Magnetism and Magnetic Anisotropy in

Exchange BIAS Systems

A. J. Freeman, K. Nakamura, M. Kim, and W. T. Geng

Department of Physics and Astronomy
Northwestern University
Evanston, IL60208
USA

Abstract

Magnetism in man made ultra-thin materials has become an area of intense activity
and excitement, mainly driven by applications in magnetic recording. For the pur-
pose of high-density magnetic recording, the understanding and control of the mag-
netic properties, such as complex magnetic ordering, enhanced magnetic moments,
magneto-crystalline anisotropy and magnetostriction are desired. In this paper, we
present results of magnetism and magnetic anisotropy in ferromagnetic NiFe thin film,
antiferromagnetic NiMn and their interface determined from first-principles FLAPW
(full-potential linearized augmented plane wave) calculations, and demonstrate how
their properties behave sensitively on the environment (such as in bulk, at surfaces or
interfaces). These results may be important for understanding exchange bias materi-
als.

I. Introduction
Magnetism research has been undergoing a renaissance over the last decade following
the discovery of a variety of new scientific phenomena associated with man-made
transition metal thin films. Among them are the theoretical prediction of enhanced
magnetic moments in ultra-thin films and at surfaces, the discovery of perpendicular
magnetic anisotropy in ultra-thin films and layered structures, and the discovery of
giant magnetoresistance (GMR) and the accompanying oscillatory exchange coupling
in multilayers made by alternating magnetic and nonmagnetic metals. Some of these
discoveries are already having a major impact on the magnetic recording industry.
First principles electronic structure studies based on local spin density functional
theory, which performs extremely complex simulations of ever increasingly realistic
Physics of Low Dimensional Systems
Edited by J. L. Morán-López, Kluwer Academic/Plenum Publishers, New York 2001 383
384 A. J. Freeman et al.

systems, plays a very important role in successfully explaining magnetism in thin

films and has led to the facing of even more challenging problems. It is known that the
calculations, especially those carried out at Northwestern using the full-potential lin-
earized augmented plane wave (FLAPW) method,1 played a significant role during the
course of developments of the field. The calculations predicted the large enhancement
of the magnetic moment for 3d-transition metal (TM) surfaces or overlayers deposited
on inert substrates, and the possible magnetization in some normally non-magnetic
materials, for which some results have already been verified experimentally. Complex
magnetic structures, especially some antiferromagnetic (AFM) configurations, can
now be predicted by comparing total energies with their equilibrium atomic geome-
tries (including multilayer relaxations and reconstructions at surfaces and interfaces)
optimized very efficiently using the atomic force approach. Significant progress has
been made very recently for the treatment of the weak spin-orbit coupling (SOC) and
now we are able to obtain very reliable results of i) the magneto-crystalline anisotropy
(MCA) energies and magnetostrictive coefficients for transition metal thin films and
ii) the magneto-optical Kerr effect (MOKE) and soft X-ray magnetic circular dichro-
ism (MCD).
The aim of this paper—dedicated to Hans-Christoph Siegmann—is to present
a case study on magnetism and magnetic anisotropy of important materials used in
magnetic recording industries, and to demonstrate how their properties depend on the
environment in bulk, at surface or interface. We first report results for ferromagnetic
(FM) NiFe thin films. The effects of capping a Ta overlayer on NiFe are also discussed,
which is commonly used in magnetic recording applications but the effects are not well
understood. We next focus on the magnetic anisotropy of antiferromagnetic (AFM)
NiMn and the FM/AFM NiFe/NiMn interface, which continues to attract great at-
tention in advanced technological applications—especially the exchange bias which is
associated with the magnetic anisotropy created at an interface between ferromagnetic
and antiferromagnetic materials.2,3 NiMn is believed to be a promising candidate as
the exchange bias AFM material,4,5 because it possesses a high crystalline anisotropy,
high Neel temperature and high corrosion resistance.
The calculations were performed by the FLAPW approach based on the local
density approximation (LDA), in which the core states are treated fully relativisti-
cally and the valence states are treated semi-relativistically. The MCA energy was
calculated by treating the SOC in a second variation way using the state tracking and
torque methods.6–8

