Increased ferromagnetic resonance linewidth and exchange anisotropy in

NiFe/FeMn bilayers
Shujuan Yuan, Baojuan Kang, Liming Yu, Shixun Cao, and Xinluo Zhao

Citation: J. Appl. Phys. 105, 063902 (2009); doi: 10.1063/1.3086292

View online: http://dx.doi.org/10.1063/1.3086292
View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v105/i6
Published by the American Institute of Physics.

Related Articles
Damping effect on resonance bounds relationship of nanostructured ferromagnets and composites
J. Appl. Phys. 111, 113912 (2012)
Beyond ferromagnetic resonance: The inertial regime of the magnetization
Appl. Phys. Lett. 100, 192407 (2012)
Towards precise measurement of oscillatory domain wall by ferromagnetic Josephson junction
Appl. Phys. Lett. 100, 152402 (2012)
Theory of ferromagnetic wires resonating in the proximity of a ground plane: Application to artificial impedance
surfaces
J. Appl. Phys. 111, 064911 (2012)
Effect of packing fraction on ferromagnetic resonance in NiFe2O4 nanocomposites
J. Appl. Phys. 111, 07E348 (2012)

Additional information on J. Appl. Phys.

Journal Homepage: http://jap.aip.org/
Journal Information: http://jap.aip.org/about/about_the_journal
Top downloads: http://jap.aip.org/features/most_downloaded
Information for Authors: http://jap.aip.org/authors

Downloaded 15 Jun 2012 to 137.132.123.69. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
 JOURNAL OF APPLIED PHYSICS 105, 063902 共2009兲

Increased ferromagnetic resonance linewidth and exchange anisotropy

in NiFe/FeMn bilayers
Shujuan Yuan,1,2,a兲 Baojuan Kang,1 Liming Yu,1,2 Shixun Cao,1,2 and Xinluo Zhao1,2
1
Department of Physics, Shanghai University, Shanghai 200444, China
2
Institute of Low-dimensional Carbons and Device Physics, Shanghai University, Shanghai 200444, China
共Received 13 November 2008; accepted 15 January 2009; published online 17 March 2009兲
In the past investigations, the exchange-biased bilayers show much larger ferromagnetic resonance
共FMR兲 linewidth than that of single ferromagnetic layer films. However, the mechanism of the large
linewidth remains controversial. In this paper, the FMR linewidths of NiFe/FeMn bilayers prepared
by dc magnetron sputtering system are systematically studied. Besides the intrinsic damping and
magnetic inhomogeneity, the extrinsic relaxation based on two-magnon scattering process should be
also considered to explain the strong in-plane angular dependence of the linewidths. The fitting of
the out-of-plane angular dependence of the linewidths shows that the intrinsic Gilbert damping
effect plays a major role in the increased linewidth in the bilayers. The value of the g factor
increases due to the effect of the exchange coupling at the NiFe/FeMn interface, resulting in an
enhancement in the Gilbert damping factor G. The fitting results indicate that the line broadening of
the exchange-biased films is related to the exchange anisotropy. The effects of FeMn layer thickness,
growth sequence of FeMn layer, and temperature on the linewidths further approve this
interpretation. © 2009 American Institute of Physics. 关DOI: 10.1063/1.3086292兴

I. INTRODUCTION intrinsic Gilbert damping effect and the broadening induced

by the magnetic inhomogeneity.15 The Gilbert damping fac-
Exchange bias refers to the shift in the hysteresis loop tor is inversely proportional to the CoNi layer and propor-
along the magnetic field axis, which is typically observed in tional to 共g − 2兲2 with the variation in the CoNi layer thick-
exchange coupling ferromagnetic 共FM兲/antiferromagnetic ness, where 共g − 2兲 is the deviation from the g factor of the
共AF兲 materials.1,2 During the past decade, FM/AF bilayers
free electron. The value of g can be influenced by the effect
have been extensively investigated both from fundamental
of exchange coupling at the FM/AF interface, such as the
and technological points of view mainly due to their appli-
in-plane uniaxial anisotropy and isotropic resonance field
cations in the development of magnetic random access
shift.7,15 Speriosu et al.5 and Stocklein et al.6 also found that
memories and giant magnetoresistance read heads based on
the AF and FM thickness dependences of the linewidth are
spin valves.3,4 However, a satisfactory understanding about
similar to that shown by the exchange field. This implies that
the exchange bias has not been developed. Recently, ferro-
the linewidth broadening is related to the exchange aniso-
magnetic resonance 共FMR兲 and Brillouin light-scattering
techniques have been widely used to study the exchange- tropy. In NiFe/FeMn/NiFe trilayer, it is found that the line-
biased FM/AF bilayers.5–15 The in-plane isotropic resonance width of the bottom NiFe layer is larger than that of the top
field shift and broadened linewidths are hot topics. The line- layer, which might be related to the stronger exchange cou-
widths measured by both techniques are more than one order pling at the bottom NiFe/FeMn interface.16 The correlation
of magnitude larger than in similar FM films without ex- between the linewidth and exchange anisotropy was first
change bias5–15 and scale with t−2 plus a constant term,12–14 studied by Stocklein based on Malozemoff’s model, in which
where t is the FM layer thickness. This has been interpreted the AF layer is broken up into a multidomain pattern due to
as the sum of two independent contributions, an intrinsic the interface roughness.17 It is believed that the linewidth
mechanism dominated by Gilbert damping and an extrinsic broadening is related to the inhomogeneous coupling be-
mechanism dominated by two-magnon scattering due to fluc- tween the FM magnetization and the AF grains. In this way,
tuations in the interlayer exchange coupling caused by inter- a systematic linewidth measurement and quantitative inter-
face roughness.12–14 Furthermore, the two-magnon scattering pretation with a theoretical model are required to establish
provides the dominant contribution and increases with de- the proper relaxation mechanism.
creasing FM layer thickness while intrinsic Gilbert damping In this work, the angular behavior, FM and AF thickness
has negligible contribution compared with that of two- dependences, and temperature dependence of the FMR line-
magnon scattering.12–14 width in exchange-biased NiFe/FeMn bilayers are systemi-
However, in our previous work, the dependence of the cally investigated. The angular dependence of the FMR spec-
linewidth on the out-of-plane angle and the FM layer thick- tra is analyzed by the Landau–Lifshitz–Gilbert 共LLG兲
ness in CoNi/FeMn bilayers can be explained in terms of the equation. It is found that the out-of-plane angular depen-
dence of the linewidths can be described by the intrinsic
a兲
Author to whom correspondence should be addressed. Electronic mail: Gilbert damping effect and the magnetic inhomogeneity. The
[email protected]. relation of the Gilbert damping factor G and exchange bias

