Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Home Search Collections Journals About Contact us My IOPscience

The sign of the internal field at 119Sn nuclei in Co2MnSn

This content has been downloaded from IOPscience. Please scroll down to see the full text.

1969 J. Phys. C: Solid State Phys. 2 2037

(http://iopscience.iop.org/0022-3719/2/11/318)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 142.104.240.194
This content was downloaded on 03/10/2015 at 19:14

Please note that terms and conditions apply.


J . PHYS. c (SOLID ST. P H Y S . ) 1, 9 6 9 , SER. 2, VOL. 2. PRINTED I N GREAT BRITAIN

The sign of the internal field at 'l9Sn nuclei in Co,MnSn


J. M. WILLIAMS
Department of Physics, University of Sheffield,? and
Physik-Department der Technischen Hochschule Miinchen, Germany
M S . received 24th May 1969, in revised form 6th June 1969

Abstract. The sign of the nuclear hyperfine field at a Sn nucleus in the ordered
ferromagnetic Heusler-type alloy CozMnSn has been determined using polarized
Mossbauer y-rays. Unlike all other fields measured at various nuclei in Heusler
alloys, the field at Sn in CozMnSn is shown to be positive. T h e measured value of
+ l o 7 +1 kOe at 4.2 'K is compared with the theoretically predicted value obtained
by assuming the field to be due to the oscillating conduction electron spin polarization
induced by localized moments on the M n atoms. This model, which yields the correct
negative sign for all other measured fields, is also shown to yield the correct positive
sign for the field at Sn in CozMnSn.

1. Introduction
In order to understand fully the origin of hyperfine fields at nuclei in ferromagnetic
materials it is necessary to know how both the magnitude and sign of such fields vary in
Ielated materials. Thus our original Mossbauer measurements of such fields at the l19Sn
nuclei in the ordered ferromagnetic Heusler-type alloy Co,MnSn (Williams 1968) have been
extended to lower temperatures and to a determination of the sign.
I n view of the possible ambiguity associated with a direct determination of the sign of
the hyperfine field He obtained by placing the absorber in a transverse magnetic field
(especially when the actual applied field, less the demagnetizing field, is small compared
with He), a second unambiguous experiment utilizing circularly polarized Mossbauer
y-rays was performed. T h e direct determination of the sign ( 4 3.1) was performed at
Sheffield, while the polarization experiment ( 4 3.2) was performed at Munich.
T h e sign and magnitude of He are found to compare favourably with the theoretical
predictions obtained by assuming a model similar to that discussed by Caroli and Blandin
(1966), and which is based on Friedel theory (Friedel 1954, 1958). On this model the
hyperfine fields arise from the oscillating spin polarization induced in the conduction band
by the magnetic moments localized on the M n atoms.

2. Sample analysis and experimental details


The preparation of the Co,MnSn absorber has been described previously (Williams
1968). An x-ray analysis indicated that the samples exhibit the typical Heusler (Cu,MnAl-
type) crystal structure. That they are ordered is illustrated by the fact that the observed
reflections fall into three types.
Type A: A set of body-centred cubic reflections with the indices doubled. These
correspond to the small body-centred cubic units (eight of which form the complete unit cell)
which have half the full cell spacing a,.
Type B: Extra reflections of the face-centred cubic type with h, k, 1 odd.
Type C: Extra reflections of the face-centred cubic type with h, k, 1 even.
As indicated by Bradley and Rogers (1934), type B reflections should be stronger than those
of type C in our alloy as Sn is the strongest scatterer of the three constituent atoms. This
was in fact observed in the photograph. T h e lattice constant was calculated to be
u o = 5.99 A, in excellent agreement with the accepted value of 5.989 h (Castelliz 1955).
For the direct determination of the sign ( 4 3.1) the apparatus was as described previously,
but with the addition of a magnet capable of producing transverse fields up to 10 kOe at
the absorber. I n the polarization experiment ( 4 3.2) the source (this time lIgMSnin Pd) and
absorber were mounted inside a superconducting magnet capable of producing longi-
tudinal fields up to 60 kOe. T h e drive and data-recording were standard equipment. T o
t Permanent address.
2037
2038 J. M . Williams
detect the 23.8 kev llgMSny-rays a Si detector together with a cooled Tennelec F.E.T.
preamplifier (model T C 130) were used. Such a counter has the advantage of being un-
affected by the presence of strong magnetic fields.
I n both parts of the experiment Pd foils were utilized as critical absorbers to reduce the
relative intensity of the 25.8 kev K x-rays of Sn, and the Mossbauer absorbers contained
25 mg cm-2 natural Co,MnSn.

