Antiferromagnetism at the YBa2Cu3O7 / La2/3Ca1/3MnO3 interface

Abstract

The magnetic properties of a series of YBa2Cu3O7-x/La2/3Ca1/3MnO3 (YBCO/LC1/3MO) superlattices grown by dc sputtering at high oxygen pressures (3.5 mbar) show the expected ferromagnetic behaviour. However, field cooled hysteresis loops at low temperature show the unexpected existence of exchange bias, effect associated with the existence of ferromagnetic/antiferromagnetic (F/AF) interfaces. The blocking temperature (TB) is found thickness dependent and the exchange bias field (HEB) is found inversely proportional to the FM layer thickness, as expected. The presence of an AF material is probably associated to interface disorder and Mn valence shift towards Mn4+. Comment: 12 pages, 2 figures, 1 table, submitted to Applied Physics Letters

Full text

Page 1

Submitted to Applied Physics Letters 09/22/03
Antiferromagnetism at the YBa2Cu3O7 / La2/3Ca1/3MnO3
interface

N. Haberkorn
Universidad Nacional del Sur, Avda. Alem 1253, Bahía Blanca, 8000 Bs. As.,
Argentina.,
Comisión Nacional de Energía Atómica, Centro Atómico Bariloche, S. C. de Bariloche,
8400 R. N., Argentina

J. Guimpel, M. Sirena, and L. B. Steren
Comisión Nacional de Energía Atómica, Centro Atómico Bariloche, S. C. de Bariloche,
8400 R. N., Argentina. and
Instituto Balseiro, Universidad Nacional de Cuyo and Comisión Nacional de Energía
Atómica, S. C. de Bariloche, 8400 R. N., Argentina

W. Saldarriaga, E. Baca and M. E. Gómez
Departamento de Física, Universidad del Valle A. A.,25360 Cali, Colombia.

The magnetic properties of a series of YBa2Cu3O7-x/La2/3Ca1/3MnO3 (YBCO/LC1/3MO) superlattices grown by dc
sputtering at high oxygen pressures (3.5 mbar) show the expected ferromagnetic behaviour. However, field cooled
hysteresis loops at low temperature show the unexpected existence of exchange bias, effect associated with the
existence of ferromagnetic/antiferromagnetic (F/AF) interfaces. The blocking temperature (TB) is found thickness
dependent and the exchange bias field (HEB) is found inversely proportional to the FM layer thickness, as expected. The
presence of an AF material is probably associated to interface disorder and Mn valence shift towards Mn4+.
75.70.Cn, 75.30.Vn, 74.76.Bz

Page 2

The interface between La1-xAxMnO3 manganites (LAxMO with A: Sr, Ca; 0.2 < x < 0.5)
and RBa2Cu3O7-x (RBCO with R: Y, Gd) superconductors affects the physical properties
of the materials.1,2 Saturation magnetization (MS), Curie (TC) and superconducting
transition (TS) temperatures are reduced from bulk values for YBCO/LC1/3MO
superlattices, effect attributed to disorder and oxygen deficiency at the interface.1-3 In
previous work,4 we studied the structure of YBCO / LC1/3MO superlattices grown on
(100) MgO through refinement of the X-ray diffraction patterns. Although the
superlattices showed epitaxial growth and lattice parameters similar to those found in
films, the interfaces were found not perfect, showing around 1.2 nm roughnesses and
30 % interdiffusion.
In order to study the magnetic structure of the interface, we analysed the magnetic
behaviour of the same series of YBCO/LC1/3MO superlattices at low temperatures. In the
following they are labelled as YBCON / LC1/3MOM, where N indicates the YBCO layer
thickness in unit cells (u.c.) (N = 3, 5, 10), and M indicates the LC1/3MO layer thickness
in u.c. (M = 9, 15, 30). These values indicate the nominal thickness of the layers, which
due to the existence of roughness could fluctuate around N and M in the above specified
amount.4 The total thickness of the superlattices was kept constant at approximately
150 nm by varying the number of bilayers. The superlattices were grown by DC
magnetron sputtering in pure oxygen at high pressures (3.5 mbar), as previously
reported.1Magnetic properties were measured in a commercial SQUID magnetometer,
with the applied magnetic field parallel to the superlattice surface. The TC was estimated
from the extrapolation to zero magnetization of the zero field cooled (ZFC) and field
cooled (FC) magnetization vs T curves. The MS was measured with isothermal hysteresis

