Influence of Cr doping on the magnetic structure of the FeAs-strips compound CaFe_{4}As_{3}: A single-crystal neutron diffraction study

Abstract

We have studied the magnetic structure of a Cr-doped iron-arsenide compound CaFe_{4}As_{3} by means of single crystal neutron diffraction. The neutron data reveal that below 90 K, an antiferromagnetic structure with propagation vector k=0 is adopted. Refinement of the magnetic structure using one of the modes allowed by symmetry analysis indicates that two of the four Fe sites, including the one where the selective substitution Fe/Cr happens, bear reduced magnetic moments. Density functional theory calculations confirm the stability of such a magnetic arrangement.

Full text

PHYSICAL REVIEW B 88, 104414 (2013)
Influence of Cr doping on the magnetic structure of the FeAs-strips compound CaFe4As3:
A single-crystal neutron diffraction study
P. Manuel,1L. C. Chapon,1,2G. Trimarchi,3I. S. Todorov,4D. Y. Chung,4B. Ouladdiaf,2M. J. Gutmann,1
A. J. Freeman,3and M. G. Kanatzidis4,5
1ISIS facility, STFC Rutherford Appleton Laboratory, Chilton, Didcot, Oxfordshire OX11 0QX, United Kingdom
2Institut Laue-Langevin, 6 Rue Jules Horowitz, BP 156, 38042 Grenoble Cedex 9, France
3Department of Physics and Astronomy, Northwestern University, Evanston, Illinois 60208, USA
4Materials Science Division, Argonne National Laboratory, Argonne, Illinois 60439, USA
5Department of Chemistry, Northwestern University, Evanston, Illinois 60208, USA
(Received 14 June 2013; revised manuscript received 12 August 2013; published 17 September 2013)
We have studied the magnetic structure of a Cr-doped iron-arsenide compound CaFe4As3by means of
single crystal neutron diffraction. The neutron data reveal that below 90 K, an antiferromagnetic structure
with propagation vector k=0 is adopted. Refinement of the magnetic structure using one of the modes allowed
by symmetry analysis indicates that two of the four Fe sites, including the one where the selective substitution
Fe/Cr happens, bear reduced magnetic moments. Density functional theory calculations confirm the stability of
such a magnetic arrangement.
DOI: 10.1103/PhysRevB.88.104414 PACS number(s): 75.10.−b, 75.25.−j
High temperature superconductivity (SC) research was
given a recent boost following the discovery of supercon-
ductivity at 26 K in noncopper based materials.1Perhaps
the most striking similarity between these new iron-based
layered compounds, such as La[O1−xFx]FeAs, and the high
temperature cuprate (HTC) superconductors is the presence
of square planar (Fe2As2or CuO2) superconducting layers.
By contrast, the nature of the parent compounds, which can
be tuned towards SC by doping,2,3or by applying pressure,
is very different for the new iron arsenide and the HTC
superconductors: the former are spin density wave (SDW)
metals while the latter are insulators. Clearly, the ability to
change the topology of the Fe2As2layers and to perturb
the SDW state is important to understand the electronic
correlations in the new pnictides. These effects can be both
realized in CaFe4As3where, instead of forming infinite layers,
the FeAs are arranged in strips along bconnected in a
rectangular cross pattern,4as shown in Fig. 1, and where Cr
doping affects the SDW state.
The undoped nonsuperconducting systems, including the
122 compounds AFe2As2(where A=Ba,Sr,Ca), evolve from
a tetragonal to an orthorhombic structure and eventually order
magnetically with commensurate antiferromagnetic (AFM)
structure, bearing ordered moments less than 1μB.5For
CaFe4As3two transitions at TN1≈90 and TN2≈26 K
are seen in magnetic susceptibility data,4,6heat capacity
measurements,6and recent Hall effect, thermopower mea-
surements and M¨
ossbauer spectroscopy.7–9In a previous
publication10 on a powder sample of CaFe4As3we identified
the second order transition at TN1with an incommensurate
(ICM) longitudinal spin-density wave along the baxis with
propagation vector k=(0,δ,0) and 0.375 <δ<0.39 and the
transition at TN2with a locking into a commensurate (CM)
state with δ=3
8where the moments remained predominantly
aligned along the baxis. A later single crystal study11 refined
the low temperature commensurate structure by mixing two
irreducible representations, revealing the existence of a small
additional component in the ac plane.