100

98
-10 -8 -6 -4 -2 2 4 6 8 1 0
Relative v e l o c i t y ("is)

(4

I I , , , . , ~, , , , ,
-10 -8 -6 -4 -2 0 2 4 6 8 10
Relative velocity (mm/s)

(4
Figure 1. Mossbauer absorption spectrum of CozMnSn a t room temperature (a) in
zero applied field and (b) in a transverse applied field Ho = 5 kOe.
The sign of the internal3eld at Il9Sn nuclei in CozMnSn 2039

3. Experimental procedure and results


3.1. The direct measurement
Mijssbauer absorption spectra were obtained with the absorber in transverse magnetic
fields ranging from 0 to 10 kOe. From the change in the measured internal field the sign of
He may, in principle, be obtained. All spectra were measured with the source and absorber
at rooin temperature, and were fitted as described previously (Williams 1968).
3.2. The polarization measurement
I n this case Mossbauer absorption spectra were obtained (i) with the source and
Co,MnSn absorber in zero applied field, and (ii) with both the source and absorber in
a large longitudinal magnetic field H , = 57 kOe. Both (i) and (ii) were measured with the
source and absorber at 4.2 OK.
,
102;

L
2 8 L

oL-- 2 4 6 8 IO
Ho (kOe)
Figure 2. T h e measured magnetic field H,,, at the 119Snnuclei in Co,SnMn plotted
as a function of the applied field Ho.

92
-8 -6 -4 -2 0 2 4 6 8
Relative velocity (mm/s)
Figure 3. Mossbauer absorption spectrum obtained with the Sn(Pd) source at 4.2 "K
in a large longitudinal applied field Ho = 57 kOe and a single-line SnMg, absorber
at room temperature in zero applied field.
2040 J. M. Williams
Additional Mossbauer absorption spectra using a single-line absorber were then
obtained as follows: (iii) with the source in zero field at 4.2 OK and an unsplit SnMg,
absorber in zero field at room temperature, and (iv) with the same arrangement as in (iii) but
with the source alone in a large longitudinal magnetic field H , = 57 kOe similar to that
used in (ii).
3.3. Results
The results of the direct measurement are summarized in figures 1 and 2.
T h e polarization measurement: spectrum (iii), showed an unsplit absorption peak,
indicating that in the absence of an applied field the Sn(Pd) source emits a single line.

-14 -12 -10 -8 -6 - 4 -2 0 2 4 6 8 IO 12 14


Relative velocity ( m m / s )

Figure 4. Mossbauer absorption spectrum obtained with both the Sn(Pd) source and
CozMnSn absorber at 4.2 OK in zero applied field. The full curve represents a computer
fit.
I

20 15 IO 5 0 5 IO 15 20
Relative velocity ( m m / s )

Figure 5 . Mossbauer absorption spectrum obtained with both the Sn(Pd) source and
Co&InSn absorber at 4.2 OK and in a large longitudinal applied field H , = 57 kOe.
T h e full curve represents the best symmetrical eight-line fit.
The sign of the internaljCield at lI9Sn nuclei in Co,MnSn 2041

Spectrum (iv) is shown in figure 3 and shows the four-line pattern expected when the source
is split by the applied field. -4 computer fit of (iv) shows that the splitting corresponds
almost exactly to a magnetic field at the source nuclei equal to the applied field. We may
therefore conclude that the field at the source in this and in the polarization spectrum (ii)
is equal in magnitude and sign to the applied field produced by the superconducting sole-
noid. Spectra (i) and (ii) are shown in figures 4 and 5 respectively. A computer fit of
spectrum (i) (figure 4), taken with the accepted value of pg = - 1,046 nuclear magnetons
yields a value of 107 5 1 k 8 e for the internal field He at l19Sn nuclei in Co,MnSn at 4.2 OK.