0021-8979/2009/105共6兲/063902/7/$25.00 105, 063902-1 © 2009 American Institute of Physics

Downloaded 15 Jun 2012 to 137.132.123.69. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
063902-2 Yuan et al. J. Appl. Phys. 105, 063902 共2009兲

z z in-plane measurements, as shown in Fig. 1共a兲, the dc mag-

H netic field H is applied in the x-y plane and the microwave
field h lies along the z axis. The in-plane angular dependence
M of measurements is performed with the magnetic field H
h θ
θH rotated in the film plane by varying the azimuthal angle ␾H
with respect to the unidirectional axis x, which varies from 0
φH
y
h
y to 2␲. For the out-of-plane measurements, as shown in Fig.
film φ H film φ 1共b兲, the magnetic field H is applied in the y-z plane and the
M microwave field h lies along the x axis. The polar angle ␪H is
x x the angle between the normal direction of the film plane and
(a) (b) the applied field H. The out-of-plane angular dependence of
measurements is performed by varying ␪H from 0 to ␲ / 2.
FIG. 1. The coordinate system used in the FM resonance measurements.
The orientation of the magnetization M is expressed in terms
Sample orientations for 共a兲 in-plane and 共b兲 out-of plane measurements are
indicated. H is the applied magnetic field and h is the external microwave of the angles ␾ and ␪. In terms of these angles, the free
field. energy per unit area can be written as
F = − HMtFM关sin ␪ sin ␪H cos共␾ − ␾H兲 + cos ␪ cos ␪H兴
field HE indicates that the broadened linewidth of biased lay-
ers result from the enhanced Gilbert damping due to ex- + 共2␲ M 2tFM − KS兲cos2 ␪ − KUtFM sin2 ␪ cos2 ␾
change coupling with the AF FeMn. + Eex共␾兲, 共1兲

II. EXPERIMENT where tFM is the thickness of the FM layer. The first term
represents the Zeeman energy. The second and the third
The samples 共5 ⫻ 0.5 cm2兲 were fabricated on Si 共100兲 terms represent the sum of the shape anisotropy and the in-
substrate by a dc magnetron sputtering system using targets terfacial perpendicular anisotropy, and the uniaxial aniso-
of Cu, Ni81Fe19, and Fe50Mn50. The base pressure before tropy energies, respectively. The fourth term is the FM/AF
deposition was typically 10−8 Torr and Ar pressure was 5 exchange coupling energy, which contains an irreversible ro-
mTorr during deposition. A permanent magnetic field of 130 tation of the AF spins, in addition to a conventional cosine
Oe was applied parallel to the film plane during deposition to term.11 KS and KU are the interfacial and in-plane uniaxial
develop the necessary exchange bias. To facilitate the obser- anisotropies, respectively.
vation of FM layer thickness effect, a wedge-shaped FM The angular dependence of the FMR spectra can be ob-
layer with thickness from 0 to 25 nm was used, whereas the tained by using the LLG equation,

冉 冊
FeMn layer and both the Cu layers had uniform thicknesses
of 15 and 30 nm, respectively. A wedge-shaped FeMn dMជ ៝
layer with thickness from 0 to 25 nm was also prepared, with = − ␥共M ជ 兲+ G M
ជ ⫻H ជ ⫻ dM , 共2兲
eff
dt ␥M 2
dt
FM uniform thickness of 10 nm. Each large sample was
cut into small pieces 共0.1⫻ 0.2 cm2兲 along the wedge direc- where G is the Gilbert damping factor and ␥ is the gyromag-
tion, thus providing many small samples prepared at the netic ratio. From Eq. 共2兲 neglecting damping, the standard
same time but varying in NiFe or FeMn thickness. For com- condition for resonance becomes