A profound effect on the magnetic properties of CaFe4As3
can be obtained by both chemical doping and pressure.12,13 For
both hydrostatic pressure and P/Yb doping on the nonmagnetic
As/Ca sites, the SDW and the ICM-CM transitions appear to be
preserved; this is also the case for Cu-doping on the Fe sites.13
By contrast doping on the magnetic Fe sites by either Co
or Cr retains the higher transition temperature but suppresses
the ICM-CM transition although in the former case, a third
transition only visible in the derivative of the magnetization
appears. These studies point towards a fairly robust SDW
order. To date there has not been any magnetic structure
determination on any of these doped compounds. In this paper,
we report on the magnetic structure of CaCr0.86Fe3.14 As3
determined by single crystal neutron diffraction.
Neutron experiments were performed at the Institut
Laue Langevin (ILL) on a needle shape single crystal of
CaCr0.86Fe3.14 As3. A tin flux grown sample (12 ×1×1mm)
was prealigned using the OrientExpress facility and placed in a
cryostat on the new neutron Laue diffractometer CYCLOPS.14
This allowed us to quickly survey a very large portion of
reciprocal space and thereby identify the magnetic propagation
vector before a more quantitative study on a conventional
four-circle diffractometer (D10, ILL) using the same sample,
as well as to perform a rapid temperature dependence. The
D10 data were collected with an incident neutron wavelength
of λ=2.36 ˚
A by using an 80 mm2two-dimensional microstrip
detector and normalized by monitor. Integration of the Bragg
peaks was obtained by a two stage process using the program
RACER (ILL). Initially, a library was built by fitting ellipsoidal
shapes to a set of strong reflections (with intensities five times
higher than the background). The integration of all reflections
was then obtained in a second pass using this library. For each
data set, the list of integrated intensities obtained was corrected
for Lorentz factors and normalized to the monitor count. A
set of 186 independent reflections, measured at 150 K, was
used to refine the nuclear structure in the paramagnetic phase.
To refine the magnetic structure, 212 independent reflections
were collected at 2 K. For both data sets, removing the (020)
104414-1
1098-0121/2013/88(10)/104414(4) ©2013 American Physical Society

P. MANUEL et al. PHYSICAL REVIEW B 88, 104414 (2013)
FIG. 1. (Color online) Crystal (left) and magnetic (right) struc-
tures of CaCr0.86Fe3.14 As3at 1.5 K. The (Fe/Cr)As polyhedra, with
the four (4c) Fe or Cr sites sitting in the middle, have been drawn to
highlight the ribbon nature of the structure.
reflection considerably improved the refinements by reducing
the extinction effect due to the needle shape of the sample
growing along the bdirection. Refinement of both the nuclear
and magnetic structures was carried out using the FullProf
program15 and symmetry (representation) analysis is presented
using Kovalev’s notation.16
Refinements in the paramagnetic phase, collected at 150 K
and consistent with a previous report,12 indicate that the
structure crystallizes in the orthorhombic space group Pnma
with a=12.16(1), b=3.766(4), and c=11.80(1) ˚
A. The
refinement gives RBragg =3.39% and the positions (x,1
4,z)
of the four independent (4c) Fe sites in the crystallographic
unit cell are given in Table I.The(4c)sites2,3,and4are
fully occupied while Cr substitutes on site 1 with a refined
occupancy of 85.6(6)% comparable with the 84% obtained on
a previous sample.12
We now turn our attention to the low temperature (1.5 K)
magnetic structure. As can be seen immediately from Fig. 2,
the magnetic structure is commensurate with k=0 in con-
trast with the parent compound CaFe4As3. The decompo-
sition of the magnetic representation for the (4c) sites is
given by
τmag(4c)=τ1+2τ2+2τ3+τ4+τ5+2τ6+2τ7+τ8.
