Document text
Inhomogeneous Magnetism in La-doped CaMnOs. (II) Mesoscopic Phase Separation
due to Lattice-coupled FM Interactions
CD. Ling, 1 ' 2 !] E. Granado, 3 ' 4 ' 5 J.J. Neumeier, 6 J.W. Lynn, 3 - 4 and D.N. Argyriou 2 - 7
1 Institut Laue-Langevin, BP 156, 38042 Grenoble Cedex 9, France
2 Materials Science Division, Argonne National Laboratory, Argonne, Illinois 60439
3 NIST Center for Neutron Research, National Institute of Standards and Technology, Gaithersburg, Maryland 20899
Center for Superconductivity Research, University of Maryland, College Park, Maryland 20742
5 Laboratorio Nacional de Luz Sincrotron, Caixa Postal 6192, CEP 13083-910, Campinas, SP, Brazil
6 Department of Physics, Montana State University, Bozeman, Montana 59717
7 Hahn-Meitner- Institut, Glienicker str. 100, Berlin D- 14 109, Germany
(Dated: February 2, 2008)
A detailed investigation of mesoscopic magnetic and crystallographic phase separation in
Cai-^La^MnOa, 0.00 < x < 0.20, is reported. Neutron powder diffraction and DC-magnetization
techniques have been used to isolate the different roles played by electrons doped into the e 9 level
as a function of their concentration x. The presence of multiple low-temperature magnetic and
crystallographic phases within individual polycrystalline samples is argued to be an intrinsic feature
of the system that follows from the shifting balance between competing FM and AFM interactions
as a function of temperature. FM double-exchange interactions associated with doped e g electrons
are favored over competing AFM interactions at higher temperatures, and couple more strongly
with the lattice via orbital polarization. These FM interactions thereby play a privileged role, even
at low e g electron concentrations, by virtue of structural modifications induced above the AFM
transition temperatures.
PACS numbers: 61.12.Ld; 75.25.+z; 75.30.Kz; 75.70.Kw; 75.70.Pa
I. INTRODUCTION
The physical properties of mixed valent perovskite
manganites such as Cai_ a; La :E Mn03 are dominated by
the strong coupling of charge-orbital and spin degrees
of freedom. This results in a family of materials that
shows pronounced physical property responses to chemi-
cal doping, temperature, pressure and magnetic field. Of
these responses, colossal magnetoresistance (CMR) - a
dramatic drop in resistivity in an applied magnetic field
- at optimal doping x ~ 0.7 has attracted the most at-
tention as it raises the possibility of applications such as
data storage devices and sensors.
At the 'electron-doped' end of the phase diagram (x ~
0) , the light doping of charges into the well understood G-
type antiferromagnetic [G-AFM, Fig^a)] ground state
provides an opportunity to test the relevance of physi-
cal models of manganites. It was originally argued by de
Gennesi that a small concentration of doped carriers into
the e g band (which is fully polarized due to the strong
Hund's coupling of localized t^g electrons) gives rise to
ferromagnetic (FM) double-exchange (DE) interactions,
which for lightly doped systems competes with AFM
super-exchange (SE) to produce a spin canted G-AFM
state. 2 This is consistent with a coexistence of FM and
G-AFM components observed by neutron powder diffrac-
tion (NPD). 3 - 4 -^ 7 . The application of simple DE across
the whole of the phase diagram, however, contradicts
experimental evidence by predicting either homogenous
canting or the pure FM state at all points, in contrast
with the rich phase diagram experimentally observed^
In fact, the (ostensibly) spin-canted state only survives
to x ~ 0.1, beyond which the degeneracy of the (d x 2_ y 2
and d 3z 2_ r 2) e g orbitals causes it to be supplanted by
C-type AFM [C-AFM, Fig []]»] In C-AFM, FM
DE becomes long-range in one dimension via delocal-
ized d 3z 2_ r 2 orbital chains, into which the doped e g elec-
trons are stabilized, while AFM SE is maintained per-
pendicular to these chains. This leads to a co-operative
Jahn- Teller (J-T) distortion along the FM chain direc-
tion, lowering the symmetry from orthorhombic Pnma
to monoclinic P2i/m.