冉冊 冋冉 冊冉 冊 冉 冊 册
parison, a large sample of Si/ Cu 共30 nm兲 / Ni81Fe19
共wedged, 0 – 25 nm兲 / Cu 共30 nm兲 was also prepared at the ␻ 2
1 ⳵ 2F ⳵ 2F ⳵ 2F 2
= − , 共3兲
same conditions. No further field annealing and cooling pro- ␥ 2
M 2tFM sin2 ␪ ⳵ ␪2 ⳵␸2 ⳵␪ ⳵ ␸
cedures were carried out.
where ␻ is the angular frequency.
The hysteresis loops measurements were performed by a
For in-plane FMR spectra of the NiFe/FeMn bilayer, as-
vibrating sample magnetometer 共VSM兲. FMR measurements
suming 4␲ M eff Ⰷ Hres Ⰷ HK , HEFMR and evaluating Eq. 共3兲 for
were carried out on a Bruker ER 200D-SRC EPR spectrom-
␪ = ␲ / 2 and ␾H = ␾, the in-plane resonance field can be sim-
eter, with a fixed microwave frequency of 9.78 GHz and
plified as10,11,18
swept external dc field. The samples were mounted on the
side of a quartz rod and a goniometer was used to vary the
angle. Thus the angular dependence of the resonance field
Hres and the peak-to-peak linewidth ⌬Hpp can be obtained.
Hres = 冉冊 ␻
␥
2
1
4␲ M eff
− HISO − HEFMR cos ␾H

− HK cos 2␾H , 共4兲

III. PHENOMENOLOGICAL DESCRIPTION
where the effective demagnetization field 4␲ M eff = 4␲ M
OF FMR
− 2KS / MtFM and the uniaxial anisotropy field HK = 2KU / M.
The coordinate systems used for measurements and HISO comes from the irreversible rotation of the AF spins.
analysis of the in-plane and out-of-plane angular depen- The exchange bias field HEFMR is the coefficient of the cosine
dences of FMR are described as follows, as shown in Fig. 1. term in −共1 / MtFM兲共d2Eex / d␾2兲. ␻2 / ␥24␲ M eff and HISO are
Let the NiFe/FeMn bilayer lie in the x-y plane, with the z independent of the angle ␾H. By using Eq. 共4兲, the experi-
axis normal to the film plane. The uniaxial and unidirectional mental in-plane angular dependence of the resonance field
axes of the bilayer are aligned along the x direction. For can be fitted and HEFMR, HISO, and HK can be determined.

Downloaded 15 Jun 2012 to 137.132.123.69. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
063902-3 Yuan et al. J. Appl. Phys. 105, 063902 共2009兲

Neglecting the in-plane anisotropy, the out-of-plane dis- (a)

persion equation can be expressed as

冉冊
1200
␻ 2
= 关Hres cos共␪H − ␪兲 + 4␲ M eff cos 2␪兴

Ηres (Oe)
␥
⫻关Hres cos共␪H − ␪兲 + 4␲ M eff cos2 ␪兴. 共5兲 1000 Single 6.5 nm
Bilayer 6.5 nm
In numerical calculations of the above dispersion equation,
Bilayer 10.6 nm
the following equilibrium equations of the magnetization Bilayer 20.5 nm
vector are required: 800
Single 6.5 nm
⳵ F/⳵ ␪ = 0, ⳵ F/⳵ ␾ = 0. 共6兲 Bilayer 6.5 nm (b)
120
Bilayer 10.6 nm
After considering the contribution of the intrinsic Gilbert
Bilayer 20.5 nm
damping ⌬Hintrin and inhomogeneity induced broadening

∆Hpp (Oe)
⌬Hinhom, one has the peak-to-peak linewidth ⌬Hpp: 90

1 1
⌬Hpp =
冑3 ⌬Hintrin + 冑3 ⌬Hinhom , 共7兲
60

⌬Hintrin = 冏 冏 冉
⳵␻
⳵H
−1
G
M 2 F␪␪ +
F␾␾
sin2 ␪
, 冊 共8兲 0 60 120 180 240 300 360
φΗ(deg)
⌬Hinhom = 冏 冏
⳵ Hres
⳵ 共4␲ M eff兲
⌬共4␲ M eff兲 +
⳵ Hres
⳵ ␪H
⌬␪H 冏 冏 FIG. 2. 共Color online兲 In-plane angular dependence of 共a兲 resonance fields

