Superexchange in molten MnF2

Solid State Communications, Vol. 28, pp. 13. ~P~rgamon Press Ltd. 1978. Printed in Great Britain.

0038—1098/78/1001—0001 $02.OO/O

SUPEREXCHANGE IN MOLTEN MnF2 Cesar Evora and Vincent Jaccarino Department of Physics, University of California, Santa Barbara, CA 93106 (Received 28 July 1978 by B. Mühlschlegel)

Utilizing the fact that fast—relaxing impurities broaden the EPR linewidth of strongly coupled paramagnets in proportion to the strength of the impurity-host exchange coupling, we have been able to show that the exchange interaction in the molten phase is as large as, if not larger than, it is in the solid state of Fe-doped MnF 2. This unique experiment distinguishes exchange from motional effects unlike resonance in the pure paramagnets.

The dipole—broadened, exchange—narrowed EPR linewidth ~H of certain paramagnetic solids was observed Cs) or decrease to either (MnF2, increase MnC12), (XMnF3, as theX melting = K, Rb or point Tm is crossed from below . However, the characteristic times for ion motion in the molten state of these salts are not known and may be of the same order as the exchange—induced decay times for the spin—spin correlation functions. Since motion could equally well narrow the EPR, an uncertainty exists as to whether or not the dynamic effects of exchange persist into the liquid state and, if they do, whether they are

larger or smaller than the solid—phase values. 2+behavior substitutional impurities, By studying the of the EPR L~Hof in both MnF2, with solid small and Fe molten states, we demonstrate that the impurity—host exchange ~IH is as large, if not larger, above than it is below TB. We utilize the fact that fast-relaxing Fe2~impurities are known to severely broaden the EPR at RbMnF2 in the solid state by an amount which is in direct proportion to the magnitude of ~ ~ In our experiments MnF2 rather th~n RbMnF3 was used for the host because the Fe2 impurities rapidly leach out of the latter sample at high

8O0~ II

/ -

MnF :Fe2+

~—Q_~

600

2

-~/

~_Th----

I

I

1

—

—a-—

-o--

—

-o.

—

—

I

400

200

1

MnF 2

0

I

0

2

4

2 °C)6

T(x10

I

I

8

10

Figure 1: variation with temperature of the X-band EPR peak-to-peak derivative linewidth of MnF2 :Fe and MnFz with Ho parallel and perpendicular (j) to c. The powder values of the linewidths just below the melting point are indicated by the syithol (P).

(Jj)

1

SUPEREXCHANGE IN 14DLTEN MnF

2

2

temperatures in our crucible configuration. Our results were obtained with a high— The system only major modification was the temperature1 EPR X-band bridge, described elsewhere. addition of an HF flow system to avoid oxidation of the fluoride samples at high temperatures. The results obtained for pure and doped (0.5% Fe2+) MnF2 are shown in Fig. 1. The intro duction of the Fe 2+ causes an increase in the EPR linewidth which is anisotropic with respect to the relative orientation of the external field and crystal axis. We interpret these results using the simple phen~menologicalmodel previously applied to the Fe2 and Co2+_doped RbMnF~ system.2’ 3 The strong exchange coupling between 2~ impurity and host oIn2 spins provides a temperature-dependent broadening the fast-relaxing Fe of the EPR. The u~eof a coupled equations-ofmotion approach2’ for the impurity and host magnetizations leads to the following expression for the EPR linewidth parallel cular (j) to the c axis:

II,j (Au)

(II)

and perpendi-

Vol. 28, No.1

Thus the only quantity which is expected to suffer an appreciable variation as Tm is crossed using Eq. of (1) its ariddependence the factors is because on discussed ~IH and above, we suggest the ratio of the impurity contributions to the linewidths obey the following relation at T Tm: -_________

[t~ — (Au)o] [Au

—

(&i)o

(X°(~T)] I m

M

]~ (x~(’Tfl11 =

(2)