All the irreducible representations (irreps) are one dimen-
sional. The τ2,τ3,τ6, and τ7irreps correspond to ordering in
the ac plane which is not consistent with the susceptibility.12
TABLE I. Amplitude of the magnetic moments (Mi) and relative
phases (i) where appropriate for the four inequivalent Fe sites (i=
1–4) extracted from the refinements at 1.5 K for CaCr0.86Fe3.14 As3
(left) and for CaFe4As3(right) from Ref. 10.
Doped 1.5K Undoped 30 K/1.5 K
Site iPosition Mi(μB)Mi(μB)i
Fe1x=0.0215(4) −0.48(6) 2.14(13)/0/
z=0.3141(4) 1.4(11) 0
Fe2x=0.0661(2) 0.46(5) 1.55(16)/0.13(2)/
z=0.5383(2) 1.61(14) 0.14(3)
Fe3x=0.3051(2) 1.71(5) 1.83(8)/0.56(4)/
z=0.1221(2) 1.67(20) 0.45(3)
Fe4x=0.3155(2) 1.58(5) 1.94(10)/0.10(4)/
z=0.7231(2) 1.84(10) 0.01(4)
FIG. 2. (Color online) Neutron Laue patterns obtained on CY-
CLOPS. For both patterns, the horizontal coverage is 225◦and the
vertical coverage is 90◦. Top: In the paramagnetic regime at 200 K.
Bottom: In the magnetically ordered state at 1.5 K. The arrows
highlight four peaks with strong intensities in the low temperature
data and which can all be indexed with k=0.
For each of the four (4c) sites, the moment directions for the
atomic positions (x,1/4,z), (−x,3/4,−z), (−x+1/2,3/4,z +
1/2), and (x+1/2,1/4,−z+1/2) along the baxis of the
remaining four irreps are arranged as follows: ++−−for τ1,
+−+− for τ4,++++ for τ5, and +−−+ for τ8.Using
Bertaut’s well known notation,17 this sequence can be rewritten
as Cy,Gy,Fy, and Ay. Refinement of the low temperature
data very clearly favors τ4with the Rffactor being two to
three times better and χ2a factor 7 to 9 times better. This
corresponds to the magnetic space group Pn
ma. The final
refinement, with the moments on site 1 tied for the Cr and
Fe, is presented in Fig. 3and gives Rf=3.92%. Moment
FIG. 3. (Color online) Experimental (data from D10) versus
calculated structure factors for CaCr0.86Fe3.14 As3at 1.5 K. The tight
distribution of black dots around the red line F2
obs =F2
calc indicates
the high quality of the refinement. Inset: Temperature dependence of
(310) peak, obtained from CYCLOPS.
104414-2

INFLUENCE OF Cr DOPING ON THE MAGNETIC ... PHYSICAL REVIEW B 88, 104414 (2013)
values extracted from the refinement are presented in Table I.
Interestingly, the magnetic structure drawn in Fig. 1bears some
similarity with that of the pure compound CaFe4As3apart
from the fact that the structure now repeats itself along the
bdirection as it is k=0 and no longer modulated along b.
Indeed, moments on sites 3 and 4 are slightly reduced while
the moment on the five-coordinated site 1 is much reduced
[0.48(6)μB] like in the case of the undoped compound in the
incommensurate phase. However, for the Cr-doped compound,
the moment on site 2 is also reduced [0.46(5)μB]. It is to be
noted that both sites 1 and 2 with reduced magnetic moments
correspond to the anchoring points between the Fe2As2
ribbons.
The temperature dependence of the (310) peak, obtained
by integrating 74 runs from CYCLOPS is presented in
the inset of Fig. 3. The extracted transition temperature of
100 K is consistent with previous magnetic susceptibility
reports.12
To shed further light on the magnetic structure of
CaCr0.86Fe3.14 As3we have used density functional theory
(DFT). The random distribution of Cr and Fe atoms on
the Fe1special site in CaCr0.86Fe3.14 As3results in magnetic
moments at the transition metal sites that are distributed
around the average values detected by the neutron diffraction
measurements. Ab-initio calculations allow one to resolve the
magnetic moments on each transition metal site and to relate
them to the local distribution of Fe and Cr atoms and to their
distance from the As atoms. We performed density functional
calculations by using the projected augmented-wave plane-
wave method as implemented in the VASP package,18–20
with the Perdew, Burke, and Ernzerhof generalized-gradient
approximation21 to the exchange and correlation functional.