Page 3

loops at 35 K. The FC hysteresis loops were acquired after cooling in 1T (HFC). Transport
measurements detect superconductivity below 40 K for YBCO5 / LC1/3MO9, below 75 K
for YBCO10 / LC1/3MO9, and below 60 K for YBCO10 / LC1/3MO30.
Figure 1 shows the ZFC and FC hysteresis loops at 5 K for a YBCO3 / LC1/3MO15
superlattice. The exchange bias signature, i.e. a field shift in the FC loop, is evident and
was observed for all studied superlattices. This phenomenon is associated to the existence
of a ferromagnetic/antiferromagnetic (F/AF) interface, and is observed when the F layer
is cooled through the AF Néel temperature (TN) in the presence of a magnetic field.5 The
unexpected presence of an AF phase can also be inferred from the data in Table 1, which
shows TC and MS. Superlattices YBCO5 / LC1/3MO9 and YBCO10 / LC1/3MO9 are not
included since there is a huge uncertainty in TC and MS, given the smallness of the
magnetic signal. The MS values are strongly reduced from the bulk value
(≈580 emu / cm3) for all superlattices, by a factor between 5 and 10, as previously found
for the same system grown on (100) SrTiO3.1 If we interpret this fact as an amount of FM
material transformed into an AF phase, we can define an effective F layer thickness (tF)
normalizing Ms to the bulk value. This effective thickness is only 1 u.c. for
YBCO3 / L1/3CMO15 and YBCO5 / LC1/3MO 15, 5 u.c. for YBCO3 / L1/3CMO 30, and 8 u.c.
for YBCO10 / LC1/3MO 30. These very low tF values for the first two superlattices
probably imply that the F layers are not continuous in these samples. This possibility is
also suggested by the thickness dependence of Hc, which reduces with layer thickness,
similar to the behaviour in small size particles,6 and contrary to the Hc ∝ t-1 one expected
for continuous films.7

Page 4

The TC are also reduced from bulk values for all superlattices, being smaller the
thinner the LC1/3MO layers. A reduced TC for small thickness was also found for
LSr0.4MO films,8 where it was attributed to dimensional effects, structural defects and
biaxial strain, and for the YBCO / LC0.67MO system,3 where it was discussed in terms of
stress and proximity effect. The TC may be influenced simultaneously by stress, valence
shift and dimensional effects.
In view of these results we picture the AF phase as mainly distorted manganite
due to manganite / cuprate interface disorder. Its origin can be rationalized in several
ways. First, it is known that interdiffusion is present. This could originate a new
intergrown structured material with an, a priori, unpredictable magnetic behaviour.4
Second, a shift in the Mn3+/Mn4+ relation towards Mn4+, and so towards the AF
manganite phase, has been found at the YBCO /LSr0.3MO interface,9 which was related
to Ba and La interdiffusion, and which could and probably is present also in the YBCO
/LC1/3MO system. Third, a Mn valence shift towards Mn4+ can be caused by a change in
the crystalline environment, originated either by a change in the oxygen content of the
material or by stress at the interfaces. The latter is known to influence the magnetic order
in superlattices.10
When analysing the behavior of MS with YBCO layer thickness, we do not find a
clear trend. Stadler et al.9 found a larger displacement towards Mn4+ in the Mn3+ / Mn4+
relation with increasing YBCO thickness in YBCO/LSr0.3MO bilayers, which should
result in smaller MS values. In our results, between YBCO3 / LC1/3MO15 and YBCO5 /
LC1/3MO15, MS is smaller for thicker YBCO layers, but between YBCO3 / LC1/3MO30 and
YBCO10 / LC1/3MO30, MS is larger for thicker YBCO layers.

Page 5

Figure 2 shows, for YBCO3 /LC1/3MO15 and YBCO3 /LC1/3MO30, the T
dependence of the exchange bias field (HEB), HEB = H1+H2/2, ZFC hysteresis loop
coercive field (HCZ) and FC hysteresis loop coercive field (HCF), HC = H1-H2/2, where
H1 and H2 are the fields for zero magnetization at both branches of the hysteresis loops,
measured after saturating the sample with 1T or -1T. For all superlattices, HEB decreases
with increasing temperature, reaching zero values at the blocking temperature (TB). We
find this temperature independent of YBCO layer thickness and superconducting
behavior, and approximately TB ≈ 30 K for both YBCON / L1/3CMO 15 superlattices, and
TB ≈ 22 K for both YBCO N / L1/3CMO30 ones. In the F/AF LCxMON/ LCyMON system, it
has been reported TB ≈ 70 K,11,12 independently of x and y doping values or layer
thickness. On the other hand, for LC0.5MO / LC1/3MO bilayers H. B. Peng et al,13 showed
that TB and the T dependence of HEB are sensitive to interface roughness. The different
TB for different LC1/3MO layer thicknesses again suggests that the AF configuration is
different, changing from a continuous layer to a granular system.
For an AF/F/AF system, the exchange energy per unit interface area is given by
δE = (MS.tF.HEB/2). The HEB values at 5 K, present indeed an inverse thickness, (tF)-1,
dependence for both LC1/3MO thicknesses, with tF the effective F layer thickness.5 The
estimated δE are independent of the YBCO layer thickness with values ≈ 0.004 erg/cm2
for theYBCON/LC1/3MO15 superlattices, and ≈ 0.006 erg/cm2 for the YBCON/LC1/3MO30
ones.
The HEB and HCF vs. T curves in Fig. 2 show an exponential dependence instead
of the usual power law.14 This behaviour was previously found in LCxMO/LCyMO
superlattices,12 and attributed to frustration effects resulting from competing F and AF