Direct observation of magnetic phase coexistence and magnetization reversal in a Gd$_{0.67}$Ca$_{0.33}$MnO$_{3}$ thin film

1.1 MB·English

by  Jeehoon Kim

Direct observation of magnetic phase coexistence and magnetization reversal in a
Gd Ca MnO thin film
0.67 0.33 3
Jeehoon Kim, Nestor Haberkorn, Leonardo Civale, Paul Dowden,
Avadh Saxena, J. D. Thompson, and Roman Movshovich
Los Alamos National Laboratory, Los Alamos, NM 87545∗
Evgeny Nazaretski
Brookhaven National Laboratory, Upton, NY 11973
2 We haveinvestigated theferrimagnetic domain structure in a Gd0.67Ca0.33MnO3 thin film using
1 magnetic force microscopy. We observe clear signs of phase separation, with magnetic islands
0 embedded in a non-magnetic matrix. We also directly visualize the reversal of magnetization of
2 ferrimagnetic domains as a function of temperature and attribute it to a change in the balance of
magnetization of anti-aligned Mn and Gd sublattices.
n
a
J
Mixed-valentperovskitemanganitesA1−xBxMnO3 (A widetemperaturerange[12]andinhomogeneousFIM-like
0
and B are rare-earth and divalent alkaline elements, re- behaviorwith anexchangebias effect close to T .[13]
1 comp
spectively), such as La-based manganites, have been Low values of the saturation magnetization (M ) sug-
S
studied extensively in recent years.[1–4] These materials gest phase coexistence.[12, 13] MFM studies described
]
l exhibit a colossal magnetoresistance (CMR) effect for a below, with the wide range of field and temperature em-
e
- widerangeofdopingscenteredatx=1/3wherethedou- ployed, allow us to visualize the magnetic structure of
r
t ble exchange mechanism is maximized.[5] The resulting GCMO and provide direct evidence of phase separation.
s combinationoffascinatingphysicalphenomenaandapo- The magnetization reversal at T of each individual
. comp
t tentialfortechnologicalapplicationshasbeenthedriving domain provides strong support for the scenario of anti-
a
m forceinsustaininghighinterestinthesecompounds.[1–4] aligned Mn and Gd sublattices with the Gd (Mn) mag-
Electronicinhomogeneityandphaseseparationareubiq- netization dominating below (above) T .
- comp
d uitous in doped manganites, and the resulting CMR ef- TheGd0.67Ca0.33MnO3 thinfilmwasgrownbypulsed-
n fectisdrivenbypercolativetransport.[6]CMRmanifests laserdeposition(PLD)onaSrTiO3 (100)substratefrom
o itself by a dramatic drop in resistivity and a discon- acommercialtargetwiththesamechemicalcomposition.
c tinuous decrease in the equilibrium Mn-O bond length Thesubstratetemperaturewaskeptat790◦Cinanoxy-
[
at a first order phase transition in an applied magnetic gen atmosphere at a pressure of 200 mTorr. After depo-

1 field.[7, 8] Their complex electronic structure anda vari- sition, the O2 pressure was increased to 200 Torr, and
v ety of competing interactions lead to a rich ensemble of the temperature was decreased to room temperature at
4 ground states in this family of compounds. a rate of 30 ◦C/min. Bulk GCMO is an orthorhombic
4
1 In this Letter we report a low temperature perovskite (Pbnm (no. 62); a = 5.52 ˚A, b = 5.34 ˚A,
2 magnetic force microscopy (MFM) investigation of c = 7.50 ˚A).[13, 14] The GCMO film was examined by
1. Gd0.67Ca0.33MnO3 (GCMO), a compound with an in- x-ray diffractometry, and was found to be single phase
sulating ferrimagnetic (FIM) ground state. Compared with a (00l) orientation. The lattice parameters (a =
0
2 to other ferromagnetic (FM) perovskite manganites, 5.55(2) ˚A, b = 5.36(2) ˚A, and c = 7.50(1) ˚A) were de-
1 GCMOexhibitsarelativelylowCurietemperature(TC), termined using (00l), (200), and (020) reflections from
: and its small structural tolerance factor t < 0.89[9, 10] a four-circle diffractometer/goniometer. No additional
v
leadstoarobustinsulatinggroundstate. Magneticprop- peaksduetosecondaryphasesordifferentcrystallineori-
i
X ertiesofthe systemreflectthoseofthe twosublatticesof entationswereobserved. Therockingcurvewidtharound
r Mn and Gd ions (see below). The different temperature the(004)peakofthefilmwas∼0.27◦. Thefilmthickness
a
dependence of magnetization of each of the two sublat- of45nmwasdeterminedbyalow-anglex-rayreflectivity
ticesresultsinachangeofsignofthetotalmagnetization measurement with an angular resolution of 0.005◦.