冏 冏
and 共b兲 linewidths for the single NiFe layer film and NiFe/FeMn bilayer
⳵ Hres with different NiFe thicknesses. The solid lines in 共a兲 are fitted results. The
+ ⌬␾H , 共9兲 inserted numbers are the NiFe layer thicknesses.
⳵ ␾H
where the coefficient of 1 / 冑3 is the correction for the differ- phenomena were discussed in our previous work.18 Here we
ence between the full width at half maximum and the peak- focused on the study of the ⌬Hpp. From Fig. 2共b兲 we can see
to-peak linewidth for the Lorentzian line shape. ⌬共4␲ M eff兲, that the ⌬Hpp in the NiFe/FeMn bilayer is more than one
⌬␪H, and ⌬␾H represent the spread of the magnitude of the order of magnitude larger than in the single NiFe layer film
effective demagnetizing field and those of the directions of with the same NiFe layer thickness and increases with de-
the crystallographic axes, respectively. creasing NiFe thickness. The ⌬Hpp of the NiFe single layer
film shows little in-plane angular dependence, while the
IV. RESULTS AND DISCUSSION ⌬Hpp for bilayers show strong in-plane angular dependence.
Furthermore, the in-plane angular behavior of ⌬Hpp is simi-
In this section, the angular behavior, the NiFe and FeMn
lar to that of Hres for the bilayers. The maximum of the
thicknesses, and the temperature dependence of the linewidth
linewidth occurs at ␾H = 180°, where the exchange bias field
are discussed. The effect of the deposition sequence of FeMn
is antiparallel to the external field.
on linewidth is also studied.
As the samples are polycrystalline, the strong in-plane
angular dependence of ⌬Hpp cannot result from the fluctua-
A. In-plane and out-of-plane angular dependences of
tions by the effective magnetization as well as the applied
linewidths
field direction and anisotropy field direction. We try to ana-
Figures 2共a兲 and 2共b兲 show typical in-plane angular de- lyze the in-plane angular dependence of ⌬Hpp by using Eqs.
pendence of the resonance fields Hres and linewidths ⌬Hpp 共4兲 and 共6兲–共9兲. Unfortunately, the fitting results do not co-
for the single NiFe layer film and the NiFe/FeMn bilayers incide with the experimental data. That is to say, only the
with different NiFe thicknesses. The single NiFe layer film intrinsic damping and magnetic inhomogeneity cannot ex-
has the same in-plane Hres at ␾H = 0°, 180°, and 360°, and the plain the strong in-plane angular dependence of ⌬Hpp. An-
angular dependence of Hres can be fitted by considering other contribution to the linewidth, the extrinsic magnetic
uniaxial anisotropy. For the NiFe/FeMn bilayers, Hres at ␾H relaxation based on two-magnon scattering process should be
= 180° is larger than that at ␾H = 0° and 360°, indicating the considered. According to the utilitarian model proposed by
existence of the exchange biasing, and thus the in-plane an- Arias and Mills, the linewidth has the largest value in the
gular dependence of Hres can be fitted using Eq. 共4兲 by con- parallel geometry and is zero in the perpendicular geometry
sidering the unidirectional and uniaxial anisotropies. The if the magnetic film exhibits an in-plane anisotropy.19 The
solid lines in Fig. 2共a兲 are the fitting results. In the fitting, large spin wave mode broadening7,8,12–14 and the similar in-
HEFMR, HISO, and HK are the fitting parameters, and plane angular dependence of linewidths8 have been observed
␻2 / ␥24␲ M eff can be obtained by fitting the out-of-plane an- and explained as the local variation in the exchange field at
gular dependence of the resonance fields. Apparently, a nega- the FM/AF interface, giving rise to relaxation via two-
tive isotropic in-plane resonance field shift does exist. The magnon scattering processes.7,8,12–14

Downloaded 15 Jun 2012 to 137.132.123.69. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
063902-4 Yuan et al. J. Appl. Phys. 105, 063902 共2009兲

800 2000

∆Hpp-max (Oe)
1200 (a) Bilayer 7.8 nm fitting
Bilayer 5.0 nm 1500
Bilayer

pp (Oe)
Bilayer 7.8 nm 600
Bilayer 20.5 nm
measured
200 Single 1000

∆Hpp (Oe)
fitted
(Oe)

900 Single 20.5 nm 400 intrinsic 500

∆H
∆4πMeff
∆θ 0
pp

600
200 100 5 10 15 20
tFM(nm)
∆H

(a)
0
0 20 40 60 80

300 θΗ(deg) 0
5 fitting

G(10 s )
4

8 -1
0 Bilayer
1200 Single
(b) 3
(Oe)

Bilayer 5.0 nm
900 2
Single 5.0 nm
(b)
pp

600 1
∆H

2.4 Bilayer
300
2.3 Single

g
0 2.2
0 30 60 90
θΗ(deg) 2.1
(c)
2.0
FIG. 3. 共Color online兲 共a兲 Out-of-plane angular dependence of linewidths 5 10 15 20
for the single NiFe layer film and NiFe/FeMn bilayers with different NiFe tFM(nm)
thicknesses. The solid lines are fitted results. Insert shows one example of
linewidth fitting of NiFe/FeMn bilayer with NiFe thickness of 5.0 nm. The FIG. 4. 共Color online兲 FM layer thickness dependence of 共a兲 average in-
contributions of the intrinsic damping and the inhomogeneity broadening plane linewidths, 共b兲 G damping factor, and 共c兲 g factor for the single NiFe
共⌬4␲ M eff and ⌬␪, respectively兲 are displayed. 共b兲 Out-of-plane angular de- layer films and NiFe/FeMn bilayers. The solid line in 共a兲 is a fit with tFM
−2

