines in figure 1 represent the fit calculated with the results of table 1. The fact that only one set ofg-ualues represents the full angular pattern conforms well to a three- or sixfold symmetry (with respect to c) of the crystal structure in conjunction with the triclinic symmetry of the centre.

The g tensor is nearly axial. The principal axis (g = 1.9832) forms an angle of = 1.2" with respect to the a-c plane. The intersection between the magnetic moment

Low-temperature phase transition in K L i S O , 2799 Table 1. The spin-Hamiltonian parameters of Mood3- and SO4- in LiKS04 crystals at various temperatures.

Euler angles WRT abc system

Principal g-values ( d e d

T ( K ) gl 22 g3 Q : e

w

MOO^^- 120 1.9393 1.9393 1.9231 0.0 0.0 0.0 78 1.9832 1.9132 1.9035 1.2 71.1 -44.7 60 1.9842 1.9190 1.8987 0.0 69.6 -44.4 (site 1) 60 1.9848 1.9169 1.9023 5.5 77.6 34.5 (site 2)

so,-

^{300 2.0294 2.0044 2.0198 0.0 0.0 0.0}

77 2.0410 2.0023 2.0078 0.0 82/98 0.0

and the a-b plane yields a g-ellipse having its principal axes parallel to a and b , respectively.

In order to connect our experiments with the published (but incomplete) EPR data on SO4 ^- we have measured the angular dependence of this radical in the b-c plane at 78 K. It turns out that the g tensor has monoclinic symmetry within the linewidths.

The principal values are given in table 1. The symmetry plane is the a-c plane. The radical shows an orthorhombic EPR spectrum at room temperature with almost the same g, value (along b ) as at 78 K but with g, and g, equal to the values expected for partial'motional averaging of the 7 8 K spectrum in the a-c plane (by an angle of approximately 39", Sastry er a1 1987).

3 . 2 . Experiments as a function of the temperature

3.2.1. T > 78 K . When the crystal, after x-irradiation at 78 K, is warmed up to a temperature of approximately 120 K the triclinic EPR spectrum vanishes and an iso- tropic line begins to emerge when the magnetic field is in the a-b plane. Because of low signal intensity it was not possible to determine unambiguously the evolution of the EPR line with H

11

c. But it seems that the spectrum is axially symmetric with respect to c. This spectrum broadens when T is further increased. U p to temperatures of approximately 150 K the process is reversible as a function of temperature. This finding is a clear indication of locally acting motional effects at these temperatures. A t even higher temperatures the centre changes irreversibly its valence state. The relevant spectra are summarised in figure 3. The arrow indicates the motionally averaged part of the spectrum.

3.2.2. T < 78 K . Upon cooling the crystal and monitoring the EPR spectrum along the c axis one observes at 65 K the sudden appearance of a second line adjacent to the one observed at and below 78 K (see figure 4). Further decrease of the temperature leads reversibly to a change of the relative intensities of the two lines. We verified that this behaviour is not due to saturation. Figure 4 also shows the splitting of the lines below 65 K when H

I/

b. A similar behaviour is observed for any direction of the magnetic field. Figure 5 shows the corresponding angular dependence of the total spectrum in the b - c and a-b planes. The angular dependence in this latter plane is of

2800 H Bill et a1

overall sixfold symmetry (within the precision of our measurements). The extrema are, however, no longer parallel to the a axis.

Parametrisation of the angular dependence of each of the partners with the aid of a triclinic Zeeman spin-Hamiltonian necessitated the use of two sets of principal values of the g tensors. They are given in table 1.

50 G c--i

A 77K

B

D

I

I

E '

Figure 3. The observation of a motionally averaged spectrum above 77 K (as shown by the a r r o w ) . Spectra A and B are recorded at 77 K with H along b and a axes respectively i n the a-b plane. Spectra C and D correspond to the same axes at 130 K while spectrum E is recorded at an arbitrary orientation in the same plane.

The E P R lines of the host SO, ^- centre were also observed in parallel to the ones of the molybdate molecule ion. They show below 6 0 K increased width with what could be interpreted as a doubling of some of the lines. The features are, however, so uncertain within the signal-to-noise ratios available to us that this study was not pursued further.

A second phase transition is observed at 38 K in the EPR spectrum of the molybdate anion. The effects were difficult to follow because of heavy saturation problems. As they are seen more clearly in the EPR spectra of Ag2+ and in the Raman experiments on the pure crystal it was decided to give a detailed account of these results elsewhere.