(~IH)M

where: S and M subscripts indicate solid and molten state, respectively, and refers to H 0 ~ direction. Using the data in Fig. 1, we (6IH) obtain 71 From the measurement on (~IH)M = (°~s) M = 1.25, and hence we dethe MflF2 we know {w 5) termine the ratio of the solid—to—molten-state

=

(Au)0

+

f

X

—4-—

~

IH1+X

(1)

where (Au)0 is the dipole—dipole, exchange-narrowed linewidth of the pure host; f = fractional concentration of impurities, X~is the “bare” Curie susceptibility of the impurity; Xj~is the exchange—enhanced of thespin—lathost, 6SL , where susceptibility ‘5SL is the impurity ~IH xtice = relaxation rate. 2/3 Z ~~IH’~ )2 s(~ 1 + 1) = is the exchange induced rate of the impurity to the host spins and u5 is the effective exchange rate, produced by superexchange and/or motion, for the pure host spins, For FeZ+_doped MnF2 the large anisotropy observed in Au is attributed to the anisotropy in a result consistent with the known anisotropy of the FeF~susceptibility above TN. What are the effects on the various para— meters which determine the linewidth, as the crystal melts? At very high temperatures (kT >> 3) the exchange—enhanced host susceptibility, will differ little from the “bare” Mn2+ susceptibility, X~. Moreover, the “spin—only” MnZ+ paramagnetism is insensitive to changes in the crystal field and therefore the melting of the crystal will not affect and Xi.°i Xjj. to theThe strength symmetry of the impurityand susceptibility is crystalline sensitive field. However, in the molten state we might expect to be roughly equal to the owder average of X~just below Tm (i.e., (X~(~, Tm)]

[x~(~Tm)])~ In the high-temperature range the factor 1 ,~is very close to unity since X>>1.

impurity host exchange couplings to be (IH)M

=

1.3

crease or decrease ~IH~ It is also reasonable to argue that if IJHH)M Eq. (3) is valid we would also expect

>

a result which is entirely consistent with the decrease in the linewidth of pure MnF2 upon melting. Thus we are led to believe that exchange continues to contribute significantly to the spin dynamics above Tm in these strongly coupled paramagnets and that exchange is as large, if not larger, above Tm as it is below in MnF2.

Our EPR measurements indicate that MnF2 melts at 875 ±10°C. (The transition to the molten state is accompanied by large fluctuations in the Q of the EPR cavity.) In this regard it is to be noted that an ambiguity concerning Tm 5’ 6, 7ranging from 867 exists in the literature; values to 920°Chave been reported. Acknowledgement - Research supported in part by NSF DMR Grant 77-20185. One of the authors (C.E.) is indebted to the Conseiho Nacional de Desenvolvimento Cientifico e Tecnologico (Brazil) for a scholarship.

DORMANN, E., HONE, D., and JACCARINO, V., Physical Review B 14, 2715 (1976). Review B 6, 58 (1972). DOpi4~N, H., and JACCARINO, V., Physical Review B 16, 56 (1977). GULLEY, J. H., and JACCARINO, V., Physical

(3)

We emphasize that motion in the molten state can play no direct role in the broadening of the EPR linewidth except insofar as it can either in-

REFERENCES

1. 2. 3.

___

Vol. 28, No.1 4.

5. 6. 7.

SUPEREXCHANGE IN MOLTEN NnF

2

Our MnF2 AH measurements differ from those reported in Ref. 1 in the vicinity of Tm. We do not observe the change of sign in the anisotropy factor (1 - ABi/ABII) just below Tm. Without the HF flow system we could not obtain reproducible linewidth measurements as Tm was crossed in either direction. We attribute this behavior to the sensitivity of NnF2 to oxidation at elevated temperatures and would probably account for the differen8e between our results and those of Ref. 1. GRIFFEL, M., and STOUT, J. W., Journal of the ~onerican Chemical Society 72, 4351 (1950). BELYP~EV, I. N., and REVINA, 0. Y., Russian Journal of Inorganic Chemistry 11, 1041 (1966). WEINBERG, E., and SRINIVASAN, N. K., Journal of Crystal Growth 26, 210 (1974).