These two papers report a detailed investigation into
the nature of, and the relationships among, the rich va-
riety of phases found in the electron-doped regime of
Cai-ajLa^MnOs. This is of interest both as a model for
spin-lattice coupling in the dilute limit of lattice polarons,
and due to reports in many systems Cai- x A x Mn03
of large magnetoresistance e&ectsi&H^lS^ and meta-
magnetic phase transitions^i^ In Part Iji& we found
using neutron scattering that for light electron-doping
0.0 < x < 0.1, the G-AFM matrix contains a well-
organized liquid distribution of FM clusters ~ 10 A in
diameter. These clusters could be aligned by an ex-
ternal applied magnetic field to produce a long-range
FM moment, as seen at the opposite end of the same
phase diagram>i2iiSiiS Higher density of these clusters
at x = 0.09 led to a spontaneous (H — 0) long-range
FM moment due to the formation of a FM cluster-
glass^ the orientation of which is coupled to the G-
AFM matrix (Fig. Part II concerns principally
the relationships among the various crystallographic (or-
thorhombic and monoclinic) and magnetic (G-AFM,
liquid-like FM clusters, FM cluster glass and C-AFM)
2
(a) G-AFM
FIG. 1: Schematic representations of the low-T magnetic
ground states of Cai-xLa^MnOa (x < 0.2). Solid lines show
the unit cell and dashed lines show nearest-neighbor Mn-Mn
interactions, (a) Ideal G-AFM. (6) C-AFM, in which e 9 elec-
trons delocalize into d 3z 2_ y 2 orbital chains along the (101)
direction, allowing 1-D FM DE while maintaining AFM in-
teractions among the chains and causing a symmetry-lowering
from orthorhombic Pnma to monoclinic P2i/m.
0.00 <x<~ 0.05 0.05 <~ x<~ 0.10
FIG. 2: Schematic representation of the growth of FM clusters
within the G-AFM matrix (left), which can be aligned by an
external applied magnetic field, into a FM cluster glass (right)
exhibiting spontaneous long-range FM, with increasing x. 16
Black arrows represent the G-AFM matrix and grey spheres
represent FM clusters.
phases. High-resolution NPD and DC-magnetization
techniques are used to address questions of sample ho-
mogeneity arising out of the observation of multiple crys-
tallographic and magnetic phases in individual polycrys-
tallinc samples^iii^ necessary in order to correctly in-
terpret local phenomenon observed using bulk probes. It
is found that the inability to attain a unique thermody-
namic ground state is an intrinsic feature of the system
resulting from the extremely fine balance between com-
peting states.
II. EXPERIMENTAL DETAILS
Ceramic samples of Cai-^La^MnOa with nominal
compositions 0.00 < x < 0.20 were prepared by solid
state reaction. Stoichiometric quantities of (99.99% pu-
rity or better) CaC03, La203, and Mn02 were weighed
(to yield 7 g samples) and mixed in an agate mortar for
15 min followed by reaction for 20 h at 1100 °C. The
specimens were reground for 10 min, reacted for 20 h
at 1150 °C, reground for 10 min, reacted for 20 h at
1250 °C, reground for 10 min, reacted for 46 h at 1300 °C,
reground for 10 min, reacted for 46 h at 1300 °C, re-
ground for 10 min, pressed into pellets, reacted for 17 h
at 1300 °C and cooled at 0.4 "Cmm" 1 to 30 °C.
DC-magnetization measurements were conducted us-
ing a commercially available SQUID magnetometer.
Specimens were cooled to 5 K in zero field, then warmed
to the highest measurement temperature in an applied
field of H = 2000 Oe. Magnetization vs. H curves were
taken at 5 K.
Temperature-dependent time-of-flight (TOF) NPD
data were collected on the Special Environment Pow-
der Diffractometer (SEPD) at Argonne National Labora-
tory's Intense Pulsed Neutron Source (IPNS). Data were
analyzed by Rietveld-refinement using the program suite
FullProf. 1-2 % wt Marokite (CaMn 2 4 )2ii2£ impurities
were included in all refinements as both nuclear and (at
low temperatures) magnetic phases (details of the low-T
AFM structure of Marokite are published elsewhere) m&
III. RESULTS AND ANALYSIS
A. DC-magnetization and Resistivity
The T-dependencies of the DC-magnetization of all
studied samples for H = 2000 Oe are shown in Fig. [3]
For x — 0.20, a peak is observed at T ~ 180 K. A similar
feature observed for a Cao^Bio.isMnOa single-crystal 2 ^
was ascribed to a change of character of the spin fluc-
tuations from FM to AFM with decreasing T, due to
the freezing of the charge carriers and the consequent
suppression of DE interactions. At lower T, a sudden
enhancement in the DC-magnetization is observed be-
low T c ~ 110 - 125 K for x < 0.12, and is ascribed to
a spin-ordering transition with a FM component below
Tc- The inset to Fig. [3] shows DC-magnetization at 5
K as a function of H for x — 0.00, 0.03 and 0.06, with
clear signatures of FM components (hysteresis). The IE-
dependence of the DC-magnetization at 5 K for H — 2000
Oe is shown in Fig. results for the large samples used
in this study (open markers) are consistent with those for
smaller samples previously studied (closed markers). 25
3
0.03
Neutron Powder Diffraction
■■ 0.00
H = 2000 Oe
0.02
0.01
x=0.20 T(G) T N (C)
-15-10 -5 5 10 15
H(kOe)
50 100 150 200 250 300
T(K)
FIG. 3: (a) Magnetization M vs. T measured at H = 2000
Oe for x = 0.00 and x = 0.20 samples. T N (C) and T N (G) are
indicated, (b) A similar plot (note the change in scale) for
x = 0.03, 0.09 and 0.12 samples, with an inset showing M vs.