as a function of temperature at a characteristic compen- A Quantum Design SQUID magnetometer was used
sation temperature Tcomp, where the Mn and Gd sub- for measurements of the global magnetization with the
lattices have magnetic moments of the same magnitude magneticfieldorientedperpendiculartothe filmsurface.
and opposite direction.[9–11] A small tolerance factor, AlllocalizedMFMmeasurementsdescribedinthisLetter
a structural distortion, and the antiferromagnetic inter- were performed in a home-built low-temperature MFM
action between Gd and Mn sublattices yield remarkable apparatus.[15] MFM images were taken in a frequency-
properties, such as a giant magnetostrictive effect in a modulated mode, with the tip-lift height of 100 nm
above the sample surface. High resolutionSSS-QMFMR
cantilevers,[16] magnetized along the tip axis in a field
of 3 T, were used for MFM measurements; the external
∗ Electronicmail: [email protected] magnetic field was always applied perpendicular to the
2
0.001 T
80 0.01 T
0.1 T
3 ] 60 0.5 T
m 1 T (a) 4 K (b) 10 K (c) 15 K
c 1
u/ 40
m Hz)
[e 20 Δf (

M
0 0
(a)
0 (d) (e) T< Tcomp : Gd (f) T> Tcomp : Gd
: Mn : Mn
Tcomp n (a.u.)
[T] 0.2 Correlatio
c -80 non-magnetic Matrix non-magnetic Matrix
H
0
0.1 FIG. 2. (Color online) (a)-(c) MFM images acquired at dif-
ferent temperatures. Solid and dashed circles represent the
samesamplearea. (d)Cross-correlation mapbetweenimages
shown in panels (a) and (c). The large negative value at the
Gd Mn
0.0 (b) centerofthemapsignifiestheanticorrelationbetweenimages.
0 20 40 60 80 100 (e) and (f) Schematical illustration of the temperature evo-
T [K] lution of phase-separated magnetic regions above and below
Tcomp ≈ 12 K in 1 mT. The field of view in theimages ((a)-
FIG.1. (Coloronline)(a)Field-cooled M(T)curvesindiffer- (d)) is 6 µm × 6 µm. Features on the left side are broader
ent magnetic fields (H). (b) Coercive field (Hc) as a function than those on the right side because the scan plane is not
of temperature obtained from magnetic hysteresis loops. perfectly parallel to thesample surface.