pendence of linewidths for the single NiFe layer film and NiFe/FeMn bi- dependence plus a constant term of 49 Oe. The dotted lines serve as guides
layer with NiFe thickness of of 5.0 nm. to the eye. The inset of 共a兲 shows FM layer thickness dependence of the
−2
maximum of the out-of-plane linewidth. The solid line is a fit with tFM
dependence plus a constant term.
The out-of-plane angular dependences of ⌬Hpp for NiFe/
FeMn bilayers and NiFe single layer films are shown in Figs. ness, which is mainly due to the increase in the intrinsic
3共a兲 and 3共b兲. The out-of-plane ⌬Hpp for biased bilayers in- damping. The line broadening caused by the spread of direc-
creases with decreasing FM thickness and is larger than that tions of the crystallographic axes is much smaller than that
for single NiFe layer film in the full range of ␪H. For larger of ⌬4␲ M eff and intrinsic damping, especially for larger FM
FM layer thickness, the ⌬Hpp at ␪H = 0° 共perpendicular兲 is layer thickness. The contribution by ⌬4␲ M eff has the largest
smaller than that at ␪H = 90° 共parallel兲, while for smaller FM value in the perpendicular geometry 共␪H = 0°兲 and increases
layer thickness, the ⌬Hpp at ␪H = 0° is larger than that at ␪H with decreasing FM layer thickness. It is apparent that when
= 90°. The ⌬Hpp in the parallel and perpendicular geometries the FM layer thickness increases, ⌬4␲ M eff decreases and
both increase with decreasing FM thickness. The increase in approaches zero. The variation of ⌬4␲ M eff can be attributed
⌬Hpp in the full range of ␪H cannot be explained by two- to the fluctuation of FM thickness 共wedge-shaped samples兲
magnon scattering because the linewidth due to two-magnon and the reduction in 4␲ M eff with FM thickness.15 The ⌬Hpp
scattering is zero around ␪H = 0°.19 The intrinsic Gilbert at ␪H = 0° increase observably with decreasing FM layer
damping and the broadening induced by magnetic inhomo- thickness both for bilayers and single layer films, as shown
geneity should be considered to explain the increased out-of- in Figs. 3共a兲 and 3共b兲. The ⌬Hpp at ␪H = 0° almost have the
plane ⌬Hpp. same value for bilayers and single layer films with the same
The experimental data of ⌬Hpp versus ␪H are analyzed FM layer thickness. This indicates that the difference in the
by using Eqs. 共5兲–共9兲. Before the linewidth fitting, the mea- linewidths between the bilayers and single layer films is
sured out-of-plane angular dependence of Hres has been fitted mainly caused by the variation of the intrinsic Gilbert damp-
by using Eqs. 共1兲, 共5兲, and 共6兲 to obtain the value of the ing.
effective demagnetizing field 4␲ M eff and the Landé g factor
共The data are not shown here兲. The inset of Fig. 3共a兲 shows
B. FM layer thickness dependence of linewidths
one example of the fitting. The fitted ⌬Hpp include the intrin-
sic Gilbert damping and the magnetic inhomogeneity, which Figure 4共a兲 shows the FM layer thickness dependence of
are proportional to the fitting parameters G, ⌬4␲ M eff, and the average in-plane ⌬Hpp for NiFe/FeMn bilayers and NiFe
⌬␪H. It can be seen that ⌬Hpp goes through a maximum single layer films. In both cases, ⌬Hpp increase with decreas-
value near ␪H = 10°. Note that near the same position the ing FM layer thickness tFM, but the ⌬Hpp of bilayers show a
difference between the orientation of the magnetization vec- dramatic FM thickness dependence. The solid curve in Fig.
−2
tor and that of the external field also has a maximum value. 4共a兲 is a fit with tFM dependence plus a constant term of 49
The fitting results show that the intrinsic damping term has Oe, which is similar to the fitting of ⌬Hpp of the NiFe/NiO
the maximum near the same position. The maximum of ⌬Hpp bilayer.12–14 The inset of Fig. 4共a兲 shows FM thickness de-
increases markedly with the decrease in the FM layer thick- pendence of the maximum of the linewidth ⌬Hpp-max for bi-

Downloaded 15 Jun 2012 to 137.132.123.69. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
063902-5 Yuan et al. J. Appl. Phys. 105, 063902 共2009兲

4 (a) 13.2 nm
G (10 s ) (a) 120 12.3 nm
9.7 nm
-1

3 100 6.2 nm
8

∆Hpp (Oe)
3.5 nm
80 0 nm
2
60
1
0.01 0.02 0.03 2 0.04 0.05 0.06
(∆ g) 40
0.25
0 60 120 180 240 300 360
(b)
φΗ(deg)
0.20
(b)
∆g

600
0.15

∆Hpp (Oe)
20.5 nm
0.10 400 10.6 nm
9.1 nm
50 100 150 200 0 nm
HE (Oe) 200
∆Hpp (Oe)

150 (c)
0
0 30 60 90
100 θΗ(deg)