The lattice vectors and atomic positions were kept fixed to
their experimental values reported in Table I. In order to model
the mixed occupancy by Fe and Cr on the fivefold coordinated
atomic site, two supercells with 1 ×2×1 and 1 ×4×1 pe-
riodicity obtained by, respectively, doubling and quadrupling
the primitive cell along the baxis were used. These supercells
enable us to model Cr-doped CaFe4As3with a Cr content
corresponding to a CaCr0.875Fe3.125 As3composition which
closely approximates the actual composition of the sample
studied. Three configurations with unique arrangements of
Fe and Cr atoms on the Fe1special site can be produced
in these supercells. The self-consistent electronic structure
calculations for each of these atom arrangements converged
to a magnetic configuration with the Fe atoms on adjacent Fe2,
Fe3, and Fe4sites having magnetic moments parallel to each
other and antiparallel to the magnetic moment of the Cr or
Fe atoms occupying the Fe1site. The AFM configuration is
lower in energy than the ferromagnetic one by 18 meV/at. The
nonmagnetic configuration which reasonably approximates
the paramagnetic phase is higher in energy than the AFM one
by 132 meV/at. Thus, the calculations reproduced the AFM
ordering observed in the experiment and shown in Fig. 1, right
panel. In all configurations, the amplitude of the magnetic
moments of the iron atoms on the Fe3sites are comprised
between 2.28μBand 2.33μB, while on the Fe4sites they are
between 2.16μBand 2.24μB. The magnetic moments on the
Fe2sites have an amplitude at least 0.3μBsmaller than those
on the Fe3and Fe4sites, but with a larger spread, nominally
FIG. 4. (Color online) Density of states projected on the iron
atomic dorbitals for all the Fe sites in the 1 ×2×1 supercell.
between 0.78μBand 1.93μB. Therefore, the significant re-
duction of the magnetic moment at the Fe2sites observed
in the experiment is reproduced in the electronic structure
calculations. We note, however, that this is not the case for the
experimentally observed reduced magnetic moment on the Fe1
site where the calculations give values of the order of 2.3μB.
Figure 4shows the projected density of states onto dorbitals of
the Fe atoms at the Fe1,Fe
2,Fe
3, and Fe4sites in the 1 ×2×1
supercell. At the Fe3and Fe4sites, the spin up Fe dorbitals
are fully occupied while the density of states in the spin down
channel at similar energies is much smaller. Such an electronic
configuration results in the large magnetic moment obtained
for these sites. The total density of states, not shown here, is
large at the Fermi energy which is consistent with the metallic
104414-3

P. MANUEL et al. PHYSICAL REVIEW B 88, 104414 (2013)
behavior observed in this compound. By contrast, at the Fe2
site, the amplitude of the magnetic moment is much reduced
owing to the similar density of states of spin up and spin down
dstates below the Fermi level.
In summary, we have determined the magnetic structure of
CaCr0.86Fe3.14 As3through single crystal neutron diffraction.
The resulting structure is no longer incommensurate as for
the undoped CaFe4As3sample but a k=0 antiferromagnet is
attained instead. Moments on the sites linking the rectangular
cross pattern of Fe2As2ribbons are reduced. This picture is
globally consistent with DFT calculations which show the
lowest energy state as an AFM state with moments reduced on
one of the linking sites. Further band structure calculations to
determine what happens to the Fermi surface, which appeared
to possess a clear nesting vector of 3b∗/8 equal to the magnetic
propagation vector in the undoped sample, are required to
fully understand the effect of doping. Indeed, our present
results should also be placed in the general context of the iron
arsenides, by discussing the relative importance of the itinerant
versus the local moment description for the magnetism where
the Fermi surface nesting dominates on the one hand and
the exchange interactions on the other. Further experimental
studies with different and smaller doping should also prove
valuable in that respect.
