This content has been downloaded from IOPscience. Please scroll down to see the full text.
IP Address: 18.104.22.168
This content was downloaded on 14/10/2013 at 22:08
Please note that terms and conditions apply.
Magnetic exchange hardening in polycrystalline GdN thin films
View the table of contents for this issue, or go to the journal homepage for more
2010 J. Phys.: Condens. Matter 22 302003
HomeSearchCollectionsJournalsAboutContact usMy IOPscience
JOURNAL OF PHYSICS: CONDENSED MATTER
J. Phys.: Condens. Matter 22 (2010) 302003 (5pp)doi:10.1088/0953-8984/22/30/302003
FAST TRACK COMMUNICATION
Magnetic exchange hardening in
polycrystalline GdN thin films
K Senapati, T Fix, M E Vickers, M G Blamire and Z H Barber
Department of Materials Science and Metallurgy, University of Cambridge,
Cambridge CB2 3QZ, UK
Received 8 June 2010
Published 12 July 2010
Online at stacks.iop.org/JPhysCM/22/302003
We report the observation of intrinsic exchange hardening in polycrystalline GdN thin films
grown at room temperature by magnetron sputtering. We find, in addition to the ferromagnetic
phase, that a fraction of GdN crystallizes in a structural polymorphic form which orders
antiferromagnetically. The relative fraction of these two phases was controlled by varying the
relative abundance of reactive species in the sputtering plasma by means of the sputtering power
and N2partial pressure. An exchange bias of ∼30 Oe was observed at 10 K. The exchange
coupling between the ferromagnetic and the antiferromagnetic phases resulted in an order of
magnitude enhancement in the coercive field in these films.
(Some figures in this article are in colour only in the electronic version)
Magnetic ordering and electronic transport in gadolinium
monopnictides have long been of interest both theoreti-
cally [1–6] and experimentally [7–10]. Despite the simple
rock-salt crystal structure, these monopnictides exhibit a wide
variety of magnetic and electrical ordering (see  for a
recent review).In this group of pnictides only GdN is
known to order ferromagnetically, with a Curie temperature
of ∼60 K , while GdP, GdAs, GdSb, and GdBi exhibit
type II antiferromagnetism (AFM) .
the better known EuO , interest in GdN has been revived
recently due to possible applications, e.g spin filters, in low
temperature spintronic devices. Electronic transport in GdN is
stillunderdebate inviewof thereportsofsemiconducting,
semimetallic  as well as insulating  behavior in
the literature. Magnetism in the Gd monopnictides arises
from the half filled 4f band of Gd atoms.
highly localized nature of the 4f shell inhibits any direct
exchange due to extremely low orbital overlap.
indirect exchange mechanisms [3, 15] have been invoked
in order to explain the bulk ferromagnetic (FM) ordering
in these compounds.Recent calculations by Duan et al
 have suggested that an antiferromagnetic superexchange
mechanism coexists in GdN along with the Ruderman–Kittel–
Kasuya–Yosida-type [17–19] indirect exchange interaction.
In common with
First-principle calculations have suggested thatthe nearest
neighbor and the next-nearest-neighbor magnetic exchange
parameters (J1and J2, respectively) are strongly influenced by
changes in the lattice parameter. While both J1and J2increase
with increasing lattice constant of GdN, J2shows a crossover
from FM to AFM.
In thispaper we provide evidence for an antiferromagnetic
phase of GdN which, under certain growth conditions,
intrinsically crystallizes within the normal ferromagnetic GdN
matrix during the deposition of polycrystalline thin films.
Several series of GdN films were prepared in varied N2
environments and with varying sputtering powers.
to other transition metal nitrides , we found a strong
dependence of the physical properties on the nitrogen partial
pressure during film growth. Insulating polycrystalline films
were obtained with saturation magnetization (Ms) ranging
from ∼2.3 μB/Gd ion to the ideal Hund’s rule value of
7 μB/Gd ion . In this paper we focus on a set of
samples prepared in a relatively low N2partial pressure of 4%.
Interestingly, we find that the hysteresis loops of GdN films
change progressively from a soft ferromagnetic behavior to a
harder FM phase by increasing the sputtering power at fixed
nitrogen partial pressure. A two-fold increase in sputtering
power (a higher power leads to an increase in Gd flux) led to
© 2010 IOP Publishing LtdPrinted in the UK & the USA
J. Phys.: Condens. Matter 22 (2010) 302003 Fast Track Communication
Figure 1. Panels (a)–(f) zero-field-cooled (ZFC) and field-cooled (FC) magnetization of GdN films deposited in 4% nitrogen partial pressure
with various deposition powers. Magnetization was measured with 0.05 T field applied in-plane to the sample. Panels (g)–(l) magnetic
hysteresis loops of corresponding films measured at 10 K.
an order of magnitude enhancement in the coercive field (Hc).