H for x = 0.00, 0.03 and 0.06. For the M vs. H loops, lines
were drawn through the low and high field regions of the data,
and the intersection taken as the saturation moment M 3a t at
5 K used in Fig. 0]
0.5
0.4
0.3
0.2
0.1
0.0
T = 5K :
-
■
/ D \
\ D
G-AFM /
C-AFM -\G-AFM
C-AFM :
: (i) (ii)/
(ink
(iv) :
0.00
0.05
0.10
x in Ca La MnO
1-x x 3
0.15
0.20
FIG. 4: Magnetic saturation moments M sa t determined from
M vs. H curves at 5 K for the samples used in the present
study (open squares) and in the study by Neumeier and
Colin— (solid squares) as a function of x. Regions (I-IV)
discussed by Neumeier and Cohn are labelled and defined by
dashed lines.
At 300 K, the crystal structures of samples at x = 0.00,
0.03, 0.06, 0.09, 0.12, 0.16 and 0.20 were Rietveld-refined
as single orthorhombic Prima phases with TOF NPD
data. In preliminary refinements, the O fractional oc-
cupancies [^(O)] were refined and found to lie between
0.98 and 1.02 for all samples, with typical standard devi-
ations ±0.008. This result supports the absence of signifi-
cant cation or oxygen vacancies, consistent with chemical
analysis performed on similarly prepared samples^ and
f(O) was subsequently fixed at 1.
Figs. HRd,g) show the crystallographic phase fraction
of the symmetry-lowered (P2i/m) phase associated with
the C-AFM state, transformed from the orthorhombic
Pnma state, for x = 0.12 and 0.20. This phase transition
arises due to the polarization of d 3z 2_ y 2 orbitals along the
(101) direction, facilitating DE along the FM chains char-
acteristic of C-AFM. A monoclinic phase fraction with a
similar T-dependence could also be refined for x = 0.09
and 0.16. (For x = 0.06, although the presence of a weak
C-AFM magnetic Bragg peak indicates the presence of
a small monoclinic phase fraction, this fraction was too
small to meaningfully refine.) The transition from the
room-temperature orthorhombic Pnma phase in the C-
AFM regime is also shown in the plots of refined lattice
parameters vs. T in Figs. EJc,/). Note that for x = 0.20
(Fig. Et/), in addition to the monoclinic distortion un-
dergone by the majority of the sample, the remaining
orthorhombic phase fraction undergoes a different low-
T distortion, characterized by an elongation along (100)
and (001). This type of distortion has been observed
for Ca 2/3 La 1/3 Mn03S2£ and G&x-xBiJAnOz (x = 0.22,
0. 25)^ at low T, and has been ascribed to superstructures
in the ac plane caused by charge and orbital ordering of
the Mn 3+ e g electrons (a 'Wigner crystal'-type or W-C-
type phase). In order to account for the enlarged unit cell
of W-C-type without excessively complicating the refine-
ment, the Oil site was split evenly across the 4/ position.
(Although the distortion clearly identifies this phase, no
corresponding superstructure Bragg peaks were identified
for our x = 0.20 sample below the orbital-ordering tem-
perature of the W-C phase T (W-C)~ 165 K, possibly
due to the small phase fraction and/or disorder.)
Final refined crystallographic phase fractions, unit
cells, atomic positions, displacement parameters, Mn-
O bond distances and magnetic moments at 20 K for
x = 0.03, 0.12 and 0.20 are given in Tabled The differ-
ences between the Mn-0 bond distances are very small
for the orthorhombic G-AFM phase at x = 0.03 and 0.12,
1. e. the MnC>6 octahedra are not significantly distorted.
The same was true of the orthorhombic phases at x =
0. 06 and 0.09 at 20 K (not shown). Conversely, for the
monoclinic C-AFM phase at x = 0.20 and 0.12, as well
the monoclinic phases at x = 0.16, 0.09, and 0.06 (not
shown), the Mnl-012 and Mn2-012 bonds are longer
than the other Mn-0 bond distances by ~ 0.01 — 0.05 A,
1. e. the MnC"6 octahedra are elongated along the (101)
4
FIG. 5: (a,e,/) T-dependence of pseudo-cubic lattice parameters (101), (101) and (010) for (a) x — 0.03, (c) x = 0.12 and (/)
x = 0.20, from Rietveld- refinement of NPD data. Note in (c) and (/) the elongation of the monoclinic unit cell along the (101)
direction of e g electron polarization at Tjv(C-AFM); and in (/) the elongation in the remaining orthorhombic phase of (100)
and (001) due to the formation of a (twinned) Wigner-crystal type phase, in contrast to the isotropic monotonic contraction
of the G-AFM orthorhombic cell in (a). (d,g) T-dependence of the phase fraction of symmetry-lowered monoclinic (P2i/m)
phase associated with the C-AFM magnetic phase, from the same Rietveld-refinement of NPD data. (b,e,h) T-dependence of
the integrated intensities of the characteristic G-AFM (110)/(011) (open circles) and C-AFM (3I3) (filled circles) magnetic
peaks from NPD data, normalized to the intensity of the strong nuclear (220)/(022) peak, for (b) x — 0.03, (e) x = 0.12 and
(h) x = 0.20.