FIG. 6. 共Color online兲 共a兲 In-plane and 共b兲 out-of-plane angular depen-
50 dences of linewidths for the NiFe/FeMn bilayers with different FeMn thick-
0 10 20 30 40 50 nesses. The insert numbers are the FeMn thicknesses.
2 2
HE /1000 (Oe )
bilayers. That is to say, the increased linewidths in the
FIG. 5. 共a兲 G vs ⌬g2 = 共g − 2兲2; 共b兲 共g − 2兲 vs HE; 共c兲 ⌬Hpp vs H2E. The lines present NiFe/FeMn bilayers are related to the exchange an-
in 共a兲, 共b兲, and 共c兲 refer to fitting results with scales of G ⬀ 共g − 2兲2, 共g − 2兲 isotropy. The linear relationship ⌬Hpp ⬀ HE2 is shown in Fig.
⬀ HE, and ⌬Hpp ⬀ H2E, respectively. 5共c兲.
−2
layers. The solid curve in the inset is a fit with tFM depen- C. AF layer thickness dependence of linewidths
dence plus a constant term of 259 Oe. The above fitting
Figures 6共a兲 and 6共b兲 show the typical in-plane and out-
calculations show that the increase in the ⌬Hpp for bilayers is
of-plane angular dependences of ⌬Hpp for NiFe/FeMn bilay-
mainly caused by the variation of the intrinsic Gilbert damp-
ers with different FeMn thicknesses and tFM = 10 nm. The
ing. This can be understood by analyzing the variation of the
in-plane and out-of-plane linewidths increase with increasing
Gilbert damping coefficient G. The fitting calculation shows
FeMn thickness. Especially, with increasing FeMn layer
that the G and g factors increase with decreasing tFM, as
thickness, the in-plane linewidths show stronger angular de-
shown in Figs. 4共b兲 and 4共c兲. More importantly, the G factor
−2 pendence, corresponding to the increase in the exchange bias
is proportional to tFM for NiFe/FeMn bilayers. The solid
−2 fields. The FeMn layer thickness dependence of the average
curve in Fig. 4共b兲 is a fit with tFM dependence plus a constant
in-plane linewidths for NiFe/FeMn bilayers is shown in Fig.
term. Moreover, G is proportional to 共g − 2兲2, as shown in
7共a兲. The linewidth increases with increasing FeMn layer
Fig. 5共a兲, which is consistent with the theoretical prediction
thickness when FeMn layer thickness is smaller than 15 nm
by Elliott.20 Here, 共g − 2兲 is the deviation from the g factor of
and hardly varies when FeMn layer thickness is larger than
the free electron. The relationship of G ⬀ 共g − 2兲2 has been
15 nm. This behavior with AF layer thickness dependence is
found in FM/AF bilayers15 and in other Cu/NiFe/Cu/Pt21 and
very similar to that of the exchange bias field HE, as shown
Co/Pt multilayers.22
in Fig. 7共b兲. It is implied that the exchange anisotropy can
As reported previously,7,15 the value of the g factor can
contribute to the line broadening mechanism.
be influenced by the effect of the exchange coupling at the
FM/AF interface for a fixed FM layer thickness. It is well
D. Effect of FeMn layer growing sequence
known that exchange anisotropy increases with decreasing
on linewidths
tFM for FM/AF bilayers. For the present NiFe/FeMn bilayers,
the exchange bias field HE. As pointed out above, the G The AF and FM layer thickness dependences of line-
factor is proportional to tFM −2
and G ⬀ 共g − 2兲2, so 共g − 2兲2 width indicate that the line broadening is related to the ex-
should be proportional to tFM. Therefore, HE and 共g − 2兲
−2
change anisotropy. To further clarify this interpretation, we
should have the linear relationship, as shown in Fig. 5共b兲. prepared two types of films: type I, Si/Cu/NiFe/FeMn/Cu,
The solid curve in Fig. 5共b兲 is linear fitting. Now, we can get and type II, Si/Cu/FeMn/NiFe/Cu. The thicknesses of FeMn
the relationship ⌬Hpp ⬀ G ⬀ 共g − 2兲2 ⬀ HE2 ⬀ tFM
−2
for the biased and NiFe are 15 and 10.6 nm, respectively. The VSM data

Downloaded 15 Jun 2012 to 137.132.123.69. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
063902-6 Yuan et al. J. Appl. Phys. 105, 063902 共2009兲

(a)
400
80
Si/Cu/NiFe/FeMn/Cu
∆Hpp (Oe)
Si/Cu/FeMn/NiFe/Cu
70 300 Si/Cu/NiFe/Cu

∆Hpp (Oe)
60
NiFe: 10.6 nm, FeMn:15 nm
200
50
(b) 100
60
(Oe)