film surface and parallel to the MFM tip. H > H the magnetization is reversed immediately be-
c
Fig.1(a)showsthe field-cooled(FC)magnetizationM low Tcomp, producing a characteristic sharp kink and a
V-shape in the data. This sharp reversal of the change
asafunctionoftemperatureatdifferentvaluesofapplied
magnetic field H. The temperature dependence of mag- in magnetization(from decreasingto increasingwith de-
netization was discussed previously by Snyder et al.[10] creasing temperature) is facilitated by a strong decrease
of H at T , as shown in Fig. 1(b), which is deter-
GCMO undergoes a phase transition from paramagnetic c comp
minedonthebasisofananalysisoffullhysteresiscurves
insulating to ferromagnetic insulating states, associated
with the ferromagnetic ordering of Mn cations, at T ≈ atdifferenttemperatures(datanotshown). Thepositive
C
offsetofM atthekinkat0.5Tand1TinFig.1(a)isdue
80 K. The local field due to FM order in the Mn sub-
lattice and the negative f-d exchange interaction on the to a paramagnetic background. All magnetic transition
temperaturesobservedin the filmareingoodagreement
Gd spins force the moments on the Gd sublattice to be
with the values previously reported for bulk polycrystal
anti-aligned to those in the Mn sublattice. The Mn sub-
and single crystal samples.[10, 13, 14] The data at 0.1
latticedominatesthemagnetizationathightemperature,
T has a clear kink as it crosses M = 0 and T , indi-
but the absolute magnitude of magnetization of the Gd comp
cating that some small number of the magnetic domains
sublattice growsfasterwhenthe temperatureis reduced.
Consequently,the totalmagnetizationM reachesa max- fliptheirorientationatTcomp. Thisisconsistentwiththe
bulkmeasurementsofH ≈0.1TatT ,andpointsto
imum value close to 50 K (see Fig. 1), starts to decrease c comp
coercivefieldinthissystembeingalocalproperty,proba-
with decreasing temperature, and goes toward zero at
T ≈15K inlow fields(T depends onH),where blydependentonthemagneticdomain’ssize,shape,and
comp comp
environment.
magnetizations of the Mn and Gd sublattices compen-
sate each other. Below T the local magnetization of TheMFMimagesdepictedinFigs.2(a)-(c)weretaken
comp
Gd is larger than that of Mn, |M |>|M |, and the sequentiallyat4K(T<T ),10K(T≈T ),and15
Gd Mn comp comp
sign of the total magnetization is determined by the di- K (T>T ), respectively, in a magnetic field of 1 mT
comp
rectionof magnetizationofthe Gd sublattice. When the (below H ) applied above T (field-cooled data). The
c C
appliedmagneticfieldH isbelowthecoercivefieldH of dashedcirclesin Figs.2(a)-(c) showthe same sample re-
c
thesystematT ,themagnetizationoftheGdsublat- gion(thermaldriftsarenegligibleforimagestakenbelow
comp
tice is locked in a direction opposite to the applied field, 15 K, see below). Regions of a non-zero magnetic sig-
andthetotalmagnetizationisnegativebelowT . For nal, either blue or red, change color as the temperature
comp
3
Figs.3(a)-(d)showMFMimagesobtainedat4Kafter
field-coolingtheGCMOsamplethroughT in0T,0.1T,
C
0.5T,and1T appliedfields. Inorderto understandthe
thermalandfieldevolutionofthesample’smagnetization
1 (a) 0 T (b) 0.1 T (c) 0.5 T
weevaluatedcross-correlationmapsfortheseimages. No
correlationwasobservedbetweendatasetsobtainedat0
Hz) T and 0.1 T (panels (a) and (b)), as shown in Fig. 3(e),
Δf ( and 0.1 T and 0.5 T (panels (b) and (c)), as shown in
Fig. 3(f). The lack of cross-correlation indicates signifi-
0 cant evolution of the spin magnetization due to the re-
220 (f) (e) (d) 1 T versal process inside FIM clusters in a field up to 0.5 T.
u.) On the other hand, magnetic domains imaged in 0.5 T
n (a. and1TFCexperimentsshowasimilarpattern,suggest-
o ing saturation of the magnetization reversal process as
elati well as a clear phase separation between ferrimagnetic
orr clusters and the paramagnetic matrix. (Data taken at
C
0 3 T, not shown, are similar to those at 1 T.) The lack
of correlation between the images (a)-(c) cannot be the
FIG. 3. (Color online) (a)-(d) Field-cooled MFM images result of thermal drift of the tip position over the sam-
taken at 4 K in different magnetic fields. (e) and (f) Cross-
ple,asthiswasobservedrepeatedlytobeunder1µmfor
correlation maps between (a)-(b), and (b)-(c), respectively;
our system (e.g., see panels (c) and (d)). The magnetic-
no correlation is observed. The field of view in the images
((a)-(f)) is 6 µm × 6 µm. Dashed circles correspond to the nonmagneticphasecoexistencecouldbeattributedtolo-
calizeddisorderor a localizedstraindistribution, similar
same sample area.