Low-temperature phase transition in KLiSO, 2801

I

^{b O X I S c axis}

I

50 G

L-

Figure 4. T h e temperature variation of the EPR spectrum below 77 K as observed with H along b and c axes.

4. Discussion

4.1. The nature of the centre and electronic ground state

The EPR results indicate that the paramagnetic centre consists of a MoOj3- structure substituting for a host anion. They further show that this centre has a triclinic magnetic moment. The question naturally arises as to whether there is a lattice defect associated with the Mood3- molecule ion, if the Jahn-Teller effect is acting in some way and/or if the charge difference to the host anion site is at the origin of this low symmetry.

Several reasons speak against the presence of an associated lattice defect. First, the molybdate molecule ion has the same valence state as the host anion during growth of the crystal. Then x-irradiation takes place at low temperatures. Finally, the crystals which were irradiated at RT exhibit a centre which differs from the presently reported one. In particular, its g tensor is different and it presents resolved superhyperfine structure, probably with the Li neighbours.

The electronic structure of the free Mood3- molecule ion implies a highest occupied one electron molecular orbital of symmetry e , resulting in a 'E ground state. The nine internal vibrations transform as

{r}

⁼ a ;

+

^{e '}

+

2t; (under Td). The vibrational coor- dinates will be written

Q(r',

^{y ' )} to represent both, the free molecule ion vibrations and, when appropriate, the effective local vibrations. An intra-molecular Jahn-Teller effect of the type E

C3

e' is expected already for the free molecule ion within its ground state. Additionally, the pseudo-JT effect between the ground state and the nearby T2 electronic state coupling to the Q(t;) modes (Stoneham and Lannoo 1969, Agresti et a1 1984) is probably important. Application of the angular overlap model (Bacci 1978)

2802

H

Bill et a1

3450

3400 3450

3400

3350

.

-

I . t

0 20 40 60 80

e (degl

Figure 5 . T h e angular variation of the EPR spectrum below T , = 64 K in the a-b and b-r planes.

to the calculation of the corresponding ( first-order) coupling constants yields (we use the notation of Bill 1985):

The parameters e, and e, represent the perturbation energies of the dominantly metal orbital by one ligand and R = metal-ligand equilibrium separation. One finds in general e, > e, (Jorgensen 1971) and therefore the pseudo-JT effect to the radial

Q(t;) mode often cannot be neglected.

Experimental information is lacking regarding the exact position of the Moo4’- in the host lattice. It is likely, however, that it occupies the trigonal position of the host anion. Thus in locally complete surroundings the M o o d 3 - ‘sees’ a trigonal ligand field in addition to the JT effect. This former field admixes to the €,(E) ground state the excited E,(T2) state. Thus, the trigonal vibrations Q(e; (e’)) and Q(er ( t i ) ) are now involved in the ground state ^JT effect. Within a weak trigonal field ( V , < A,, =

Low-temperature phase transition in K L i S 0 4 2803 electronic cubic field splitting) the coupling constant to the former vibration is approxi- mately given by ( l a ) where as the one to the latter vibration is (approximately)

Introducing plausible values for the different parameters into this expression and into (la), one finds VtE r= VtET. This result shows that neglecting coupling to Q(e((t;)) is not a good approximation. We did not work out the formal model fully (see, e.g., Stoneham and Lannoo 1969, Agresti et a1 1984) but in the strong coupling limit a triclinic site symmetry is obtained in C4 or C 3 factor group symmetry. This symmetry is effectively realised at and below 78 K.

The motional effects which set in around 120 K correspond to a rapid movement between the minima of the JT potential sheet. Indeed, in spite of the higher mass the molybdate centre seems to show dynamical effects at lower temperature than the SO4 ^- moiety.

4.2. The relation to the crystal structure and dynamics

The question arises if the observed temperature dependence of the EPR spectra below 78 K is due to phase transitions or if the local dynamics is at its origin. The following essential facts have to be considered:

(i) the specific orientation of the g-ellipse in the a-b plane at 78 K and the fact that only one tensor is needed to explain the observed EPR spectra;

(ii) below 65 K two different g-tensors are needed and the one of the 78 K spectrum does not seem to be described as a weighted average between the two former ones;

(iii) the transition between the ‘high-temperature’ spectrum and the ‘low-tem- perature’ one is reversible;

(iv) the

(v) the symmetry of the S O 4 - centre.