II. Ferromagnetic NiFe Thin Film

In this study, clean surfaces of a 5-layer slab with x = 0.5 and 0.75 and
a Ta overlayer of half coverage on each side of the slab were adopted. For the clean
surfaces, two different possible geometries for each composition were considered: with
either Fe or Ni layers on the surface for x = 0.5 in the structure and
and with both Fe-Ni mixed layers or a Ni layer on the surface for x = 0.75 in
the structure Experimental bulk lattice constants
were employed for the in-plane lattice spacing for both x = 0.5 (6.77 a.u.) and
0.75 (6.72 a.u.). The structural optimization was accomplished for each system by
performing atomic force calculations9 to obtain the fully relaxed vertical geometry.
For the Ta overlayer case, total energies of several different adsorption sites of the
surface were compared; this revealed that the hollow site is energetically
Magnetism and Magnetic Anisotropy in Exchange BIAS Systems 385

the most favorable site for the Ta overlayer rather than the Fe or Ni atomic sites by
several eV. Note that there are two different hollow sites for the with
either an Fe or a Ni subsurface atom under the Ta adatom.
The optimized positions and the magnetic moments of atoms in the surface (S), the
subsurface (S-l) and the center (C) layers are given for the 5-layer clean surface films
in Table I. We found that the surface and subsurface layers undergo a large downwards
contraction for the clean surfaces (cf. Table I), which results in the reduction of the
interlayer distance (between the surface (S) and subsurface (S-l) layers) and
(between the subsurface (S-l) and center (C) layers), by 8 to 11% and 3 to 7%,
respectively, compared to the bulk alloy. As a result of full relaxation, there appears a
slight buckling in the mixed layers of Fe and Ni. The Ni atoms suffer a more downwards
relaxation compared to Fe mainly due to the fact that Ni has a smaller atomic volume
than Fe. The difference in the vertical positions of Fe and Ni is larger for the mixed
layer on the surface (0.09 a.u. for compared to the mixed layer on the
subsurface (0.03 a.u. for
Due to surface effects and relaxation, the magnetic moments of these thin films
appear to be different from those of the bulk alloys (which were 2.69 (Fe) and 0.68 (Ni)
for in the structure and 2.94 (Fe) and 0.67 (Ni) for
in the structure according to our calculations). It was found that the magnetic
moments of the Fe atoms are very sensitive to the environment while those of the Ni
386 A. J. Freeman et al.

atoms are not much influenced. For the film with Fe on the surface, the
magnetic moment of the surface Fe is enhanced by and that of the center Fe
is reduced by compared to the bulk value with the same lattice spacing while
the subsurface Ni magnetic moment is not far from the bulk value (cf. the values in
parentheses in Table I versus bulk values). In the case of the film with the Ni atom
on the surface, the magnetic moment of the surface Ni atom does not exhibit a large
change compared to the bulk value.
Upon full structural optimization, the magnetic moment of Fe tends to be reduced
while that of Ni increases slightly for x = 0.5 and decreases for x = 0.75. This may be
caused by increased interatomic interactions due to the relaxation; the smaller Fe-Fe
or Ni-Ni interatomic distance is supposed to reduce the magnetic moments while the
increase of interatomic interactions between Ni and Fe is known to enhance the Ni
magnetic moments and to reduce the Fe magnetic moments.10 Therefore, the relax-
ation acts to reduce the Fe magnetic moments while its influence on the Ni magnetic
moments depends on the system; for large Ni composition, the Ni-Ni interatomic in-
teraction dominates over that of Fe-Ni and thus the relaxation reduces the magnetic
moment of Ni.
It is clearly seen from Table I that the increase of Ni composition enhances the
magnetic moment of Fe as already found from the bulk alloy results. The dependence
of magnetic properties on the atomic composition is more clearly demonstrated from
magnetocrystalline anisotropy (MCA) calculations. As revealed in previous investiga-
tions, MCA is extremely sensitive to the environment. The calculated MCA, defined
by the difference in total energies for the magnetic moments oriented perpendicular
and in-plane, are given in Table II. These reveal that the MCA changes sensitively de-
pending on the nature of the surface atoms and the atomic composition. For
we found that the bulk alloy has the perpendicular (001) direction as the easy axis
with MCA = 0.06 meV. For the 5-layer films with Fe on the surface, the perpendicular
direction is still the easy axis with an enhanced MCA, while it switches to in-plane
with Ni on the surface. The case of which in the bulk case has negligible
MCA due to its cubic symmetry, now exhibits a large magnetic anisotropy in keeping
with the lowered symmetry, or surface effect, and shows an in-plane preference for the
magnetization direction. Consistent with the result of slabs, the
thin film with the Fe-Ni mixed surface layer on the surface has a very small MCA,
while the one with Ni on the surface shows a large MCA. Also noting that the actual
atomic composition in one unit cell is different for each system (as shown in Table II),
we can see that the MCA changes with atomic composition; with more Ni included,
the in-plane preference becomes stronger (cf. and
The optimization results and the magnetic moments of a Ta overlayer on the
substrate show several interesting things. The interlayer distance between
Magnetism and Magnetic Anisotropy in Exchange BIAS Systems 387