to observations in Y- and Pr-based manganites with a
comparably low tolerance factor [Y2/3Ca1/3MnO3 (t ∼
changes from 4 K to 15 K, but the green areas remain 0.884) and Pr2/3Ca1/3MnO3 (t ∼ 0.91)].[19–22] Results
of x-ray diffraction measurements on our thin-film sam-
green in all images (a)-(c) in Fig. 2. The sample, there-
ple,however,areclosetothoseonbulksamplesandtend
fore,isphaseseparatedintoFIM(blueandred)andnon-
to rule out a strain mechanism of phase separation.
magnetic(green)regions.[17]At4K(Fig.2(a))Gddom-
In conclusion, we have performed MFM experiments
inates the magnetization of FIM domains, as depicted
onaferrimagneticGCMOthinfilmanddirectlyobserved
schematicallyinFig.2(e). RedislandsinFig.2(a),there-
phase separation in the sample, with magnetic (FIM)
fore, represent those parts of the sample where Gd mag-
regions of characteristic dimensions between 0.1 to 0.5
netic moments point “down”, and the blue regions are
µmembeddedinanon-magneticmatrix. Thebehaviorof
thosewithGdmagneticmomentspointing“up”. At15K
magnetic regions is consistent with the presence of anti-
(Fig. 2(c)) allofthe red regionsswitchto blue, signaling
alignedMn andGd magneticsublattices, forminga FIM
areversalintheirmagnetization,asMnmagnetizationis
state. The observed magnetization reversal in the FIM
dominantaboveT ≈12K.Thissituationisdepicted
comp domains as a function of temperature, for small external
schematically in Fig. 2(f). The small magnetic contrast
magneticfield,isconsistentwiththeMnsublatticebeing
across the sample at 10 K (see Fig. 2(b)) indicates al-
dominant at T > T ≈ 15 K, but the Gd sublattice
mostequalmagneticcontributionsoftheanti-alignedGd comp
(with magnetization locked to be antiparallel to a small
and Mn sublattices in FIM regions near T . In addi-
comp applied field) is dominant for T < T . We attribute
tion, Fig. 2(b) demonstrates the domains’ breakup and comp
the phase separation to localized disorder rather than
a reduction in their size close to T (at 10 K). This
comp a strained state of the sample. These results will have
phenomenon is consistent with the exchange bias effect
significantbearing onthe potential utilizationofGCMO
previously observed in single crystals.[13] The reduction
andotherrelatedcompoundsinmagneticmemorydevice
ofthe size ofFIM domains closeto T alsoleads to a
comp applications.
decrease of the coercive field (see Fig. 1(b)).[14, 18]
Work at LANL (sample fabrication, SQUID measure-
Fig. 2(d) shows a cross-correlation map between im-
ments, MFM, data analysis, and manuscript prepara-
ages(a)and(c)andallowsustoinvestigatequalitatively
tion) was supported by the US Department of Energy,
the temperature evolution of magnetic domains in the
Office of Basic Energy Sciences, Division of Materials
sample. The large negative value in the center of the
Sciences and Engineering. Work at Brookhaven (data
cross-correlation map demonstrates the anti-correlation
analysis and manuscript preparation) was supported by
between 4 K and 15 K magnetization data in Fig. 1(a),
the US Department of Energy under Contract No. DE-
indicating thatredandblue islands reversetheirmagne-
AC02-98CH10886. N.H. is a member of CONICET (Ar-
tization (and colors) upon the temperature change from
gentina).