The EPR spectrum at 78 K is a superposition of contributions of all the different impurity sites. The overall symmetry of its angular plot shows that the crystal has trigonal or hexagonal symmetry but that the magnetic moment of the individual impurity has triclinic symmetry. The fact that when H

11

c all the lines coalesce into one indicates that all the possible sites in the unit cell are mutually symmetry related by the space group elements and are thus group theoretically equivalent. They form one set of Wyckoff positions. Several of the possible trigonal or hexagonal groups fulfil those conditions within the constraint of the distribution of the host atoms in the unit cell. The groups include P31c, P63, P63mc (see the International Tables for Crystallography 1983).

The remarkable facts that the g-ellipse has its principal axes parallel to a and b , and that a-c is the monoclinic symmetry plane of the SO4- radical suggest the existence of symmetry planes among the elements of the space group. But none of the trigonal or hexagonal polar groups possesses a pure mirror plane of the type a-c and allows only equivalence between the sites (!I,$, z ) and

(3,

^Q, z

+

⁴⁾ (but not between

(i,:,

z )

and

( 3 ,

+, z ) ! ) if only small changes of the room temperature unit cell are assumed.

Thus, either the symmetry found in the EPR spectra is accidental (then, the groups P63, P3cl are suitable space groups) or it is in fact the b-c plane which is the relevant symmetry plane in the EPR spectra. This assumption implies that the SO4 - radical performs small librations still at 78 K with respect to the b-c plane. If this assumption

centre seems to involve an axial g-tensor above 120 K:

2804 H Bill et a1

applies then the space group P63mc would be appropriate. But then the phase transition would be of a reconstructive type. It is thus more likely that the molybdate molecule ion still undergoes librational motion and that the first hypothesis applies.

The modification of the EPR spectrum at 64 K is due to a phase transition. The fact that when H

11

^c two different lines are observed within a small temperature interval is pointing to collective effects and not to local motional dynamics. The fact that the anion set of Wyckoff positions splits into two inequivalent groups, each with the same number of sites (equal intensities of the EPR lines below approximately 45 K), indicates that (i) the two anion positions of the pure lattice are no longer equivalent, and (ii) the b-C-type glide planes have disappeared. In addition the g tensor ellipses of the moieties in the a-b plane have principal axes which are no longer parallel to a , b. If, effectively, the low-temperature symmetry group is translationsgleich the new space group corresponds to P3 or an orthorhombic space group but with a deformation so weak that it is not observable in the angular plot of the EPR spectrum. The evolution, as a function of the temperature, of the EPR lines along the c axis suggests the phase transition to be of first order.

The modification of the EPR spectrum at 64 K can be produced reversibly at 78 K when appropriate stress is applied to the sample. Preliminary experiments were performed on a crystal under uniaxial stress applied along the c axis. There is not a gradual change of the EPR line structure with increasing stress but a rather abrupt yet fully reversible ‘commutation’ between the single line and the doublet (always HI/ a and at T = 78 K). This behaviour is not observed in systems that display local motional effects.

Acknowledgment

This research was supported by the Swiss National Science Foundation.

References

Abello L, Chhor K and Pommier C 1985 J. Chem. Thermodyn. 17 1023 Agresti A , Ammeter J H and Bacci M 1984 J . Chem. Phys. 81 1861 Bacci M 1978 Chem. Phys. Lett. 58 537

Balagurov A M. Mroz B, Popa N C and Savenko B N 1986 Phys. Starus Solidi a 96 25 Bansal M L, Deb S K, Roy A P and Sahni V C 1980 Solid Stare Commun. 36 1047

Bill H 1985 in The Dynamical J T Effecr in Localized Systems ed. Yu E Perlin and M Wagner (Amsterdam:

Fonseca C H A. Ribeiro G M , Gazzinelli R and Chaves A S 1983 Solid Stare Commun. 46 221 Holuz F and Drozdowski M 1981 Ferroelectrics 36 379

International Tables for Crystallography 1983 vol A (Dordrecht: Reidel)

Jorgensen C K 1971 Modern Aspects of Ligand Field Theory (Amsterdam: North-Holland) Karppinen M, Lundgren J 0 and Liminga R 1983 Acta Crystallogr. C 39 34

Sastry M D , Dalvi A G I and Bansal M L 1987 J . Phys. C: Solid State Phys. 20 1185 Shibata T, Abe R and Fukui M 1986 J . Phys. Soc. Japan 55 2088

Sorge G and Hempel H 1986 Phys. Slatus Solidi a 97 431

Stoneham A M and Lannoo M 1969 J . Phys. Chem. Solids 30 1769 North-Holland) ch 13