4. Interpretation
4.1. The direct measurement
T h e increase of the measured hyperfine field with applied field, shown in figure 2,
indicates that the internal field has a positive sign. However, in view of a possible anomalous
increase in the splitting which could be caused by the splitting of the source in a small stray
magnetic field (see Maletta et al. 1969), it is necessary to confirm the sign by an analysis of
the polarization experiment.
4.2. The polai4zation measusement
T h e angular dependence of the y-radiation for allowed transitions in the hyperfine
+
Zeeman spectrum for an I = 2 to I = transition is given in table 1. If the source or
absorber is magnetically ordered three special arrangements may be distinguished: the
relative intensities are 3 : 0 : 1 : 1 : 0 : 3 where the propagation of the y-rays is parallel
or antiparallel to the internal magnetic field, 3 : 4 : 1 : 1 : 4 : 3 where the y-rays travel
perpendicular to the internal field, and 3 : 2 : 1 : 1 : 2 : 3 in the case of random field
directions. In the case of a magnetically split source and absorber, however, the resulting
Mossbauer absorption spectrum will exhibit peaks with relative intensities which depend
on the angular dependence of the y-rays and the relative orientation of the magnetic fields
(Fraunfelder et al. 1962).

Table 1. Angular dependence of the allowed transitions in the nuclear Zeeman


pattern for dipole radiation (I, = i, Zg = i)

Transition hm Angular dependence


* $ -+ 16 T1 ~(l+COS~O)
+&-.
- 19 0 sin%
T+ + 19 11 *(l +cos*@

Owing to the extinction of Am = 0 transitions, the simplest spectra arise when the direc-
tion of observation coincides with that of the magnetic field. When both source and
absorber lie in such a longitudinal magnetic field the y-rays emitted by the source along
the field axis are circularly polarized, the usual convention being to call a y-ray right
circularly polarized if its spin lies in the direction of motion. Thus, for example, in the
+
source transition m = - 4 to m = - the y-ray must have a x component of angular
momentum - 1. Therefore if emitted along the + x direction it is left circularly polarized,
and such a y-ray may only be absorbed in an absorber transition which requires angular
momentum - 1. Thus only eight absorption peaks are expected in the Mossbauer spectrum,
but their positions and relative intensities will differ greatly depending on whether the
magnetic field at the source is parallel or antiparallel to that in the absorber. I t is this
difference which enables us to determine the actual sign of the field in the CoJTnSn
absorber. X detailed analysis of the possible absorption lines, their positions and relative
intensities for the case of 57Fe,may be found in the papers by Blum and Grodzins (1964)
and Gonser (1967).
As we know that the sign of the field at the source nuclei is positive, i.e. in the same
direction as the applied field (see 4 3.2), a comparison of our polarization spectrum, figure 5,
with the expected absorption spectra for positive and negative absorber fields should
2042 J. M . Williams
yield the correct sign. T h e relative intensities for such polarization spectra as tabulated by
Gonser (1967) apply only to the case of a thin source and thin absorber. These conditions
are not applicable to our measurement, but as we have already measured the source and
absorber spectra separately (figures 3 and 4 respectively) we can, from their computer fits,
obtain estimates of the actual source and absorber intensities in the magnetic field by simply
assuming that the absorber relative intensities will be the same as for figure 4, but with the
Am = 0 absorption lines absent.
In this way we can predict the actual relative intensities to be expected in the polariza-
tion spectrum (figure 5). From the data obtained from the above-mentioned fits we may
also estimate the line positions to be expected for the two possible field directions at the
absorber nuclei. T h e field at the source is known to be Hs = + 5 5 kOe, while that at the
absorber nuclei is given by
Hn,, = +He+(Ho-HD) (1)
depending on whether He has a negative or positive sign. (H,.is the applied field and H D
the demagnetizing field.) T h e only factor not determined from the fits is the demagnetizing
field in the absorber. This, however, need not be known very accurately as it will be small
compared with H , and may be estimated from the results of 4 3.1. I n figure 2 it is seen
that until the applied field is about 5 kOe the measured field at the nucleus shows no change ;
we may therefore assume that the demagnetizing field is approximately 5 kOe. Thus for a
positive internal field He, the effective field at the Sn nuclei, when in an external field of

Relative v e l o c i t y (mm/s)