direction. This follows from the co-operative Jafm- Teller
ordering along (10T) that allows FM DE in the C-AFM
state, and is consistent with previous diffraction stud-
ies for electron-doped CaMn03i^S£ Finally, for the or-
thorhombic W-C-type phase at x = 0.20, bond distances
from Mnl to the split Oil site illustrate the alternately
shortened and elongated bonds along (101) and (101)
characterizing this structure type and causing the elon-
gation of a and c relative to bj\[2 [see FigEJ/)].
The magnetic phases of the samples were also identified
and Rietveld- refined by NPD, magnetic Bragg peaks be-
ing observed at low-T for all samples. Observed magnetic
reflections were consistent with G-AFM for x = 0.00 and
0.03, and with C-AFM for x = 0.20 and x = 0.16. For
intermediate dopings x — 0.06, 0.09, and 0.12, G- and
C-AFM Bragg reflections were observed simultaneously
at low-T (an extremely weak G-AFM peak was also ob-
served for x = 0.16). Fig. shows a portion of the
TOF NPD pattern at high d-spacing (low-Q) at 20 K for
x = 0.12, illustrating the coexistence of Bragg peaks from
distinct C- and G-AFM structures (dots); the solid lines
in the upper set correspond to calculated and difference
profiles using a magnetic model with phase-separated G-
and C-AFM structures. Deficiencies in the fit shown in
the upper set of Fig. EJare accounted for when a FM sub-
lattice with spins perpendicular to those of the G-AFM
lattice is included in the magnetic model (lower set) . Sig-
nificant FM intensities were also observed for x = 0.06
and 0.09 (not shown). This is consistent with our DC-
magnetization measurements, where the strongest FM
signal was observed for x between 0.06 and 0.12 (see Fig.
2J. No evidence was found for a FM moment in the mon-
oclinic phase, as expected, all e g electrons participating
in FM DE along the (10T) chain directions of C-AFM
rather than forming FM clusters.
Figs. \^b,e,h) show the T-dependencies of character-
istic Bragg reflections associated with G- and C-AFM
spin structures for x = 0.03, 0.12, and 0.20 respectively.
The magnetic ordering temperatures for G-AFM, T)v(G),
correspond to the FM Tq observed by DC-magnetization
5
TABLE I: Results from Rietveld-refinements of TOF NPD data collected at 20 K for Cai-^La^MnOs samples at selected values
of x. Refinements were carried out in space groups Pnma (#62) (atomic positions: Oil in 8d, 021 in 4c) and P2i/m (#12)
(atomic positions :_011 in 4/, 012 in 4/, 021 in 2e, 022 in 2e). G-AFM moments refined along (001), FM along (010) and
C-AFM along (101). Superscript letters indicate constraints.
X
0.03
0.12
0.20
Pli3.se fraction
1
0.516(14)
0.484(14)
0.807(3)
0.193(3)
Space group
Prima
Pnma
P2i/m
P2i/m
Pnma
G-AFM 2.47(3)
G-AFM 2.45(8)
+ FM 0.9(2)
C-AFM 2.29(7)
C-AFM 2.89(5)
n (k\
a (A)
o./i y^±u iy )
K 'iAAQ^f 1 7 s !