4Kto15K.Thecentrallocationofthecross-correlation
minimum also demonstrates the small thermal drift in
our MFM apparatus.
4
[1] S.Jin, T. H. Tiefel, M McCormack, R. A. Fastnacht,R. [12] V.F.Correa, N.Haberkorn,G.Nieva,D.J.Garcia,and
Ramesh, and L. H.Chen, Science 264, 413 (1994). B. Alascio, arXiv:1109.0259v1.
[2] Patrick A. Lee, Naoto Nagaosa, and Xiao-Gang Wen, [13] N. Haberkorn,S.Larr´egola, D. Franco, and G. Nieva,J.
Rev.Mod. Phys. 77, 721 (2005). Magn. Magn. Mater. 321, 1133 (2009).
[3] Weida Wu,Casey Israel, Namjung Hur, Soonyong Park, [14] Yanwei Ma, M. Guilloux-Viry, P. Barahona, O. Pen˜a,
Sang-Wook Cheong, and Alex De Lozanne, Nat. Mater. and C. Moure, Appl.Phys.Lett. 86, 062506 (2005).
5, 881 (2006). [15] E. Nazaretski, K. S. Graham, J. D. Thompson, J.
[4] J. P. Zhou, J. T. McDevitt, J. S. Zhou, H.Q. Yin, J. B. A. Wright, D. V. Pelekhov, P. C. Hammel, and R.
Goodenough, Y. Gim, and Q. X. Jia, Appl. Phys. Lett. Movshovich, Rev.Sci. Instrum.80, 083074 (2009).
75 1146 (1999). [16] A SSS-QMFMRcantilever, Nanosensors, Inc.
[5] LevP.Gor’kovandVladimirZ.Kresinc,Phys.Rep.400 [17] Y. H. Sun, Y. G. Zhao, H. F. Tian, C. M. Xiong, B. T.
149 (2004). Xie,M.H.Zhu,S.Park,WeidaWu,J.Q.Li,andQiLi,
[6] E. Dagotto, T. Hotta and A. Moreo, Phys. Rep. 344, 1 Phys. Rev.B. 78, 024412 (2008).
(2001), and references therein. [18] B. D. Cullity and C. D. Graham, Introduction To Mag-
[7] Liuwan Zhang, Casey Israel, Amlan Biswas, R. L. neticMaterials,TheInstituteofElectricalandElectronic
Greene, and Alex deLozanne, Science 298, 805 (2002). Engineers, Inc. (2009).
[8] John B. Goodenough, J. Appl.Phys. 81, 5330 (1997). [19] R. Mathieu, P. Nordblad, D. N. H. Nam, N. X. Phuc,
[9] H.Y.Hwang,S-W.Cheong,P.G.Radaelli, M.Marezio, and N.V. Khiem, Phys.Rev.B 63, 174405 (2001).
and B. Batlogg, Phys. Rev.Lett. 75, 914 (1995). [20] P. G. Radaelli, R. M. Ibberson, D. N. Argyriou, H.
[10] G. Jeffrey Snyder,C. H. Booth, F. Bridges, Ron Hiskes, Casalta, K. H. Andersen, S. W. Cheong, and J. F.
SteveDiCarolis,M.R.Beasley,andT.H.Geballe,Phys. Mitchell, Phys. Rev.B 63, 172419 (2001).
Rev.B. 55, 6453 (1997). [21] V. N. Smolyaninova, A. Biswas, P. Fournier, S. Lofland,
[11] Octavio Pen˜a, Mona Bahouta, Karim Ghanimia, Pe- X. Zhang, G. Zhao, and R.L. Greene, Phys.Rev.B 65,
dro Duranb, Dionisio Gutierrezb, and Carlos Moureb, 104419 (2002).
J. Mater. Chem. 12, 2480 (2002). [22] D. Saurel, A.Bruˆlet, A. Heinemann, C. Martin, S. Mer-
cone, and C. Simon, Phys. Rev.B 73, 094438 (2006).