(a>

I
-20 -15 -10 -5 0 5 IO 15 20
Relative v e l o c i t y (mm/s)

(b)
Figure 6. T h e predicted Mossbauer polarization spectrum constructed assuming
(a) a negative and ( b ) a positive internal field at the Il9Sn nuclei in CozMnSn.
The sign of the internalJield at l19Sn nuclei in Co,MnSn 2043

57 kOe, would be
H,,, = 107+57-5 = 159 kOe
while for a negative internal field
H,,, = -107+57-5 = -55 kOe.
T h e predicted polarization spectra for positive and negative absorber fields constructed in
this way are shown in figure 6, and from a visual comparison alone it is clear the the correct
spectrum is that assuming a positive internal magnetic field at the Sn nuclei in Co,MnSn.
T h e estimated and experimentally determined splittings for the four pairs of symmetrical
absorption lines are given in table 2.

Table 2. Estimated and experimentally measured splittings of the four pairs


of symmetrical absorption lines in the polarization experiment (figure 5)

Splittings relative to centre of


Method spectrum (mm s - l)
estimated assuming H e negatice F1.9 k2.0 k3.6 k7.5
estimated assuming H e positice k3.3 F7.0 rt8.9 k12.5
experimentally measured k3.1 i6.9 k8.9 k12.6

5. Theory
Samoilov et al. (1959, 1960) first showed that large hyperfine fields exist at diamagnetic
impurities in ferromagnetic lattices, and such fields have since received much attention.
While similar fields at magnetic atoms in ferromagnetic lattices have been fairly well
described by polarizations of the core and conduction electrons (Watson and Freeman 1961),
there exists no general treatment for non-magnetic atoms. As a general rule such fields are
negative and proportional to the magnetic moment of the host (Shirley and Westenburger
1965). ,IIore recent measurements, such as those by Samoilov et al. (1965) and Frankel
et al. (1965)’ show, however, that this general rule regarding the sign breaks down for solute
atoms with atomic number 50 dissolved in Fe and Ni. These results also show that Sn
(atomic number 50) is near the cross-over point for the sign of the induced field. However,
until now the only positive field found at Sn nuclei is that in Ni (Boyle et al. 1960), and
this is very small ( + 18.5 kOe). T h e present observation of a relatively large positive field
( + 107 kOe) is therefore unusual in the case of Sn.
All previous measurements of hyperfine fields at various nuclei in Heusler alloys have
yielded a negative sign, and such fields have been best described using the model of Caroli
and Blandin (1966). On this model the fields are attributed to oscillations in the conduction
electron spin polarization induced by the magnetic moments located on M n atoms. T h e
model is based on Friedel’s partial wave method (Friedel 1958, Daniel and Friedel 1965)
and has been summarized by Daniel (1967). The results of such measurements and theoreti-
cal calculations are summarized in table 3.
I n view of the relative success obtained in predicting such hyperfine fields, it seems
worth while to adapt such a model to the case of Co,MnSn which, although not being a
true Heusler alloy (CO being normally ferromagnetic), is in many respects very similar to
alloys of this class. It has the typical ordered crystal structure and its magnetic moment of
4.2 pB is reportedly carried mainly by the M n atoms (Shinohara and Watanabe 1966).
Caroli and Blandin assume that only one of the seven 3d54f2M n electrons goes into the
conduction band, the remaining six being localized in a virtual bound state in such a way as
to produce the right screening charge and magnetic moment (4.0 pBon the M n in Co,MnSn).
Assuming that these M n atoms are sufficiently far apart, so that they may be treated
as independently producing their own spin polarization in the conduction band, the hyper-
fine field at a certain Sn nucleus a distance d away is given by
2044 J . M . Williams
Table 3. Measured and predicted hyperfine fields in Heusler alloys
Theoretical magnetic Experimental magnetic
Alloy Nucleus field t field
Value Reference Value Reference Method
We) We)
Cu2MnAl 63Cu/85Cu -199 Caroli and -212 magnitude and n.m.r. decrease
Blandin sign: Ogawa of resonance
(1966) and Smit, (1969) frequency with
increasing
applied
magnetic field
CuzMnAl 27Al - 168 Caroli and k 68 magnitude Ogawa n.m.r.
Blandin and Smit (1969) :
(1966) sign: not
determined
Cu2MnIn 63Cu/65Cu - 21 3 Caroli and - 197 magnitude n.m.r. decrease
Blandin Sugibuchi and of resonance
(1966) Endo (1964) frequency with
sign : Titman, increasing
private com- applied
munication magnetic field
Cu2MnSn l19Sn i:200 magnitude Mossbauer
Chekin et al. effect
(1967)
sign: not
determined
Pd2MnSn llQSn -66 Kanekar - 48 magnitude and Mossbauer
et al. (1968) sign : Kanekar effect: actual
et al. (1968) method not
given
t All the predicted fields are calculated using the model of Caroli and Blandin (1966).