40 0
0 30 60 90
FMR

θH (deg)
ΗE

20
FIG. 8. 共Color online兲 Out-of-plane angular dependence of linewidths for
0 the single NiFe layer film and NiFe/FeMn and FeMn/NiFe bilayers. The
NiFe and FeMn layer thicknesses are 10.6 and 15 nm, respectively. The
0 10 20 30 solid lines are fitted results using Eqs. 共5兲–共9兲.
FeMn thickness (nm)
Since ⌬Hpp ⬀ G ⬀ 共g − 2兲2 ⬀ HE2 , the enhanced ⌬Hpp of the
FIG. 7. FeMn layer thickness dependence of the 共a兲 average in-plane line- NiFe/FeMn bilayer 共type I兲 is mainly caused by the greater
widths and 共b兲 exchange bias field for NiFe/FeMn bilayer with NiFe thick-
ness of 10 nm. The exchange bias field HFMR was obtained from the FMR
exchange coupling energy.
E
measurement.
E. Temperature dependence of linewidths

show that HE is 86 Oe for type I and 16 Oe for type II The temperature dependence of the in-plane Hres and
samples. The large difference in exchange bias values ob- ⌬Hpp of NiFe single layer film and NiFe/FeMn bilayer with
served for the two types of films is related to two reasons. On tFM = 7.8 mm are shown in Figs. 9共a兲 and 9共b兲. In the FMR
the one hand, the present samples are as-prepared samples experiments, resonance spectra at each temperature are re-
and no further field annealing and cooling are performed. A
unidirectional deposition field is applied during the deposi- 1100
tion process to establish the exchange anisotropy. Thus, dur- 1000
ing the deposition of NiFe/FeMn film 共type I兲, the domain
Hres (Oe)

900
formation in the FeMn film is influenced by a net moment on
800
the NiFe surface. During this process, the FeMn film under-
goes a field-cooling process. However, for the FeMn/NiFe 700 single, 7.5 nm
film 共type II兲, FeMn has already grown which can only be bilayer, 7.5 nm
600 (a)
weakly influenced by the top NiFe layer or by the weak 250
deposition field. Therefore, the spin configurations of the
∆Hpp (Oe)

200 single, 7.5 nm

FeMn layer for NiFe/FeMn and FeMn/NiFe interfaces bilayer, 7.5 nm
should be different. On the other hand, the microstructure of 150
FeMn at the NiFe/FeMn and FeMn/NiFe interfaces might be 100
different, which can influence the exchange coupling be-
tween the AF/FM interfaces.23,24 50
(b)
Figure 8 shows the out-of-plane angular dependence of 0
⌬Hpp for the two types of films and the single NiFe layer 200
HE by VSM
(Oe)

film with NiFe thickness of 10.6 nm. From Fig. 8, one can
see that ⌬Hpp of both NiFe/FeMn and FeMn/NiFe films are
150
∆Ηpp(difference)
pp
HE&∆Η

larger than that of the single NiFe layer film, and ⌬Hpp of the 100
NiFe/FeMn film is larger than that of the FeMn/NiFe film.
50
The measured ⌬Hpp versus ␪H are fitted by Eqs. 共1兲, 共5兲, and
(c)
共6兲. The best-fitted parameters are g = 2.1296, G = 2.001 0
⫻ 108 s−1, and ⌬4␲ M eff = 80 G, ⌬␪ = 0.06° for the NiFe/ 100 150 200 250 300 350
FeMn bilayer, g = 2.1095, G = 1.484⫻ 108 s−1, ⌬4␲ M eff Temperature (K)
= 200 G, and ⌬␪ = 0.05° for the FeMn/NiFe bilayer, and g
FIG. 9. 共Color online兲 Temperature dependence of 共a兲 in-plane resonance
= 2.1048, G = 1.083⫻ 108 s−1, ⌬4␲ M eff = 180 G, and ⌬␪
fields and 共b兲 linewidths for NiFe single layer film and NiFe/FeMn bilayer
= 0.05° for the single NiFe layer. So the difference in ⌬Hpp is with tFM = 7.5 nm. 共c兲 ⌬Hpp共difference兲 = ⌬Hpp共bilayer兲 − ⌬Hpp共single兲 and
mainly caused by the variation in the G damping factor. exchange bias field measured by VSM as a function of temperature.

Downloaded 15 Jun 2012 to 137.132.123.69. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions
063902-7 Yuan et al. J. Appl. Phys. 105, 063902 共2009兲