Ta and the surface layer of the substrate (S) shows large relaxation compared to the
average interlayer distances of Ta and NiFe (001) bulk (3.26 and 3.24 for
and respectively), ranging from 4 to 8%. Due to the Ta overlayer, the
relaxation of the substrate (between S and S-l) is largely reduced compared to
those of the clean surfaces, reaching only up to 6% of the bulk values at most. The
magnetism of the Ta overlayer changes dramatically depending on the surface layer;
for the systems with an Fe layer and a Fe-Ni mixed layer on the surface, Ta couples
antiferromagnetically with the substrate while it orders ferromagnetically for the Ni
surfaces. Surprisingly, the induced magnetic moment of Ta is large, ranging from 0.34
to 0.56 in magnitude. The effect of the Ta overlayer on the magnetic moment
of the substrate is also quite interesting: it acts to reduce significantly the magnetic
moment of the nearest (i.e. surface) atom; the surface Fe and Ni magnetic moments
are reduced to about half of their clean surface values. This detrimental effect of the
Ta overlayer seems to be confined only to the surface layer and exhibits an efficient
metallic screening of the NiFe substrate.

III. Antiferromagnetic NiMn and

Ferromagnetic/Antiferromagnetic NiFe/NiMn Interface

Calculations were carried out for bulk NiMn and for five-layer slabs of NiMn, namely
and assumed in the structure with the experimental lattice
parameters, a = 3.70 Å, c = 3.54 Å. For the interface study, NiFe/NiMn and
superlattice structures were employed. The L10 atomic ordering in
both the NiFe and NiMn layers was assumed with a (001)/(001) crystallographic ori-
entation at the interface, and again the experimental lattice parameters for bulk NiMn
were adopted. The lattice parameters in the basal plane of the NiFe were assumed
to match those of NiMn, but with the c/a ratio chosen to conserve the experimental
atomic volume of the Ni0.5Fe0.5 alloy. We confirmed that this structural approximation
does not significantly alter the magnetic properties of NiFe. Although very important,
the effect of lattice relaxation is outside the scope of the present first step study, and
is therefore not discussed.
Table III shows the calculated spin magnetic moments for bulk NiMn, and for
five-layer films of NiMn. Note that the Mn atoms are located on two sublattices with
moments that have the same magnitude but opposite signs. For bulk NiMn, although
the calculated Mn moment is smaller than experiment ', the
calculations predict the AFM ordering observed in experiments. The magnetic mo-
ments at the surface in NiMn are very sensitive to the environment. For the film with
Mn on the surface, the magnetic moment is enhanced by 0.28 Interestingly, if the
Ni atom is at the surface, the magnetic moment does not exhibit a large change, as
observed in NiFe case (Table I). The calculated magnetic moments for the NiFe/NiMn
and superlattice structures are also given in Table III. FM ordering
of the Fe magnetic moments in the NiFe layer and AFM ordering of the Mn mag-
netic moments in the NiMn layer were found. For except for the Ni
atom at the interface, the magnetic moments recover almost identical values to the
bulk. Note that the non-vanishing moment of the Ni atom at the interface, with a
ferromagnetic moment alignment, was induced.
Let us now discuss the magnetic anisotropy. Table IV shows the calculated MCA
energy, , defined by the difference in total energies for the magnetic moments
oriented perpendicular (c-axis) and in-plane (a-axis). Contrary to bulk NiFe that has
388 A. J. Freeman et al.

Table III
Calculated magnetic moment for b u l k antiferromagnetic NiMn, for five-layer
films of N i M n , and for NiFe/NiMn and superlattice structures. S,
S–1 and C for the five-layer films represent the surface, subsurface and center layers,
respectively. For the superlattices, the interface Ni atom between the NiFe and NiMn
layers are represented by N i ( I ) . Because antiferromagnetic structures consist of two
sublattices, values in parentheses are the moments in the other sublattice.