We thank the Science and Technology Facilities Coun-
cil for providing the neutron beam time at the ILL and
Marek Jura for assistance with preliminary x-ray diffrac-
tion (XRD) data. The work at Argonne National Lab-
oratory was supported by the US Department of En-
ergy, Office of Basic Energy Sciences, under Contract
No. DE-AC02-06CH11357.
1Y. Kamihara, T. Watanabe, M. Hirano, and H. Hosono, J. Am.
Chem. Soc. 130, 3296 (2008).
2R.Zhi-An,L.Wei,Y.Jie,Y.Wei,S.Xiao-Li,L.Zheng-Cai,
C. Guang-Can, D. Xiao-Li, S. Li-Ling, Z. Fang et al.,Ch. Phys.
Lett. 25, 2215 (2008).
3M. Rotter, M. Tegel, and D. Johrendt, Phys.Rev.Lett.101, 107006
(2008).
4I. Todorov, D. Y. Chung, C. D. Malliakas, Q. Li, T. Bakas,
A. Douvalis, G. Trimarchi, K. Gray, J. F. Mitchell, A. J. Freeman,
and M. G. Kanatzidis, J. Am. Chem. Soc. 131, 5405 (2009).
5Q. Huang, Y. Qiu, W. Bao, M. A. Green, J. W. Lynn, Y. C.
Gasparovic, T. Wu, G. Wu, and X. H. Chen, Phys. Rev. Lett. 101,
257003 (2008).
6L. L. Zhao, T. Yi, J. C. Fettinger, S. M. Kauzlarich, and E. Morosan,
Phys.Rev.B80 020404(R) (2009).
7A. B. Karki, G. T. McCandless, S. Stadler, Y. M. Xiong, J. Li,
J. Y. Chan, and R. Jin, Phys.Rev.B84, 054412 (2011).
8M. S. Kim, Z. P. Yin, L. L. Zhao, E. Morosan, G. Kotliar, and M. C.
Aronson, Phys.Rev.B84, 075112 (2011).
9I. Nowik, I. Felner, A. B. Karki, and R. Jin, Phys.Rev.B84 212402
(2011).
10P. Manuel, L. C. Chapon, I. S. Todorov, D. Y. Chung, J. P. Castellan,
S. Rosenkranz, R. Osborn, P. Toledano, and M. G. Kanatzidis, Phys.
Rev. B 81 184402 (2010).
11Y. Nambu, L. L. Zhao, E. Morosan, K. Kim, G. Kotliar, P. Zajdel,
M. A. Green, W. Ratcliff, J. A. Rodriguez-Rivera, and C. Broholm,
Phys. Rev. Lett. 106, 037201 (2011).
12I. Todorov, D. Y. Chung, H. Claus, K. E. Gray, Q. Li, J. Schleuter,
T. Bakas, A. P. Douvalis, M. J. Gutmann, and M. G. Kanatzidis,
Chem. Mat. 22, 4996 (2010).
13L.L.Zhao,S.K.Kim,G.T.McCandless,M.S.Torikachvili,
P. C. Canfield, J. Y. Chan, and E. Morosan, Phys.Rev.B84, 104444
(2011).
14B. Ouladdiaf, J. Archer, J. R. Allibon, P. Decarpentrie, M-H. Lemee-
Cailleau, J. Rodriguez-Carvajal, A. W. Hewatt, S. York, D. Brau,
and G. J. McIntyre, J. Appl. Cryst. 44, 392 (2011).
15J. Rodriguez Carvajal, Physica B 192, 55 (1993).
16O. V. Kovalev, Representations of the Crystallographic Space
Groups, 2nd ed. (Gordon and Breach Science Publishers, Switzer-
land, 1993).
17E. W. Bertaut, Acta Cryst. A 24, 217 (1968).
18G. Kresse and J. Hafner, Phys.Rev.B47, 558 (1993).
19G. Kresse and J. Furthmuller, Phys. Rev. B 54, 11169
(1996).
20G. Kresse and D. Joubert, Phys. Rev. B 59, 1758
(1999).
21J. P. Perdew, K. Burke, and M. Ernzerhof, Phys.Rev.Lett.77, 3865
(1996).
104414-4