X-ray diffraction studies reveal the coexistence of two cubic
phases of GdN in these films, with slightly different lattice
All GdN films were deposited in a UHV deposition
chamber at room temperature on 5 × 5 mm2Si chips with
a 250 nm thermally grown SiO2 layer.
formation of GdSi2at the interface the substrates were coated
with a 50 nm thick NbN layer before the deposition of GdN.
We find that GdN films grown directly onto the Si/SiO2
substrate react rapidly and flake off the substrate within a
few days. Since GdN is extremely reactive to atmospheric
moisture, all samples were also capped with a 50 nm thick
protective NbN layer. In all cases the superconducting
transition temperature of the NbN was below 5 K (due
to non-optimal deposition conditions) and did not interfere
In order to avoid
with the magnetization measurements performed at 10 K
and above. Since the Curie temperature (Tc) of GdN is
adversely affected by very low levels of oxygen dopants ,
the background oxygen level in our deposition chamber was
minimized via prolonged sublimation pumping before each
deposition, attaining a base pressure of ∼2 × 10−9mbar. GdN
was deposited in a sputtering gas environment containing 4%
N2mixed with argon at a pressure of 1.5 Pa. Sputtering power
density was varied from 0.49 to 0.96 W cm−2at the target
surface.The magnetization measurements were performed
in a vibrating sample magnetometer based on a variable
temperature insert from Cryogenic Limited.
In figure 1 we compare the equilibrium magnetization of a
series of GdN films grown in 4% N2with varying sputtering
power.The Curie temperature increased with increasing
sputtering power, as evident from panels (a)–(f) of figure 1.
J. Phys.: Condens. Matter 22 (2010) 302003 Fast Track Communication
Figure 2. The in-plane coercive field and the saturation
magnetization of GdN films at 10 K is plotted as a function of
sputtering power. The inset shows the high-field hysteresis loops of
these films measured at 10 K. Solid lines are to guide the eye.
The corresponding isothermal hysteresisloops (panels (g)–(l)),
measured at 10 K, become wider at higher powers. Samples
grown at higher powers do not saturate upto a field of 5 T. The
gradual increase in the coercive field with sputtering power
was found to be a common feature of samples deposited in
any fixed nitrogen partial pressure.
the power dependence of the coercive field and saturation
magnetization at 4% N2. Mswas calculated by extrapolating
the linear high-field moment to zero field, after background
subtraction. The high-field hysteresis loops are shown in the
inset of figure 2. From figure 2 we note that the gain in
coercivity isaccompanied by a lossin saturation magnetization
in this series of samples. We mention here that the maximum
value of Ms for a fixed N2approaches the theoretical value
of 7 μB/Gd only at the high N2partial pressure. This may
indicate some N2deficiency in these samples. Preliminary x-
ray photo-emission spectroscopy (XPS) measurements (on 6%
N2samples) showed a sub-stoichiometric nitrogen fraction at
the highest sputtering powers.
In figure 3 we show the high angle 2θ–θ x-ray diffraction
patterns of a series of GdN films grown in 4% N2. Rietveld
refinement of four strongest reflections of GdN for the
sample deposited at lowest power (soft FM) revealed a
lattice parameter of 5.034(5) ˚ A, which is slightly larger than
the reported  lattice parameter of GdN (4.999 ˚ A).
shown in figure 3, increasing sputtering power results in a
gradual broadening of the GdN peaks.
the strongest (111) peak of GdN reveals that the broadening
is also accompanied by the appearance of a second peak
corresponding to a slightly larger lattice parameter. These two
peak positions are marked by the dashed lines in figure 3.
If the increasing abundance of Gd ions in the sputtering
plasma leads to an increased Gd content in a spatially random
manner in the fcc lattice of GdN, then one would expect a
gradual broadening of thex-ray peaks along witha lateral shift.
However, the evolution of a second peak close to the original
shows separation into at least two cubic phases with different
In figure 2 we show
A closer look at
Figure 3. 2θ–θ x-ray diffraction patterns of a series of GdN films
deposited in 4% N2environment with power densities 0.49, 0.56,
0.63, 0.69, 0.76, 0.83, 0.89 and 0.96 W cm−2. The curves are shifted
on the y-axis for clarity.
cell parameters. Based on the position of this second peak,
the lattice parameter of the second phase of GdN (denoted
as GdN-II hereafter) was estimated to be ∼5.12 ˚ A. At the
highest power densities used in this study, a small fraction
of pure Gd condenses in the film as evident from figure 3.