o . oy J-O )
b (A)
1 .4I4U^y ( )
7 A 1 { CZ\
i .4 / 1 (O )
7 /I c^Q7f 7"\
/ .400 ( \( )
7 1 7f 9*1
; .401 < \ Z)
7 A t;Qt;/ A
/ .4000(0 )
c (A)
P ( )
o.2oy f o(o ;
C 1777/ "3 \
0.^ ( r ' (,3)
0.3218(o;
O.o34t o(18J
0.30ol(z )
y0.o4o7(loj
yi.3ioy(iyj
LaCal x
0.0327(3)
0.0323(14)
0.030(3)
0.019(2)
0.0360(16)
LaCal z
U.yyoD^o )
0.99o4(19J
o.y»y(3j
0.004(2 )
0.9877(16 )
j_jav_/az x
v .ozoyz )
n ^ 1 o r ^ S
^nQ(9\
Oil x
0.2860(2)
0.2879(9)
0.2799(17)
0.2799(11)
0.288(12) /
0.271(5)
Oil a
0.03433(19)
0.0314(5)
0.0378(15)
0.0402(10)
0.022(6)/
0.042(6)
Oil z
0.7128(3)
0.7116(8)
0.7162(17)
0.7214(12)
0.743(6)/
0.699(6)
012 x
-
-
0.7807(18)
0.7839(6)"
-
012 y
—
—
0.0355(12)
0.0294(8)
—
012 z
-
-
0.7796(17)
0.7839(6)"
-
021 a
0.4899(4)
0.4866(12)
0.495(3)
0.4864(19)
0.494(4)
021 2
0.0678(5)
0.0678(13)
0.065(2)
0.0639(18)
0.053(4)
022 a;
—
—
0.998(2)
0.9951(18)
—
022 z
—
0.447(2)
0.438(2)
Mn B i50
0.20(3)
0.22(7) d
0.20(6) b
LaCa
0.39(2)
" '
0.35(7) e
0.36(6) c
Oil B iso
0.29(9)
0.19(7)'
0.19(7)'
0.50(8)
0.3(2)
012 B iso
—
-
0.19(7)'
0.29(7)
—
021/22 B iBO
0.45(4)
0.19(7)'
0.19(7)'
0.31(8)
0.4(2)
Mnl-Oll (A)
1.8970(13)
1.904(5)
1.888(9)
1.908(6)
2.03(2) /
-II ( 101 )
1.83(3)
Mnl-012 (A)
1.9034(13)
1.906(4)
1.921(9)
1.938(3)
-II (101)
Mnl-022 (A)
1.8947(5)
1.9035(13)
1.885(2)
1.895(2)
1.886(3)
-II (oio)
Mn2-OH (A)
1.916(9)
1.898(6)
-II (101)
Mn2-012 (A)
1.928(9)
1.939(3)
-II (101)
Mn2-021 (A)
1.896(2)
1.8980(18)
-II (oio)
Rb
0.0672
0.0612
0.0833
wR b
0.0700
0.0574
0.0843
x 2
1.69
1.88
3.16
measurements (see Fig. 121 within experimental error. In
contrast, the C-AFM order parameter for x = 0.12, be-
sides showing a rather peculiar T-dependence, is far more
evident in the NPD than in the magnetization data. Note
nonetheless that for x = 0.20, T N {C) ~ 180 K obtained
by NPD is clearly associated with the peak in the DC-
magnetization (compare Figs. Ef 1 ) and Eta))-
IV. DISCUSSION
A. Sample Homogeneity
The relationships among crystal lattices, crystallo-
graphic phase fractions and magnetic order parameters
as functions of T presented in Fig. El reveal relation-
ships that bear on the compositional homogeneity of the
polycrystalline samples used in this study. In particu-
lar, for 0.12 and 0.20 [Figs. EKflO], the growth of the
monoclinic phase fraction does not continue down to the
lowest temperatures. For x — 0.12 the phase fraction
does not change significantly below ~ T/v (G-AFM), and
for a; = 0.20 it does not change significantly below Tb(W-
C). Furthermore, for x = 0.12, T/v (G-AFM) (marked by
the dashed line) appears to influence both the C-AFM
magnetization [Fig. |3Je)] and the monoclinic lattice pa-
rameters [Fig. EJc)]. These relationships imply that the
different magnetic states are competing for the same do-
mains within the sample, rather than simply forming in
mutually exclusive, compositionally segregated, domains.
The wide x-interval over which Pnma and P2i/m
crystallographic phases coexist, which is typical
of doped manganites e.g. Cai-^Bi^MnOs 4 and
Cai-^Sm^MnOs, 13 cannot be understood in terms
6
x=0.12
T = 20 K
8000
■f 6000
-S. 4000
I 2000
No FM in model 2^
j y G-AFM C-AFM
(200)/(002) (020) _
(121)/(121) (1 01 01 )
Nuclear+FM Nuclear+FM
I I
d-spacing (A)
FIG. 6: Observed (+), calculated and difference (below)
plots of Rietveld-refined time-of-flight NPD data (60 ° detec-
tor bank of SEPD) for the x — 0.12 sample at 20 K. Prominent
magnetic reflections are labelled. The upper set shows the
fit using a G-AFM + C-AFM magnetic model only, and the
bottom set shows the fit with an additional FM component
perpendicular to the G-AFM moment. Magnetic reflections
due to the AFM Marokite impurity phase are marked (*).
of mesoscopic inhomogeneities in composition x within
polycrystalline samples; rather, mesoscopic phase sepa-
ration at low-T is an intrinsic feature of electron-doped
manganite perovskites, or at least of these systems where
electron-doping is accomplished by compositional varia-
tion (phase separation might be favored by local chemical
variations). The very recent study of Cai-xSm^MnOa,
x = 0.15, by Algabarel et alM provides similar evidence
for the existence of monoclinic C-AFM and orthorhom-
bic FM-canted G-AFM in phase separated regions of
compositionally homogeneous samples by demonstrating
that their relative phase fractions could be influenced by
an external applied magnetic field. There may be some
compositional separation at low- a; because lighter doping
gives a higher probability of La-clustering (as recently
demonstrated in Lai-^Sr^MnOg^), however, the signifi-
cance of this decreases with increasing x. The competing
magnetic states are extremely finely balanced over a
broad crossover regime 0.06 < x < 0.16, within which
samples do not settle into single thermodynamically
stable phases at low-T.