the perpendicular easy magnetic axis with a small magnitude of the MCA energy, bulk
NiMn exhibits an in-plane easy axis with a relatively large MCA energy of 0.5 meV.
In order to discuss its origin, the MCA energy, of NiMn, is plotted in Fig. 1
with respect to the change in the number of valence electrons by changing the band
filling resulting from varying the highest occupied states around In this band
filling dependence, the change in the number of valence electrons, is given rela-
tive to the physical values, i.e., it equals zero when the number of valence electrons
is that of NiMn. The figure also shows the and contributions to the
MCA energy, which are calculated from the coupling between occupied spin-down
and unoccupied spin-down states, and that between occupied spin-up and unoccupied
spin-down states, respectively. When the negative MCA is dominated by
negative contributions. As increases, the negative contribution grad-
ually decreases, while the positive contribution is developed because the empty
spin-down state is gradually occupied.
For the surface and interface studies, we found that the MCA is sensitive to the
environment, as seen in Table IV. The MCA in surface cases exhibits qualitatively
the same behavior as bulk that shows in-plane anisotropy, but when the Ni atom is
at the surface there is a larger negative MCA energy that creates a strong in-plane
anisotropy. The NiFe/NiMn and superlattice cases show in-plane
Magnetism and Magnetic Anisotropy in Exchange BIAS Systems 389

Table IV
Magnetocrystalline anisotropy energy, , (in meV/cell) for bulk antiferromag-
netic NiMn, for five-layer films of NiMn, and for the NiFe/NiMn and
superlattice structures.

magnetic orientation, mainly dominated by the in-plane MCA from the NiMn layer.
However, the interface effects on the MCA may not be negligible. In Fig. 2, the
band filling dependence of and contributions for NiFe/NiMn and
with respect to the number of their valence electrons, are plotted.
Although similar behavior of the band filling dependence in with
that of bulk NiMn were observed (since the contribution from the NiMn layer tends
to dominate the MCA energy), the changes in the band f i l l i n g dependence from that
of bulk are emphasized especially in the N i F e / N i M n case—which might be impor-
tant for discussing the exchange bias. This is because the magnetic anisotropy at the
FM/AFM interface, as well as the interface exchange coupling, may play a crucial role
in determining the exchange bias phenomenon.
390 A. J. Freeman et al.

IV. Conclusion
We presented results of magnetism and magnetic anisotropy in FM NiMn thin film,
AFM NiMn, and their interface obtained from first-principles FLAPW calculations.
The magnetic moments of Fe in NiFe and Mn in N i M n were found to depend sensitively
on the environment, i.e., surface, relaxation and atomic composition while those of
Ni do not change much. Upon contact with the Ta overlayer, t h e surface atoms of the
substrate suffer a significant reduction of their magnetic moments, which is confined
to the surface layer due to the efficient metallic screening. The large induced magnetic
moment of Ta and the change of its magnetic ordering, due to the surface layer,
Magnetism and Magnetic Anisotropy in Exchange BIAS Systems 391

imply a possible new finding of 5d magnetism. Their MCA values are also drastically
influenced by their environment. Bulk NiFe and NiMn show slightly perpendicular and
in-plane easy magnetic axes, respectively. However, the effects of the surface change
their MCA direction and strength. For the interface between NiFe and NiMn, the
interface effects on the MCA are not negligible. These researches, coupled with the
micro-magnetic modeling at the device level and the microstructure control allowed
by advanced process techniques, are expected to offer possible improvement of the
magnetic materials desired by high density magnetic recording and sensor materials.

References

1. E. Wimmer, H. Krakauer, M. Weinert, and A. J. Freeman, Phys. Rev. B 24, 864

( 1 9 8 1 ) and references therein; M. Weinert, J. Math. Phys. 22, 2433 (1981).
2. J . Nogues and I. K. Schuller, J. Magn. Magn. Mater. 192, 203 (1999).
3. A. E. Berkowits and K. Takano, J. Magn. Magn. Mater. 200, 552 (1999).
4. T. Lin, D. Mauri, N. Staud, C. Hawng, J. K. Howard, and G. L. Gorman, Appl.
Phys. Lett. 65, 1183 (1994).
5. X. Portier, A. K. Petford-Long, and T. C. Anthony, IEEE Trans. on Magn. 33,
3679 (1997).
6. R . Wu and A. J. Freeman, J. Magn. Magn. Mater. 200, 498 (1999).
7. D . S. Wang, R. Wu, and A. J. Freeman, Phys. Rev. Lett. 70, 869 (1993).
8. X . D. Wang, R. Wu, D. S. Wang, and A. J . Freeman, Phys. Rev. B 54, 61 (1996).
9. R. Yu, D. Singh, and H. Krakauer, Phys. Rev. B 43, 6411 (1992).
10. R. Wu and A. J. Freeman, Phys. Rev. B 45, 7205 (1992).
11. J . S. Kasper and J. S. Kouvel, J. Phys. Chem. Solids 11, 238 (1959).
12. L. Pal, E. Kren, C. Kadar, P. Szabo, and T. Tarnoczi, J. Appl. Phys. 59, 538
(1968).