The appearance of pure Gd in the films deposited at the
highest powers (figure 3) may be argued to play a role in the
observed coercivity enhancement. However, we emphasize
the fact that Hc enhancement, as a function of sputtering
power, is observed well before the appearance of Gd in the
films. In fact, Hcenhancement was found to be a common
feature in samples (not discussed in this paper) deposited at
6%, 8% and 10% N2partial pressures, which did not show
any signature of Gd in the x-ray diffraction measurements.
In the most likely scenario, the randomly distributed Gd
nano-granules may add a superparamagnetic contribution to
the overall hysteresis loop of GdN films.
a contribution may increase the saturation field of the GdN
samples, all superparamagnetic systems are characterized by
extremely low coercivity. Condensation of larger grains of Gd,
which can order ferromagnetically, is also unlikely to enhance
the Hcbecause pure Gd is known  to have extremely low
(<20 Oe) coercivity. Therefore, we exclude any contribution
of Gd to the enhanced coercivity observed in our samples.
In view of the continuous evolution of the GdN-II phase
with increasing sputtering power, it is reasonable to assume
that under all sputtering conditions polycrystalline GdN films
contain both GdN and GdN-II phases with varied fraction.
Magnetization measurements show that increasing fraction
of GdN-II, with ∼1.8% larger lattice parameter, directly
correlates with the increasing coercive field. Simultaneously,
the ZFC magnetization shows a thermal blocking behavior at
low temperature. These observations suggest that GdN-II is
either a relatively harder ferromagnet or an antiferromagnet.
However, as evident from figure 1, the ratio of remnant and
saturation magnetization (MR/MS) of GdN films decreases
considerably with increasing power and hence with increasing
J. Phys.: Condens. Matter 22 (2010) 302003 Fast Track Communication
GdN-II fraction. For the samplesdepositedat 0.96W cm−2the
MR/MSvalue drops below 0.5. If GdN-II was magnetically
harder than GdN, one would expect the (MR/MS) to approach
unity with increasing fraction of GdN-II. This excludes the
possibility of a hard ferromagnetic nature of GdN-II. On the
other hand, the larger lattice parameter of the GdN-II is more
favorable for an antiferromagnetic ordering due to increased
exchange parameters J1and J2. The loss in high-field
Ms(see figure 2) as a function of evolving fraction of GdN-II
(figure 3) is strongly suggestive of an antiferromagnetic nature
A more objective proof of the antiferromagnetic nature
of GdN-II can be obtained from the exchange biasing effect
of these films. In figure 4(a) we compare the temperature
dependence of the coercive fields (HcPosand HcNeg) of GdN
films grown at 0.49 and 0.83 W cm−2.
are the coercive fields obtained from the field increasing and
field decreasing branch of the hysteresis loop, respectively.
Here we mention that within our instrumental uncertainties
(±5 Oe) the zfc and fc hysteresis loops did not show any
relative shift along the field axis, which is the hallmark of
all FM–AFM systems.However, the hysteresis loops of
the high power samples (with larger GdN-II fraction) were
found to be shifted from origin. We believe that due to the
extremely soft FM nature of the GdN phase in the samples,
the residual field of our magnetometer (∼25 Oe) was sufficient
to switch a significant fraction of the FM phase even in the
zfc condition. As a result the relative loop shift between
zfc and fc magnetizations were smeared out. Therefore we
have measured the hysteresis loops at various temperatures
across the N´ eel temperature to isolate the absolute value of
exchange bias from the instrumental artifact. As shown in
figure 4(a), the HcPosand HcNegdid not have any measurable
difference for the sample deposited at 0.49 W cm−2while
there is a clear split between HcPos(T) and HcNeg(T) curves
at low temperatures for the GdN-II dominant (0.83 W cm−2)
sample. Below this temperature the GdN-II phase orders
antiferromagnetically and the resulting exchange interaction
between GdN and GdN-II phases causes an enhancement of
Hc. Such exchange hardening has previously been reported
in composite FM–AFM systems .
for the 0.83 W cm−2sample, calculated from figure 4(a) as
|(HcPos− HcNeg)/2|, is plotted in the inset of figure 4(a) as a
function of temperature. FollowingMalozemoff’s  random
field model for FM–AFM layers withcubic anisotropy of AFM
layers, the temperature dependence of exchange bias in a FM–
AFM system can be written as HEx =√AAFMKAFM(0)(1 −
T/TN) assuming the exchange stiffness AAFMis constant with
KAFM is the anisotropy energy and TN is the
N´ eel temperature. This model fits reasonably to our data at
low temperature, as shown by the solid line in the inset of
figure 4(a). The N´ eel temperature obtained from this fit was
36±1.5K. The standard modelof coercivity enhancement 
associated with exchange bias assumes that the interfacial
exchange couplingofAFM particlesdistributedinaFM matrix
is larger than the anisotropy energy of the particles. Therefore,
the effective anisotropy energy (and hence the coercivity)
experienced by the FM scales with the volume fraction of the
HcPos and HcNeg
The exchange bias
Figure 4. (a) Temperature dependent coercive field of two samples
deposited in 4% N2partial pressure with 0.49 and 0.83 W cm−2
sputtering power. The lines are guides to the eye. Inset: temperature
dependence of exchange bias for the sample deposited at
0.83 W cm−2. The solid line is a fit to Malozemoff’s model .
(b) Coercivity as a function of sputtering power, compared with the
relative fraction of AFM (GdN-II) phase for samples deposited in 4%
nitrogen. The solid line is to guide the eye.