B. Spin-lattice Coupling and Frustration
The competition between orthorhombic G-AFM and
monoclinic C-AFM [Figs. Et^)] reflects the balance of
gains and losses associated with the co-operative J-T dis-
tortion of the latter; a lowering in exchange energy on
the one hand, and an increase in elastic energy on the
other. This balance is affected by the relative strengths
of FM DE and AFM SE interactions. The formation
of the monoclinic phase below Tn (C-AFM) corresponds
to the ordering of FM DE interactions (which exist as
short-range fluctuations above Tn (C-AFM)2i) into in-
finite 1-D chains, by AFM SE interactions perpendic-
ular to them. The monoclinic phase fraction grows as
T decreases because these AFM SE interactions become
stronger, decreasing the exchange energy of the mono-
clinic phase and making its total energy less than that of
the paramagnetic orthorhombic phase. At TV(G-AFM),
however, the monoclinic phase fraction of the x = 0.12
sample stops growing because AFM SE interactions be-
come strong enough to stabilize G-AFM in the remaining
orthorhombic phase fraction. This lowers the exchange
energy and therefore the total energy of the orthorhombic
phase, restoring its status as the more stable crystallo-
graphic polymorph. Similarly, in the x = 0.20 sample
[Figs. Efd)], the monoclinic phase fraction stops growing
at To (W-C) because the distortion required to form the
W-C phase is less energetically costly than that required
to form C-AFM.
The effects of T N (G-AFM) on the distortion of the
pseudo-cubic lattices, as seen in Fig. EI a and c), are
also of interest. The G-AFM structure is isotropic and
therefore should not effect the lattice, as is indeed the
case for x — 0.03 [Fig. EI a )]- For x — 0.12, however,
there seems to be a small but significant magnetostric-
tive effect at T N (G-AFM) [Fig. GJc)], whereby the (010)
axis elongates slightly relative to (101). If this effect is
real, it is presumably related to (010) being the direc-
tion of net FM in the cluster glassi& (see Section IHI Bl) ;
however, since the effect is small, speculation on a mech-
anism will be avoided. More surprising is the large ef-
fect of TV (G-AFM) on the distortion of the monoclinic
lattice [Fig. Et c )L where no FM clusters are involved.
Note firstly that the monoclinic phase does not adopt the
fully-ordered C-AFM state immediately upon symmetry-
lowering. This is clear for the x = 0.20 sample, for which
the magnetic order parameter [Fig. EI' 1 )] and monoclinic
distortion [Fig. EI/)] show a strong T-dependencies be-
low T (W-C) [Fig. Et<?)], despite the fact that the mon-
oclinic phase fraction no longer grows. In this light, the
refined monoclinic cell for x = 0.12 [Fig. EI C )] might
actually represent an average monoclinic cell, for which
a reduction in the monoclinic distortion would not nec-
essarily represent a deterioration of established C-AFM
ordered domains. It could simply be a convolution of the
delay between symmetry-lowering and the establishment
of a fully-ordered C-AFM state on the one hand, and
the increasing strength of the competing G-AFM state
on the other.
An intriguing extension of this argument is the possi-
bility that the increasing strength of AFM SE interac-
tions as T decreases not only slows the establishment of
long-range FM DE (i.e. C-AFM) in the monoclinic phase,
but actually leads to the establishment of G-AFM there
instead. In the extreme case, G-AFM might actually re-
place established C-AFM domains. While there is no
direct evidence for this in the present data, the reader's
attention is brought to the highly analogous 'bi-layered'
manganite perovskite La2_2sSri + 2 a ;Mn207, where the C-
AFM phase also requires a symmetry-lowering transition
(from tetragonal to orthorhombic). 31 A 10 % electron-
doped sample in this system (x = 0.90) exhibited not
only a structural transition followed by two magnetic
transitions IV (C- AFM) = 110 K and Tjv(G-AFM) = 60
K, but a clear decrease in the C-AFM order parameter
below (G-AFM); i.e., the G-AFM state 'colonizes' the
monoclinic regions established by C-AFM orbital polar-
ization at higher-T.
Low-T phase inhomogeneities ultimately arise because
short-range FM DE correlations appear at higher T than
AFM SE correlations, as has been noted in studies of
weak diffuse neutron scattering above the magnetic long-
range-ordering transition temperatures .iSiSiSii The C-
AFM magnetic state can form at a higher temperature
than the G-AFM state because the AFM interactions
only have to be strong enough to create AFM order in
2-D, rather than 3-D. At the same time, the strength of
the FM SE interactions is obviously related to the con-
centration of e g electrons (x) facilitating it. Magnetic
and crystallographic ground states are frustrated in the
region where 1-D FM SE interactions are strong enough
to cause e g orbital polarization and symmetry-lowering
at Tjv(C), but where AFM DE is strong enough to cre-
ate 3-D order below TV (G-AFM). This frustration is il-
lustrated by Fig where the phase diagram [Fig[7j6)]
shows orthorhombic G-AFM to be the ground state for
x up to 0.16, but Fig[7{a) shows that less than 20 % of
the x = 0.16 sample is actually in this state at low-T.
V. CONCLUSION
This study (Part II) has used TOF NPD data in con-
junction with physical property measurements to identify
and characterize the low-T phases present in samples of
Cai-^La^MnOs (0.00 < x < 0.20) studied by neutron
scattering in Part I^ The samples appear to be com-
positionally homogeneous and yet display multiple low-T
magnetic states, exemplifying the delicate balance among
competing interactions characteristic of the CMR man-
ganites. The results, in conjunction with those of Part
I, have been used to construct a ground state phase dia-
gram.
The theme that emerges from this phase diagram is
thee strong effect that the introduction of FM DE inter-
actions has on the AFM SE (for ideal G-AFM at x = 0).
In orthorhombic G-AFM these FM interactions create 0-
D correlations; they have no influence on the long-range
magnetic structure until the AFM SE interactions be-
come strong enough below Tv (G-AFM) to create long-
range 3-D AFM order, at which point they are 'frozen
in' as isolated FM clusters, or (in sufficient densities) as
an FM cluster-glass. In monoclinic C-AFM they become
1-D in character, leading to d^ z 2_ r -2 orbital polarization
and hence symmetry-lowering. At the low- a; end of the
C-AFM regime, the actual thermodynamic ground state
is G-AFM, but this is frustrated by the irreversible struc-
7
0.00 0.05 0.10 0.15 0.20
xinCa La MnO
1-x x 3
FIG. 7: (a) Refined monoclinic phase fraction at 20 K
as a function is x. (b) Ground state phase diagram of
Cai-^La^MnOa, 0.0 < x < 0.2, mapped onto the crystallo-
graphic and magnetic phase transitions determined from NPD
data. Closed circles represent the coincident magnetic transi-
tions Tjv (G-AFM) and Tc, and open circles represent the co-
incident magnetic transition Tn (C-AFM) and structural tran-
sition To- Monoclinic (as opposed to orthorhombic) regions
are dark grey. The presence of FM clusters is indicated by
diagonal black lines, and the FM cluster glass by black diag-
onal cross-hatching. The W-C-type phase is black. Note that
at the high-a; end of the G-AFM regime, the stability of the
C-AFM state is very high, and therefore very little G-AFM
is actually observed (a); the same is true for the W-C-type
at x = 0.20, and the converse is true at the low- a; end of the
C-AFM regime.
tural phase transition favoring C-AFM. In each case, FM
DE interactions play a privileged role because they ap-
pear at higher-T than the competing AFM SE interac-
tions, allowing them to influence the structure on cooling
and pre-dispose the system to a particular low-T state.
The changing dimensionality of the FM DE inter-
actions with x, from 0-D in the electron-doped G-
AFM regime to 1-D in the C-AFM regime, foreshad-
ows the subsequent change to 2-D (for the A- AFM state
at x ~ h in some manganite perovskite systems e.g.
Sri-^Pr^MnOg 14 ) and finally 3-D (for the FM state
at i < x < 1 in most such systems). Between the
fully-electron-doped (exhibiting a different type of A-
8
AFM) and stoichiometric CaMn0 3 (G-AFM) end mem-
bers, these changes in DE dimensionality with e g elec-
tron concentration underlie the magnetic phase diagram
of the CMR manganites. At the same time, the order-
ing of these e g electrons via the J-T effect underlies the
crystallographic phase diagram. The C-AFM, A-AFM
and FM magnetic states are examples of co-operation
between these spin and orbital ordering effects, while the
phase diagram is also punctuated by regions in which
they compete, notably the C-E state at x ~ 0.5 and its
W-C variants between the C-E and C-AFM regimes.
der contract W-31-109-ENG-38 (CDL, DNA), by the Na-
tional Science Foundation (NSF) under CAREER grant
DMR 9982834 (JJN) and by FAPESP-Brazil (EG). Work
at the university of Maryland was supported in part by
the NSF MRSEC, DMR 00-80008. The authors thank S.
Short of 1PNS for technical support in the collection of
NPD data.
Acknowledgments
This work was supported by the U.S. Department of
Energy, Basic Energy Sciences - Materials Sciences, un-
* Electronic address: [email protected]
1 P.-G. de Gennes, Phys. Rev. 118, 141 (1960).
2 S.-W. Cheong and H. Y. Hwang (Gordan and Breach, Lon-
don, 1999).
3 K. Hagdorn, D. Hohlwein, J. Ihringer, K. Knorr,
W. Prandl, H. Ritter, H. Schmid, and T. Zeiske, Eur. Phys.
J. B 11, 243 (1999).
4 P. N. Santhosh, J. Goldberger, P. M. Woodward, T. Vogt,
W. P. Lee, and A. J. Epstein, Phys. Rev. B 62, 14928
(2000).
5 C. Martin, A. Maignan, M. Hervieu, B. Raveau, Z. Jirak,
A. Kurbakov, V. Trounov, G. Andre, and F. Bouree, Jour-
nal of Magnetism and Magnetic Materials 205, 184 (1999).
6 M. Respaud, J. M. Broto, H. Rakoto, J. Vanacken, P. Wag-
ner, C. Martin, A. Maignan, and B. Raveau, Phys. Rev. B
63, 144426 (2001).
7 C. Martin, A. Maignan, M. Hervieu, B. Raveau, Z. Jirak,
M. M. Savosta, A. Kurbakov, V. Trounov, G. Andre, and
F. Bouree, Phys. Rev. B 62, 6442 (2000).
s E. O. Wollan and W. C. Koehler, Phys. Rev. 100, 545
(1955).
9 J. B. Goodenough, Phys. Rev. 100, 564 (1955).
10 A. Maignan, C. Martin, F. Damay, and B. Raveau, Chem.
Mater. 10, 950 (1998).
11 H. Chiba, M. Kikuchi, K. Kusaba, Y. Muraoka, and
Y. Syono, Solid State Communications 99, 499 (1996).
12 H. Aliaga, M. T. Causa, B. Alascio, H. Salva, M. Tovar,
D. Vega, G. Polla, G. Leyva, and P. Konig, J. Magn. Magn.
Mater. 226, 791 (2001).
13 R. Mahendiran, A. Maignan, C. Martin, M. Hervieu, and
B. Raveau, Phys. Rev. B 62, 11644 (2000).
14 C. Martin, A. Maignan, M. Hervieu, and B. Raveau, Phys.
Rev. B 60, 12191 (1999).
15 D. A. Filippov, R. Z. Levitin, A. N. Vasil'ev, T. N.
Voloshok, H. Kageyama, and R. Suryanarayanan, Phys.
Rev. B 65, 100404 (2002).
16 E. Granado, C. D. Ling, J. J. Neumeier, D. N. Argyriou,
J. W. Lynn, and P. L. Lee (submitted).
17 M. Hennion, F. Moussa, G. Biotteau, J. Rodriguez-
Carvajal, L. Pinsard, and A. Revcolevschi, Phys. Rev.
Lett. 81, 1957 (1998).
18 M. Hennion, F. Moussa, J. Rodriguez-Carvajal, L. Pinsard,
and A. Revcolevschi, Phys. Rev. B 56, R497 (1997).
19 M. Hennion, F. Moussa, G. Biotteau, J. Rodriguez-
Carvajal, L. Pinsard, and A. Revcolevschi, Phys. Rev. B
61, 9513 (2000).
20 A. Maignan, C. Martin, F. Damay, B. Raveau, and J. Hejt-
manek, Phys. Rev. B 58, 2758 (1998).
21 M.-M. Couffon, G. Rocher, and J. Protas, C. R. Hebd.
Seances Acad. Sci. 258, 1847 (1964).
22 G. Lepicard and J. Protas, Bull. Soc. Fr. Mineral. Crystal-
logr. 89, 318 (1966).
23 C. D. Ling, J. J. Neumeier, and D. N. Argyriou, J. Solid
State Chem. 160, 167 (2001).
24 W. Bao, J. D. Axe, C. H. Chen, and S.-W. Cheong, Phys.
Rev. Lett. 78, 543 (1997).
25 J. J. Neumeier and J. L. Cohn, Phys. Rev. B 62, 14319
(2000).
26 P. G. Radaelli, D. E. Cox, L. Capogna, S.-W. Cheong, and
M. Marezio, Phys. Rev. B 59, 14440 (1999).
27 M. T. Fernandez-Diaz, J. L. Martinez, J. M. Alonso, and
E. Herrero, Phys. Rev. B 59, 1277 (1999).
28 M. Pissas, G. Kallias, M. Hofmann, and D. M. Tobbens,
Phys. Rev. B 65, 064413 (2002).
29 P. H. Algarabel, J. M. D. Teresa, B. Garci'a-Landa,
L. Morellon, M. R. Ibarra, C. Ritter, R. Mahendiran,
A. Maignan, M. Hervieu, C. Martin, et al., Phys. Rev.
B 65, 104437 (2002).
30 T. Shibata, B. Bunker, J. F. Mitchell, and P. Schiffer, Phys.
Rev. Lett. 88, 207205 (2002).
31 C. D. Ling, J. E. Millburn, J.F.Mitchell, D. N. Argyriou,
J. Linton, and H. N. Bordallo, Phys.Rev. B 62, 15096
(2000).