where p B is the Bohr magneton, (/$(0)/2)Fthe average density of conduction electrons at


the Fermi level on the Sn atom, Clo the volume per atom and n ( d ) the spin density polariza-
tion due to the M n moment. For 2 = 6 and p = 4.0 p B this polarization is given bp
5
n(d)= - (3)
where k, is the Fermi momentum for the free conduction electrons given by

k, = 3n2- (4)
n being the number of conduction electrons per atom. If we assume that the conduction
band is made up of one electron per Mn atom, four electrons per Sn atom and two electrons
per CO atom, we obtain a value of K F = 0.905 A.u.-'.
T o estimate (l$(0)12)F is, however, more difficult,but if we assume it to be the same as in
pure Sn, equation (2) may be written as

Here E = PF/PA,where PF = (/$(0)2))Fand PA = (I$A(0)2/)are the probability densities


in the metal and free atom respectively, a is the hyperfine coupling constant for an s electron
in the free atom and g, the nuclear ground state gyromagnetic ratio for lI9Sn.
T h e value of /g,j (= 1-56x erg G - ~ ) is known accurately (Muir et al. 1966).
f s n ( = 0.41) may be determined from electronic specific heat measurements and is given by
The sign of the internal$eld at 1 1 9nuclei
S ~ ~in Co,AlnSn 2045

Knight (1956). T h e value of la,,J (=0.904 x erg) is more uncertain and in our case
is that calculated by Knight (1956) using the Goudsmit formula (Goudsmit 1933). Using
these values equation ( 5 ) reduces to
He = 1.072 x lo8 n(d)
so that the total field at a Sn nucleus will be given by summing over all the surrounding
M n shells:
2
He = 1.072 x lo8 n(d). (6)
The various contributions are listed in table 4. Owing to the d - 3 dependence, only the first
three shells are important, and the total estimated field is approximately + 32 kOe.

Table 4. Estimated contributions to the hyperfine field at Sn nuclei due to the


surrounding Mn atoms
No. of Mn
atoms in the d n(d) per Mn He per Mn Total H e per
Mn shell shell (AX.) atom atom (kOe) shell (kOe)
1st n.n.t 6 ao/2 = 5.65 -5.75 x ~ O - ~ -6.16 -37.0
2nd n.n. 8 1/3a0/2 = 9.79 $6.95 x +7.45 ~59.6
3rd n.n. 6 3ao/2 = 16.95 $1.50 x ~ O - ~ $0.6 +9*6
4th n.n. 8 4 3 a o = 19.6 -6.78 x ~ O - ~ - 0.07 -0.5
Total He (Theory) N +32 kOe.
f n.n., nearest neighbour.

6. Conclusion
Thus, while only approximately predicting the magnitude of the internal field, this
model does yield the correct sign. As seen from equation (5) the magnitude depends
critically on the values used for the parameters E and a. T h e estimates used here neces-
sarily carry a large error (Knight (1956) attributes a 50% possible error to a alone). How-
ever, the sign on this model depends only on Cn(d)which may be estimated more accurately
given the lattice constant and the number of conduction electrons per atom. Errors are also
to be expected, however, as a result of inherent deficiencies in the model, for example in the
use of the asymptotic form of the spin density oscillations up to the nearest neighbours,
and as a result of neglecting the small moment on the CO atoms in our case.
We therefore conclude that, whereas the agreement between theoretical and experi-
mental magnitudes listed in table 3 is sometimes fortuitous, the prediction of the sign is
correct for all cases in which it is known. T o complete the survey the sign of the field at
Sn in Cu,MnSn will shortly be determined.