AFM phase. A roughestimationof the relative volume fraction
of the AFM phase in the series of samples deposited in 4%
N2was done using the relation f = 1 − Ms/Ms(Max), where
Ms(Max)is the maximum saturation moment obtained for the
4% samples. As shown in figure 4(b), Hcindeed follows the
same pattern as the AFM volume fraction f .
To summarize, we have studied the magnetic behavior
of polycrystalline GdN thin films deposited with varying
sputtering power and N2partial pressure. A gradual soft-to-
hard FM crossover was observed as a function of increasing
sputtering power at fixed N2partial pressure. X-ray diffraction
studies showed that two separate cubic phases of GdN coexist
in these films, with ∼1.8% difference in lattice parameter.
The GdN-II phase, with a larger lattice parameter, was found
to be antiferromagnetic with a N´ eel temperature ∼36 K.
Exchange coupling between the antiferromagnetic GdN-II
J. Phys.: Condens. Matter 22 (2010) 302003Fast Track Communication
and ferromagnetic GdN resulted in an effective magnetic
hardening, enhancing the coercivity by an order of magnitude.
The single step control of coercivity in thin films of GdN, with
native FM and AFM phases, may find use in low temperature
We thank Dr A Walton for the XPS measurements at the
Leeds EPSRC Nanoscience and Nanotechnology Facility
(LENNF). Financial support for this work was provided by the
 Hasegawa A and Yanase A 1997 J. Phys. Soc. Japan 42 492
 Petukhov A G, Lambrecht W R L and Segall B 1994 Phys. Rev.
B 50 7800
 Kasuya T and Li D X 1997 J. Magn. Magn. Mater. 167 L1
 Larson P and Lambrecht W R L 2006 J. Phys.: Condens.
Matter 18 11333
 Doll K 2008 J. Phys.: Condens. Matter 20 075214
 Mitra C and Lambrecht W R L 2008 Phys. Rev. B 78 134421
 Wachter P and Kaldis E 1980 Solid State Commun. 34 241
 Gerlach J W, Mennig J and Rauschenbach B 2007 Appl. Phys.
Lett. 90 061919
 Leuenberger F, Parge A, Felsch W, Fauth K and
Hessler M 2005 Phys. Rev. B 72 014427
 Li D X, Haga Y, Shida H, Suzuki T, Kwon Y S and
Kido G 1997 J. Phys.: Condens. Matter 9 10777
 Duan C, Sabirianov R F, Mei W N, Dowben P A, Jaswal S S
and Tsymbal E Y 2007 J. Phys.: Condens. Matter
 Schmehl A et al 2007 Nat. Mater. 6 882
 Kaldis E and Zurcher C 1974 Helv. Phys. Acta 47 421
 Xiao J Q and Chien C L 1996 Phys. Rev. Lett. 76 1727
 Wachter P, Kaldis E and Hauger R 1978 Phys. Rev. Lett.
 Duan C, Sabiryanov R F, Mei W N, Dowben P A, Jaswal S S
and Tsymbal E Y 2006 Appl. Phys. Lett. 88 182505
 Ruderman M A and Kittel C 1954 Phys. Rev. 96 99
 Kasuya T 1956 Prog. Theor. Phys. 16 45
 Yosida K 1957 Phys. Rev. 106 893
 Duan C, Sabiryanov R F, Liu J, Mei W N, Dowben P A and
Hardy J R 2005 Phys. Rev. Lett. 94 237201
 Bacon D D, English A T, Nakahara S, Peters F G, Schreiber H,
Sinclair W R and van Dover R B 1983 J. Appl. Phys.
 Rhyne J J and McGuire T R 1972 IEEE Trans. Magn. 8 105
 Barth A, Treubel F, Marszalek M, Evenson W, Hellwig O,
Borschel C, Albrecht M and and Schatz G 2008 J. Phys.:
Condens. Matter 20 395232
 Sort J, Nogu´ es J, Suri˜ nach S, Mu˜ noz J S, Bar´ o M D, Chappel E,
Dupont F and Chouteau G 2001 Appl. Phys. Lett. 79 1142
 Malozemoff A P 1987 Phys. Rev. B 35 3679
 Stiles M D and McMichael R D 2001 Phys. Rev. B 63 064405