corded with the applied field oriented parallel and antiparal- ACKNOWLEDGMENTS
lel to the orientation of the deposition field. Thus, the plotted This work was supported by the National Natural Sci-
values of Hres and ⌬Hpp in Figs. 9共a兲 and 9共b兲 are the average ence Foundation of China 共Grant Nos. 10504019 and
of the two resonance fields and linewidths obtained at each
10674092兲 and the Science and Technology Innovation Fund
temperature. For the single NiFe layer film, ⌬Hpp hardly in-
of the Shanghai Education Committee 共Grant No. 09ZZ95兲.
crease with decreasing the temperature. While the ⌬Hpp of
the biased bilayer show a dramatic increase with decreasing 1
W. H. Meiklejohn and C. P. Bean, Phys. Rev. 102, 1413 共1956兲; 105, 904
the temperature. At 110 K, the ⌬Hpp of the NiFe/FeMn bi- 共1957兲.
2
layer is 205 Oe, while the ⌬Hpp of the NiFe single layer film J. Nogués and I. K. Schuller, J. Magn. Magn. Mater. 192, 203 共1999兲.
3
B. Dieny, V. S. Speriosu, S. S. P. Parkin, B. A. Gurney, D. R. Wilhoit, and
is 43 Oe. Here, we define ⌬H pp共difference兲 = ⌬Hpp共bilayer兲 D. Mauri, Phys. Rev. B 43, 1297 共1991兲.
− ⌬Hpp共single兲. The temperature dependence of 4
S. Tehrani, J. M. Slaughter, M. Deherrera, B. N. Engel, N. D. Rizzo, J.
⌬Hpp共difference兲 is shown in Fig. 9共c兲. It is found that Slater, M. Durlam, R. W. Dave, J. Janesky, B. Butcher, K. Smith, and G.
⌬Hpp共difference兲 increase monotonically and almost linearly Grynkewich, Proc. IEEE 91, 703 共2003兲.
5
V. S. Speriosu, S. S. P. Parkin, and C. H. Wilts, IEEE Trans. Magn. 23,
with decreasing temperature down to 110 K. Interestingly, 2999 共1987兲.
this phenomenon is very similar to the temperature depen- 6
W. Stoecklein, S. S. P. Parkin, and J. C. Scott, Phys. Rev. B 38, 6847
dence of the exchange anisotropy field HE for the NiFe/ 共1988兲.
7
FeMn bilayer, as shown in Fig. 9共c兲. This means that ex- R. D. McMichael, M. D. Stiles, P. J. Chen, and W. F. Egelhoff, Jr., Phys.
Rev. B 58, 8605 共1998兲; J. Appl. Phys. 83, 7037 共1998兲.
change anisotropy field has direct relation with the increased 8
C. Mathieu, M. Bauer, B. Hillebrands, J. Fassbender, G. Güntherodt, R.
linewidth for the NiFe/FeMn bilayer. Jungblut, J. Kohlhepp, and A. Reiders, J. Appl. Phys. 83, 2863 共1998兲.
9
P. Miltényi, M. Gruyters, G. Güntherodt, J. Nogués, and I. K. Schuller,
Phys. Rev. B 59, 3333 共1999兲.
10
M. Rubinstein, P. Lubitz, and S. F. Cheng, J. Magn. Magn. Mater. 195, 299
V. CONCLUSIONS
共1999兲.
11
H. W. Xi, K. R. Mountfield, and R. M. White, J. Appl. Phys. 87, 4367
In conclusion, the FMR linewidths in the exchange- 共2000兲.
biased NiFe/FeMn bilayers are systematically studied to 12
S. M. Rezende, A. Azevedo, M. A. Lucena, and F. M. de Aguiar, Phys.
clarify the mechanism of the increased linewidth. In addition Rev. B 63, 214418 共2001兲.
13
to the intrinsic damping and magnetic inhomogeneity, the S. M. Rezende, M. A. Lucena, A. Azevedo, F. M. de Aguiar, J. R. Fermin,
and S. S. P. Parkin, J. Appl. Phys. 93, 7717 共2003兲.
extrinsic damping based on two-magnon scattering process 14
B. K. Kuanr, R. E. Camley, and Z. Celinski, J. Appl. Phys. 93, 7723
should be also considered in the interpretation of the strong 共2003兲.
15
in-plane angular dependence of the linewidths. The out-of- S. J. Yuan, L. Wang, R. Shan, and S. M. Zhou, Appl. Phys. A: Mater. Sci.
plane angular dependence of the linewidths can be fitted by Process. 79, 701 共2004兲.
16
S. J. Yuan, K. Xu, L. M. Yu, S. X. Cao, C. Jing, and J. C. Zhang, J. Appl.
the intrinsic Gilbert damping effect and the magnetic inho- Phys. 101, 113915 共2007兲.
mogeneity. The value of the g factor increases due to the 17
A. P. Malozemoff, Phys. Rev. B 35, 3679 共1987兲; 37, 7673 共1988兲.
18
effect of the exchange coupling at the NiFe/FeMn interface, S. J. Yuan, Y. X. Sui, and S. M. Zhou, Eur. Phys. J. B 44, 557 共2005兲.
19
resulting in an enhancement in the Gilbert damping factor G. R. Arias and D. L. Mills, Phys. Rev. B 60, 7395 共1999兲.
20
R. J. Elliott, Phys. Rev. 96, 266 共1954兲.
Thus, the relationship with ⌬Hpp ⬀ HE2 can be gotten from the 21
S. Mizukami, Y. Ando, and T. Miyazaki, Phys. Rev. B 66, 104413 共2002兲.
fitting. The effects of FeMn layer thickness, growth sequence 22
S. J. Yuan, L. Sun, H. Sang, J. Du, and S. M. Zhou, Phys. Rev. B 68,
of FeMn layer, and temperature on the FMR linewidths of 134443 共2003兲.
23
S. Bae, J. H. Judy, W. F. Egelhoff, Jr., and P. J. Chen, J. Appl. Phys. 87,
NiFe /FeMn bilayers further confirm that the increased line-
6650 共2000兲.
width of the NiFe/FeMn bilayers is related to the exchange 24
M. Pakala, Y. Huai, G. Anderson, and L. Miloslavsky, J. Appl. Phys. 87,
anisotropy. 6653 共2000兲.

Downloaded 15 Jun 2012 to 137.132.123.69. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions