Spiral magnetic structure of Fe in Van der Waals gapped FeOCl and polyaniline-intercalated FeOCl

Article abstract of DOI:10.1103/PhysRevB.62.14157

High-resolution and magnetic neutron-diffraction measurements were performed to investigate the crystal and magnetic structures of bilayered FeOCl and polyaniline-intercalated FeOCl. A quasi-two-dimensional crystallographic structure, where charge neutral (Fe₂O₂Cl₂)_n lamellas are weakly linked via Van der Waals interactions, has made FeOCl a good host for accommodating guest molecules. A three-dimensional long-range ordering of the Fe spins in FeOCl develops below 80 K, with a magnetic unit cell 28 times the size of the nuclear one. A spiral magnetic structure was obtained, reflecting the competition between antiferromagnetic superexchange coupling and ferromagnetic direct exchange coupling. Polyaniline intercalation interrupts the magnetic correlations between the neighboring bilayers, rendering a coupled-bilayer quasi-two-dimensional magnetic order for the Fe spins in (C₆D₄ND)_0.16FeOCl. No significant change on the ordering temperature of the Fe spins by the intercalation reaction was observed.

Full text of DOI:10.1103/PhysRevB.62.14157

PHYSICAL REVIEW B
VOLUME 62, NUMBER 21
1 DECEMBER 2000-I
Spiral magnetic structure of Fe in Van der Waals gapped FeOCl
and polyaniline-intercalated FeOCl
S. R. Hwang, W.-H. Li, and K. C. Lee
Department of Physics, National Central University, Chung-Li, Taiwan 32054, Republic of China
J. W. Lynn
NIST Center for Neutron Research, NIST, Gaithersburg, Maryland 20899-8562
C.-G. Wu
Department of Chemistry, National Central University, Chung-Li, Taiwan 32054, Republic of China
͑Received 16 February 2000; revised manuscript received 21 June 2000͒
High-resolution and magnetic neutron-diffraction measurements were performed to investigate the crystal
and magnetic structures of bilayered FeOCl and polyaniline-intercalated FeOCl. A quasi-two-dimensional
crystallographic structure, where charge neutral (Fe2O2Cl2)n lamellas are weakly linked via Van der Waals
interactions, has made FeOCl a good host for accommodating guest molecules. A three-dimensional long-
range ordering of the Fe spins in FeOCl develops below 80 K, with a magnetic unit cell 28 times the size of
the nuclear one. A spiral magnetic structure was obtained, reflecting the competition between antiferromagnetic
superexchange coupling and ferromagnetic direct exchange coupling. Polyaniline intercalation interrupts the
magnetic correlations between the neighboring bilayers, rendering a coupled-bilayer quasi-two-dimensional
magnetic order for the Fe spins in (C6D4ND)0.16FeOCl. No significant change on the ordering temperature of
the Fe spins by the intercalation reaction was observed.
I. INTRODUCTION
provide valuable information regarding to the effects of the
presence of electrically conducting guest to the magnetic
properties of the host. We have found, in FeOCl, a spiral
spin arrangement along the shortest crystallographic axis di-
rection, that we believe is a direct result of the competition
between the ferromagnetic Fe-Fe direct exchange coupling
and the antiferromagnetic Fe-Fe superexchange coupling
mediated through the O ions located between them. For spins
along the other two crystallographic directions, where the
strength of the Fe-O-Fe superexchange coupling was geo-
metrically reduced, simpler spin arrangements are observed.
A bilayered Q2D magnetic order was found for the Fe spins
in polyaniline-intercalated FeOCl, showing that the interca-
lation reaction interrupts the magnetic interaction between
the neighboring (Fe2O2Cl2)n bilayers.
As a promising material for practical applications in a
variety of devices, electrically conductive polymers have re-
cently received considerable attention. One way to reduce
tangling in polymer molecules and to obtain orientated poly-
mer chains is to grow them inside a structurally organized
host framework, with accessible oriented tunnels. The quasi-
two-dimensional ͑Q2D͒ nature, that is a result of the appear-
ance of Van der Waals crystallographic gaps,1 has made
FeOCl an excellent inorganic host for intercalation
reactions.2 Various electrically active materials, such as
tetrathiafulvalene,3
polypyrrole,4
and
polyaniline
(C6H4NH),5,6 have been intercalated into FeOCl to form
good low-dimensional conductors. Lithium ion intercalated
FeOCl has also been used7–9 as cathode materials for re-
chargeable lithium batteries. The topochemical substitution
of the Cl layers in FeOCl with OH layers has been found10–12
to help stabilize the intercalated Liϩ ions and to improve the
cathode performance.
II. CRYSTAL STRUCTURE
Polycrystalline iron-oxychloride FeOCl was prepared by
the chemical vapor transport technique.15 High-purity Fe2O3
and FeCl3 powders with a mole ratio of 1:1.3 were mixed
well and then sealed in an evacuated glass tube, followed by
heating at 380 °C for 40 h. They were then slowly cooled to
room temperature. The resultant product was washed thor-
oughly with water-free acetone to remove the excess FeCl3,
and dried naturally in vacuum. The grains in the resultant
polycrystalline sample had a lamellar, rather than spherical,
shape reflecting the anisotropic crystallographic structure of
the compound. FeOCl is rather sensitive to humidity, thus
isolation from the atmospheric environment at all times is
essential to keep the compound from being hydrated. For
polyaniline-intercalated FeOCl, 100 ml of a 3% deuterated
aniline solution ͑32.3 mmol͒ in acetonitrile was stirred with
Until recently, many of the studies made on the FeOCl
related compounds had concentrated on their electrochemical
properties, relatively fewer on their basic properties, which
remain poorly understood. Phonon spectra of FeOCl for en-
ergies below 80 meV have been reported13 to resolve six
broad vibrational modes using inelastic neutron-scattering
measurements. A spiral magnetic structure, characterized by
a propagation vector of ͑0.275 0.5 0.5͒, for the Fe spins in
FeOCl has been proposed14 based on neutron-diffraction
measurements. In the present work, we have performed
neutron-diffraction measurements to explore the detailed
crystal and magnetic structures of naturally bilayered FeOCl
as well as polyaniline-intercalated FeOCl. The results may
0163-1829/2000/62͑21͒/14157͑7͒/$15.00
PRB 62
14 157
©2000 The American Physical Society

14 158
HWANG, LI, LEE, LYNN, AND WU
PRB 62
5.0 g ͑4.66 mmol͒ FeOCl in air for a week. The black shiny
microcrystalline product was then isolated by filtration,
washed with acetone and dried naturally in vacuum. The
intercalation reaction was deemed complete when the ͑0 0 k͒
peaks of the starting FeOCl disappeared from the x-ray
powder-diffraction pattern. Elemental analysis shows a
C6D4ND to FeOCl ratio of 0.16:1, hence a chemical formula
of (C6D4ND)0.16FeOCl, for the resultant product. This
polyaniline intercalated FeOCl compound is hereafter de-
noted as (PANI)0.16FeOCl.
X-ray-diffraction measurements were used first to charac-
terize the sample, and the detailed crystal structure was de-
termined by a complete structural analysis using neutron-
diffraction measurements and Rietveld analysis.16,17 A high-
resolution neutron-diffraction pattern, covering a scattering-
angle range from 3° to 165° and taken at room temperature,
was collected on BT-1, the 32-detector powder diffracto-
meter at the NIST Center for Neutron Research. The neu-
trons had a wavelength of ␭ϭ1.5401 Å, defined by a
Cu͑311͒ monochromator crystal. Angular collimators with
FIG. 1. Observed ͑crosses͒ and fitted ͑solid lines͒ neutron
powder-diffraction patterns of FeOCl collected at room tempera-
ture. The differences are plotted at the bottom. The short vertical
lines shown below the pattern mark the calculated positions of the
Bragg reflections.
horizontal divergences of 15 , 20 , and 7 full width at half
Ј
Ј
Ј
maximum ͑FWHM͒ acceptances were used for the in-pile,
monochromatic, and diffracted beams, respectively. During
the measurements, the sample was loaded into a cylindrical
vanadium can, which produces no measurable neutron-
diffraction peaks, and was slowly rotating about the cylindri-
cal axis to account for the anisotropic sharp of the grains in
the sample.
The crystal structure of FeOCl may be viewed as a stack-
ing of charge-neutral Q2D (Fe2O2Cl2)n lamellas, where the
adjacent blocks are weakly bonded23 via Van der Waals in-
teractions, with a gap separation of 2.69617͑3͒ Å at room
temperature. We note that the Fe-O-Fe bond angles and the
Fe-Fe separations along the three crystallographic directions
are quite different, resulting in different magnetic interac-
tions and resultant configurations for the Fe spins along the
The diffraction pattern was analyzed using the GSAS
͑General Structure Analysis System͒ program,18 following
the Rietveld refinement method, with coherent scattering am-
plitudes of 0.954, 0.580, and 0.958ϫ10Ϫ12 cm for Fe, O, and
Cl, respectively. Several models with different symmetries
were assumed during the preliminary analysis. The refine-
ment was finally carried out by assuming an orthorhombic
symmetry of space group Pmmn, with a unit cell consisting
of two pairs of distorted Cl-Fe-O chains. All structural and
lattice parameters were allowed to vary simultaneously until
Rw , the weighted R factor, differed by less than one part in a
thousand within two successive cycles. Figure 1 shows the
observed ͑crosses͒ and fitted ͑solid lines͒ patterns of FeOCl,
with their differences plotted at the bottom. Clearly, the cal-
culated and the observed patterns agree very well, showing
that the observed pattern can be adequately described by the
proposed structure shown in Fig. 2, which is the same one as
has been proposed in other studies.1,19,20 Table I lists the
refined structural parameters and selected bond angles, which
are in good agreement, but are more precise than those re-
ported in other studies.1,19 Uncertainties quoted here and
throughout this article are statistical and represent one stan-
dard deviation. Note that the isotropic temperature parameter
B, which indicates the mean-square displacement of the at-
oms from their average sites, is much larger for Cl atoms
than for Fe and O atoms, showing that the Cl atoms are
relatively weakly bound to the other atoms. Note that there is
essentially no improvement made for refinements using an-
isotropic temperature parameters. Employing the Dollase and
March model,21,22 a factor of 1.57 was obtained for c-axis
preferred orientation, accounting for the anisotropic shape of
the grains in the polycrystalline sample. We estimated any
impurity phases in the sample to be less than 1%.
FIG. 2. Proposed crystal structure of FeOCl, where a unit cell
contains two sets of distorted Cl-Fe-O chains.

PRB 62
SPIRAL MAGNETIC STRUCTURE OF Fe IN VAN DER . . .
14 159
TABLE I. Refined structural parameters, selected bond lengths,
and bond angles for FeOCl at room temperature. B represents the
2
2
2
isotropic temperature parameter with Bϵ␲
␮
, where
␮
is
͘
͗
͘
͗
the mean-square displacement.
FeOCl, space group Pmmn, room temperature
aϭ3.29921(3) Å, bϭ3.77450(3) Å, cϭ7.90852(11) Å
Atom
x
y
z
B ͑Å2͒
Occupancy
Fe
O
Cl
0.75
0.25
0.25
0.25
0.25
0.25
0.1154͑1͒
0.9510͑2͒
0.3301͑1͒
0.76͑2͒
0.85͑3͒
1.57͑2͒
0.99͑1͒
0.99͑1͒
1.00͑1͒
Rp(%)ϭ5.11
Rw(%)ϭ6.15
␹ϭ1.31
Fe-O-Fe bond angle
103.49͑4͒°
Fe-Fe bond length
3.29921͑2͒ Å
3.77450͑3͒ Å
3.10065͑6͒ Å
Along a
Along b
Along c
148.89͑4͒°
99.56͑6͒°
three directions ͑see below͒. Apparently, the Q2D structural
nature makes FeOCl a good inorganic host26 for accommo-
dating guest molecules in the Van der Waals gaps, and form-
ing intercalated FeOCl. The basic (Fe2O2Cl2)n lamellas con-
sist of edge-sharing distorted FeCl2O4 octahedrons, with the
Cl layers outermost on either side of a central (FeO͒2 bilayer,
forming a bilayered Q2D Fe-O-Fe network ͑designated as
the a-b plane͒, while the network along the c-axis direction is
interrupted by two weakly bonded Cl layers.
Intercalation reaction of polyaniline into FeOCl results in
a 103% increase in the gap separation, apparently due to the
existence of polyaniline in the Van der Waals gaps, and a
shift, by an amount of (a/2 b/2), in the relative positions
between the two neighboring bilayers separated by polyani-
line. A 230% increase in the c axis parameter hence resulted.
The lateral shift in the neighboring bilayers found for the
present (PANI)0.16FeOCl is similar to what has been
observed3 in a tetrathiafulvalene-intercalated FeOCl. On the
other hand, the crystal structure within the bilayered
(Fe2O2Cl2)n lamellas remains, with the a and b axis param-
eters both increase by ϳ1.5%. Unfortunately, the positional
parameters for polyaniline in the present compound can not
be uniquely determined from the present data, mainly due to
the lack of a crystalline periodicity in the polyaniline them-
selves.
FIG. 3. Neutron-diffraction patterns of FeOCl taken at ͑a͒ 190 K
and ͑b͒ 10 K. The magnetic peaks are indexed using the notation of
(h k l)Ϯϵ(h k l)Ϯ(2/7 1/2 1/2).
lindrical aluminum can filled with helium exchange gas to
facilitate thermal conduction at low temperatures. The
sample was cooled using a closed-cycle refrigerator, oper-
ated between 10 and 200 K.
A. FeOCl
At temperatures well above the magnetic ordering tem-
perature, the spins of unpaired electrons scatter neutrons in-
coherently, so that the magnetic scattering appears as a para-
magnetic background in the diffraction pattern. When
magnetic order develops, at temperatures below the ordering
temperature, the magnetic scattering evolves into Bragg
peaks. Figures 3͑a͒ and 3͑b͒ show the diffraction patterns
taken at 190 and 10 K, respectively. The solid curves are fits
of the data to the Gaussian instrumental resolution function.
All the peaks in the pattern taken at 190 K can be indexed ͑as
shown͒ based on the crystallographic unit cell listed in Table
I, but using somewhat smaller lattice constants due to ther-
mal contraction. A series of new resolution-limited peaks ͑as
indicated by the arrows͒ developed as the temperature was
reduced to 10 K, signifying the long-range ordering of the Fe
spins.
III. MAGNETIC STRUCTURE
Neutron magnetic diffraction experiments were also con-
ducted at NIST, using the BT-9 and BT-4 triple-axis spec-
trometers operated in the double-axis mode. The magnetic
diffraction of FeOCl was carried out on the BT-9, whereas
the BT-4 was used to collect data for (PANI)0.16FeOCl. The
neutrons used had a wavelength of ␭ϭ2.359 Å defined
by using a pyrolytic graphite PG͑002͒ monochromator crys-
tal. A PG filter placed before the monochromator position
was employed to suppress any higher-order wavelength con-
tamination. For the measurements performed at BT-9 ͑BT-
4͒, angular collimators of horizontal divergences 40Ј-48Ј-48Ј
͑60Ј-40Ј-40Ј͒ FWHM acceptance for the in-pile, monochro-
matic, and diffracted beams, respectively, were employed.
During the measurements, the sample was loaded into a cy-

14 160
HWANG, LI, LEE, LYNN, AND WU
PRB 62
FIG. 5. Proposed magnetic structure for Fe spins in the a-c
plane, showing a spiral spin arrangement along the a-axis direction.
It is certainly not a simple matter to determine the spin
arrangement. We employed the symmetry group analysis
method25 to model the magnetic structure. The spin vectors
of the Fe ions can be decomposed into a superposition of
magnetic basis functions, which work as the basic magnetic
configuration modes. In terms of the magnetic basis func-
tions, the spin vector of the jth Fe ion in a designated basis
nuclear unit cell may be expressed as
S ϭ
bj
C m ,
k k
͚
where Ck is the superposition coefficient for the magnetic
basis function mk . The spin vector of the jth Fe ion in the
nth nuclear unit cell, which can be reached from the basis
unit cell by the application of a translation vector Rn , then
takes the form of
FIG. 4. The observed ͑solid lines͒ and calculated ͑dashed lines͒
Bragg angles for the propagation vectors (h k l)Ϯϵ(h k l)
Ϯ(kx 1/2 1/2) as a function of the modulation parameter kx . The
arrows indicate the positions where the observed and calculated
lines meet.
S ϭ
S exp i␣ ͒,
͑
bj n
͚
rj
where ␣nϭK•Rn and the summation takes over the two
We employed a diagrammatic method,24 based on sym-
metry considerations, to find the appropriate propagation
vectors, (hkl)Ϯϵ(hkl)ϮK, where Kϭ(kxkykz) is the
modulation vector for the magnetic peaks. The only K that fit
the observed magnetic peaks properly is shown in Fig. 4.
The horizontal solid lines represent the Bragg positions of
the observed magnetic peaks and the dashed lines mark the
calculated ones for a modulation vector, Kϭ(kx 0.5 0.5).
Apparently, at kxϭ0.2826ϭ2/7 all five calculated lines in-
tersect the observed ones, as indicated by the arrows. The
vector ͑0.2826 0.5 0.5͒ obtained for the modulation param-
eters is the only one that fits the data, as is expected from the
uniqueness of the irreducible representation,24 and is essen-
tially the same with the one reported by Adam and
Buisson.14 Accordingly, the magnetic peaks were then in-
dexed by adopting the notation (hkl)Ϯϵ(hkl)
Ϯ(2/7 1/2 1/2), as marked in Fig. 3͑b͒. The magnetic unit
cell is then double the nuclear one in both the b and c axis
directions, while it is seven times longer in the a-axis direc-
tion. There are therefore 28 nuclear unit cells and 56 Fe ions
in a magnetic unit cell.
magnetic modulation vectors ϮK.
For a symmetry of space group Pmmn, there is one irre-
ducible representation.26 The star of K contains two arms for
the two modulation vectors ϮK, each of which is associated
with two row vectors. There are therefore 12 magnetic basis
functions for each Fe ion. We then calculated the intensity24
for all possible combinations of the magnetic basis functions,
and found that only the set of spin vectors where
TABLE II. Calculated and observed magnetic integrated inten-
sities, where h k l Ϯϵ h k l Ϯ(2/7 1/2 1/2). The intensities are
͕
͖
͕
͖
normalized with respect to the peak at 23.2°.
Peak index h k l
2␪
Iobs
Ical
͕
͖
obs
0 0 0 Ϯϩ 0 0 1 Ϫϩ 0 1 0 Ϫϩ 0 1 1
23.20° 1.00͑1͒
1
Ϫ
͕
͖
͕
͖
͕
͖
͕
͖
0 0 1 ϩϩ 0 0 2 ϩ 0 1 1 ϩϩ 0 1 2
Ϫ
Ϫ
Ϫ
¯
34.00° 0.32͑2͒ 0.36
͕
͕
͖
͖
¯
͖
͕
͕
͖
͖
¯
͖
͕
͖
͖
͕
͕
͖
͖
1 0 0 Ϫϩ 1 1 0 Ϫϩ 1 0 1 Ϫϩ 1 1 1
36.15° 0.27͑2͒ 0.27
44.19° 0.15͑3͒ 0.18
͕
1 0 1 ϩ 1 1 1 Ϫϩ 1 0 2 Ϫϩ 1 1 2
Ϫ
Ϫ
Ϫ
͕
͕
͕
͕
͖
͕
͖
͖
0 0 2 ϩ 0 1 2 Ϫϩ 0 0 3 Ϫϩ 0 1 3
ϩ
¯
͖
49.54° 0.11͑4͒ 0.16
͖
͕
͕
͖
͕

PRB 62
SPIRAL MAGNETIC STRUCTURE OF Fe IN VAN DER . . .
14 161
FIG. 6. Temperature dependence of the mag-
netic intensity measured at a Bragg angle of
23.2°, where the circles and triangles mark, re-
spectively, the data taken without and with an
analyzer-crystal. The ordering temperature for the
Fe spins in FeOCl is TNϷ80 K.
agreement is excellent, showing that the proposed spin struc-
ture fits to the observations well.
Ϫsin ␣n
cos ␣ Ϫ␲K͒
͑
n
0
0
S
n1ϭ
and Sn2ϭ
ͩ ͪ
ͩ
ͪ
The quite distinct spin configurations obtained for the Fe
ions along the three crystallographic directions can be
readily understood from the crystallographic anisotropy of
the Fe and O ions. In edge-sharing FeCl2O4 octahedrons, the
d orbitals of Fe3ϩ are split by the crystalline electrical field
effect into eg and half-filled t2g orbitals, while O2Ϫ occupies
the filled p orbitals.27,28 Two types of magnetic couplings
between the Fe ions may then be anticipated, a ferromagnetic
direct-exchange ͑FMDE͒ coupling between two neighboring
Fe eg orbitals, and an antiferromagnetic superexchange
͑AFMSE͒ coupling between two Fe half-filled t2g orbitals
mediated through the O p orbitals located between them.29
Clearly, the coupling strength depends strongly on the geom-
etry of the Fe-O-Fe bonding. A wider Fe-O-Fe bond angle
gives rise to a stronger AFMSE interaction, while a shorter
Fe-Fe separation results in a stronger FMDE interaction.
Along the b-axis direction, a simple antiferromagnetic ar-
rangement was observed, showing that the AFMSE coupling
dominates the FMDE coupling. This behavior can be under-
stood, since the Fe-O-Fe bond angle ͑ϳ148.9°͒ is large and
the Fe-Fe are well separated ͑ϳ3.763 Å at 10 K͒ along this
direction. Along the a-axis direction, on the other hand, the
Fe-O-Fe bond angle is smaller ͑ϳ103.5°͒ and has a shorter
cos ␣n
sin ␣ Ϫ␲K͒
͑
n
fits the observations. Both of these spin vectors do not carry
a component along the b-axis direction, showing that the
moment lies in the a-c plane. Figure 5 shows the proposed
magnetic structure at low temperature for the Fe spins in the
a-c plane. The moment directions for the neighboring spins
along the a-axis direction are 2ϫ360°/7Ϸ102.85° apart, ac-
counting for the kxϭ2/7 obtained for the modulation along
the a-axis direction. Unfortunately, the relative orientations
of the moments between the neighboring layers ͑along the
c-axis direction͒ within each bilayer cannot be uniquely de-
termined from the intensity calculations, as has been pointed
out by Adam and Buisson.14 The relative orientations of
these moments shown in Fig. 5 are arbitrarily chosen. Nev-
ertheless, the moments between the neighboring bilayers are
antiferromagnetically coupled, as shown in Fig. 5. Along the
b-axis direction, the neighboring spins are antiparallel ͑not
shown͒. Comparisons between the observed and calculated
magnetic intensities are listed in Table II, where the factor of
1.57 ͑as discussed above͒ for the c-axis preferred orientation
has been taken into account, and the intensities are normal-
ized with respect to the isolated peak at 23.2°. Clearly, the
FIG. 7. Magnetic diffraction pattern of
(PANI)0.16FeOCl obtained at 10 K, showing a
standard coupled-bilayer Q2D scattering profile.

14 162
HWANG, LI, LEE, LYNN, AND WU
PRB 62
tion of ϳ1 meV. By comparing the magnetic intensities to
the nuclear ones, we obtained a saturated moment of ␮
͗
͘
Z
ϭ4.01(5)␮ for the Fe ions, which is in good agreement
B
with the expected moment of 4.0␮B .
B. „PANI…0.16FeOCl
In Fig. 7, we show the magnetic diffraction pattern of
͑PANI͒0.16FeOCl obtained at 10 K, where the diffraction pat-
tern taken at 100 K, serving as the nonmagnetic background,
has been subtracted. This pattern reveals peaks that devel-
oped as the temperature was reduced from 100 to 10 K,
signifying the Fe spin ordering. As seen in Fig. 7, the ob-
served magnetic reflections have a sharply rising leading
edge followed by a trailing edge extending to large scattering
angles. Such an asymmetric scattering profile with an oscil-
latory intensity is characteristic of coupled-bilayer Q2D
behavior.30–32 The solid curve shown in Fig. 7 is a fit of the
data to the coupled-bilayer 2D magnetic scattering profile32
convoluted with the Gaussian instrumental resolution func-
tion. Clearly, the fit is very good, showing that the observed
pattern can be described well using the proposed bilayer pro-
file. The magnetic peaks were identified as originating from
the ͑1/3 1/2͒ rod of a coupled-bilayer spiral spin configura-
tion shown in Fig. 8. Again, the relative orientations of the
moments between the neighboring layers cannot be uniquely
determined from the present data. Accordingly, the magnetic
unit cell triples the nuclear one along the a-axis direction and
doubles that along the b-axis direction, while no significant
magnetic correlations between the neighboring bilayers were
observed.
Evidently, the interbilayer couplings in ͑PANI͒0.16FeOCl
are interrupted by the appearance of polyaniline in the Van
der Waals gaps that double the gap separation, rendering the
dipolar interactions between the neighboring bilayers insig-
nificant. The angular separation between the two neighboring
Fe moments along the a-axis direction in ͑PANI͒0.16FeOCl
͑360°/3ϭ120°͒ is larger than that in FeOCl (2ϫ360°/7
Ϸ102.85°), showing a stronger AFMSE interaction for the
former than for the latter. This observation can be under-
stood, since in ͑PANI͒0.16FeOCl the a-axis Fe-O-Fe bond
angle ͑ϳ104.8°͒ is 1.3% wider and the Fe-Fe are 1.5%
FIG. 8. Proposed magnetic structure for the Fe spins in
(PANI)0.16FeOCl at low temperatures. The two neighboring Fe mo-
ments along the a-axis direction are 120° apart, whereas that along
the b axis are antiparallel.
Fe-Fe separation ͑ϳ3.290 Å at 10 K͒. A stronger FMDE
interaction and a weaker AFMSE interaction hence result.
Competition between these two types of couplings then gives
rise to a noncollinear spin arrangement, so that the two
neighboring moments are 2ϫ360°/7Ϸ102.85° apart. The
coupling along the c-axis direction is crystallographically
separated into intrabilayer and interbilayer interactions. The
intrabilayer interaction is similar to that along the a-axis di-
rection, but has a more compact Fe-O-Fe bonding, with a
bond angle of ϳ99.6° and a Fe-Fe separation of ϳ3.092 Å at
10 K. A smaller angular separation between the two neigh-
boring moments along the a-axis direction can be expected,
but cannot be uniquely resolved from the present data. Most
likely, the interbilayer coupling is dominated by the dipolar
interaction, since the neighboring bilayers are bonded via a
weak Van der Waals interaction, rendering the superex-
change interaction insignificant.
Figure 6 shows the temperature dependence of the mag-
netic peak at 2␪ϭ23.2°, where the circles and the triangles
mark, respectively, the data taken without ͑energy-
integrated͒ and with ͑elastic͒ an analyzer crystal. At low tem-
peratures, there is essentially no difference between the data
taken with or without an analyzer crystal, which reveals a
typical order-parameter curve for polycrystalline samples.
The ordering temperature of the Fe spins, as determined by
the inflection point of the order-parameter curve, is TN
Ϸ80 K. Above TN , the two curves are obviously different,
indicating that there is a substantial contribution from dy-
namical spin fluctuations which exceeds the energy resolu-
FIG. 9. Variation of the ͑1/3 1/2͒ peak inten-
sity with temperature, showing a TNϷ80 K for
the Fe spins in (PANI)0.16FeOCl.

PRB 62
SPIRAL MAGNETIC STRUCTURE OF Fe IN VAN DER . . .
14 163
further apart ͑ϳ3.339 Å at 10 K͒, as compared to that in
FeOCl. Apparently, the spin arrangement along the b-axis
direction is not affected by the intercalation reaction that
increases the Fe-O-Fe bond angle ͑ϳ149.2°͒ by 0.2% and the
Fe-Fe separation ͑ϳ3.822 Å at 10 K͒ by 1.5%, along this
axis direction. The ordering temperature, on the other hand,
was not significantly affected by the appearance of polyani-
line in the gaps, as seen in the temperature dependence of the
͑1/3 1/2͒ peak intensity, shown in Fig. 9, indicating a TN
Ϸ80 K for the Fe spins in ͑PANI͒0.16FeOCl as well. It is the
intrabilayer couplings that determine the ordering tempera-
ture, as expected. The saturated moment obtained for the Fe
ions in ͑PANI͒0.16FeOCl is 3.98(7)␮B , which is essentially
the same value as obtained for the Fe ions in FeOCl.
dimensional long-range ordering of the Fe spins in FeOCl
develops below 80 K, where the moment becomes saturated
at below ϳ20 K. Competition between the antiferromagnetic
Fe-O-Fe superexchange coupling and the ferromagnetic
Fe-Fe direct exchange coupling gives rise to a noncollinear
magnetic structure for the Fe spins. The arrangement of the
Fe moments is controlled by the Fe-O-Fe bond angle. An
anisotropic Fe-O-Fe network results in different configura-
tions for spins along the three crystallographic directions.
Dipolar interactions, across the Van der Waals gaps, are also
necessary for 3D long-range ordering to occur. This interbi-
layer interaction, however, becomes insignificant in
͑PANI͒0.16FeOCl, where
a coupled-bilayer quasi-two-
dimensional order of the Fe spins was observed. The exis-
tence of polyaniline in the gaps interrupts, rather than en-
hances, the magnetic interactions between the neighboring
(Fe2O2Cl2)n bilayers, indicating that there is no significant
magnetic interaction between the polyaniline and the
(Fe2O2Cl2)n lamellas.
IV. CONCLUSION
The crystal and magnetic structures of layered FeOCl
have been investigated in detail. The iron-oxychloride FeOCl
structure consists of Q2D charge-neutral (Fe2O2Cl2)n lamel-
las linked via Van der Waals interactions. This Van der
Waals gaped Q2D nature makes FeOCl a good host for in-
tercalation reactions. The basic (Fe2O2Cl2)n lamellas con-
sisted of edge-sharing distorted FeCl2O4 octahedrons form-
ing a bilayered anisotropic Fe-O-Fe network. A three-
ACKNOWLEDGMENTS
This work at NCU was supported by the National Science
Council of the Republic of China under Grant No. NSC 89-
2112-M-008-051.
1 S. M. Kauzlarich, J. L. Stanton, J. Faber, Jr., and B. A. Averill, J.
Am. Chem. Soc. 108, 7946 ͑1986͒.
tingham and A. J. Jacobson ͑Academic, New York, 1982͒, p.
377.
2 R. S. Bannwart, J. E. Phillips, and R. H. Herber, J. Solid State
Chem. 71, 540 ͑1987͒.
16 H. M. Rietveld, J. Appl. Crystallogr. 2, 65 ͑1969͒.
17 The Rietveld Method, edited by R. A. Young ͑Oxford University
Press, New York, 1993͒.
3 S. M. Kauzlarich, B. K. Teo, B. A. Averill, Inorg. Chem. 25,
1209 ͑1986͒.
4 M. G. Kanatzidis, L. M. Tonge, T. J. Marks, H. O. Marcy, and C.
R. Kanaewurf, J. Am. Chem. Soc. 109, 3797 ͑1987͒.
5 M. G. Kanatzidis, C.-G. Wu, H. O. Marcy, D. C. DeGroot, and C.
R. Kannewurf, Adv. Mater. 2, 364 ͑1990͒.
18 A. C. Larson and R. B. Von Dreele, Los Alamos National Labo-
ratory, Report No. LA-UR-86-748, 1990 ͑unpublished͒.
19 J. Rouxel and P. Palvadeau, Rev. Chim. Miner. 19, 317 ͑1982͒.
20 J.-H. Choy, J.-B. Yoon, D.-K. Kim, and S.-H. Hwang, Inorg.
Chem. 34, 6524 ͑1995͒.
6 C.-G. Wu, D. C. DeGroot, H. O. Marcy, J. L. Schindler, C. R.
Kannewurf, T. Bakas, V. Papaefthymiou, W. Hirpo, J. P.
Yesinowski, Y.-J. Liu, and M. G. Kanatzidis, J. Am. Chem. Soc.
117, 9229 ͑1995͒.
21 W. A. Dollase, J. Appl. Crystallogr. 19, 267 ͑1986͒.
22 A. March, Z. Kristallogr. 81, 285 ͑1932͒.
23 M. D. Lind, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst.
Chem. 20, 1058 ͑1970͒.
7 Z. Takehara, K. Kanamura, N. Imanishi, and C. Zhen, Bull.
Chem. Soc. Jpn. 62, 1567 ͑1989͒.
24 G. E. Bacon, Neutron Diffraction, 3rd ed. ͑Clarendon, Oxford,
1975͒.
8 T. R. Halbert and J. Scanlon, Mater. Res. Bull. 14, 415 ͑1979͒.
9 F. Kanawaru, and M. Koizumi, Jpn. J. Appl. Phys., Part 1 13,
1319 ͑1974͒.
25 O. V. Kovalev, Irreducible Representations of the Space Groups
͑Gordon and Breach, New York, 1965͒.
26
¨
Yu. A. Izyumov, V. E. Zaish, R. P. Ozerov, and Joachim Buch-
10 Z. Takehara, H. Sakaebe, and K. Kanamura, J. Power Sources
43–44, 627 ͑1993͒.
ner, Neutron Diffraction of Magnetic Materials ͑Consultants Bu-
reau, New York, 1991͒.
11 H. Sakaebe, S. Higushi, K. Kanamura, and H. Fujimoto, J. Power
Sources 56, 165 ͑1995͒.
27 J. B. Goodenough, Phys. Rev. 117, 1442 ͑1960͒.
28 E. O. Wollan, Phys. Rev. 117, 387 ͑1960͒.
29 P. W. Anderson, Phys. Rev. 79, 350 ͑1950͒.
30 B. E. Warren, Phys. Rev. 59, 693 ͑1941͒.
31 J. K. Kjems, L. Passell, H. Taub, J. G. Dash, and A. D. Novaco,
Phys. Rev. B 13, 1446 ͑1976͒.
12 S. Kikkawa, F. Kanawaru, and M. Koizumii, Bull. Chem. Soc.
Jpn. 52, 963 ͑1979͒.
13 K. Prassides, C. J. Bell, A. J. Dianoux, C.-G. Wu, and M. G.
Kanatzidis, Physica B 180&181, 668 ͑1992͒.
14 A. Adam and G. Buisson, Phys. Status Solidi A 30, 323 ͑1975͒.
15 M. R. Halbert, in Introduction Chemistry, edited by M. S. Whit-
32 Details can be found in H. Zhang, J. W. Lynn, and D. E. Morris,
Phys. Rev. B 45, 10 022 ͑1992͒.