Acknowledgments
The author wishes to thank Professor R. L. Mossbauer, Physik-Department, Technische
Hochschule Munchen, for the opportunity to spend a year at his Institute, for his hospitality
during this period and for his interest in this work. Financial support from the German
Federal Ministry of Science for this period is also gratefully acknowledged.
Thanks are also due to Herrn G. Kaindl of the above-mentioned Institute for his
assistance during the use of the superconducting magnet cryostat and for valuable
discussions.

References
BLUM,N,, and GRODZINS, L., 1964, Phys. Rev., 136, A133-7.
BOYLE,A. J. F., BUNBURY, D . ST.P., and EDWARDS, C., 1960, Phys. Rea. Lett., 5 , 553-6.
BRADLEY,A. J., and ROGERS, J. W., 1934, Proc. R. Soc. A., 144, 340-59.
R., and BLANDIN,
CAROLI, A., 1966, J . Phys. Chem. Solids, 27, 503-8.
CASTELLIZ,L., 1955, 2. Metall., 46, 198.
CHEKIN,V. V., DANILENKO, L. E., and KAPILENKO, A. I., 1967, Soa. Phys.-JETP, 24,472-4.
2046 J. M . Williams
DANIEL,E., 1967, Hyperfine Interactions, Eds A. J. Freeman and R. B. Frankel (New York, London:
Academic Press), pp. 712-23.
DANIEL,E., and FRIEDEL, J., 1965, Low TemperaturePhysics,LT9, part B (New York: Plenum Press),
p. 933.
FRANKEL, R. B., HUNTZICKER, J., MATTHIAS, E., ROSENBLUM, S. S., SHIRLEY,D. A., and STONE, N. J.,
1965, Phys. Lett., 15, 163-5.
FRAUNFELDER, ~ , R. D., COCHRAN,
H., h T ~D. ~E., ~TAYLOR, D. R. F., and VISSCHER, W. M., 1962,
.Phys. Rev., 126, 1065-75.
FRIEDEL,J., 1954, Adn. Phys., 3,446-507.
- 1958, Nuovo Cim. (Suppl.),7, 287-311.
GONSER, U., 1967, Hyperfine Interactions, Eds A. J. Freeman and R. B. Frankel (New York,
London: Academic Press), pp. 696-711.
GOUDSMIT, S., 1933, Phys. Rev., 43, 636-9.
KANEKAR, C. R., MALLIKARJUNA RAO,K. R. P., and UDAYA SHANKAR RAO,V., 1968, Phys. Lett., 28A,
220-1.
KNIGHT, W. D., 1956, Solid State Physics, Vol. 2. Eds F. Seitz and D. Turnbull (New York:
Academic Press), pp. 93-136.
MALETTA, H., FRANKEL, R. B., HENNING, W., and MOSSBAUER, R. L., 1969, Phys. Lett., 28A, 557-8.
MUIR,JR,A. H., ANDO,K. J., and COOGAN, H. M., 1966, Mossbauer Effect Data Index (1958-1965)
(New York: Interscience), p. 126.
OGAWA, S., and SMIT,J., 1969, J. Phys. Chem. Solids, 30, 657-63.
SAMOILOV, B. N., SKLYAREVSKII, V. V., and STEPANOV, E. P., 1959, Zh. Eksper. Teor. Fix., 36,1944-6.
- 1960, Zk.Bksper. Teor. Fix., 38, 359-71.
SAMOILOV, B. N., AGUREEV, V. N., GOROBCHENKO, V. D., SKLYAREVSKII, V. V., and CHILASHVILI,
0 . A., 1965, Low Temperature Physics, LT9, part B (New York: Plenum Press), pp. 925-8.
SHINOHARA, T., and WATAKABE, H., 1966, J . Phys. Soc. Japan, 21, 1658-62.
SHIRLEY,D. A., and WESTENBARGER, G. A., 1965, Phys. Rev., 138, A170-6.
SUGIBUCHI, K., and ENDO,K., 1964, J. Phys. Chem. Solids, 25, 1217-20.
WATSON, R. E., and FREEMAN, A. J., 1961, Phys. Rev., 123, 2027-47.
WILLIAMS, J. M., 1968, J. Phys. C (Proc. Phys. Soc.), [ 2 ] 1, 473-7.

You might also like