Evidence of spin-density-wave order in RFeAsO
1-x
F
x
from measurements
of thermoelectric power
M. Matusiak,
1
T. Plackowski,
1
Z. Bukowski,
2
N. D. Zhigadlo,
2
and J. Karpinski
2
1
Institute of Low Temperature and Structure Research, Polish Academy of Sciences, P.O. Box 1410, 50-950 Wroclaw, Poland
2
Laboratory for Solid State Physics, ETH Zurich, 8093 Zurich, Switzerland
Received 16 December 2008; revised manuscript received 30 January 2009; published 2 June 2009
Data on the magnetothermopower and specific heat of three compounds belonging to “1111” oxypnictides
family are reported. One specimen SmAsFeO
0.8
F
0.2
is a superconductor with T
c
=53 K, while two others
SmAsFeO and NdAsFeO are nonsuperconducting parent compounds. Our results confirm that spin-density-
wave SDW order is present in SmAsFeO and NdAsFeO. In these two samples a strict connection between the
thermoelectric power and electronic specific heat is found in the vicinity of SDW transition, which indicates
that the chemical potential of charge carriers strongly depends on temperature in this region. Low-temperature
data suggest presence of significant contribution magnon drag to the thermoelectric power.
DOI: 10.1103/PhysRevB.79.212502 PACS numbers: 74.25.Fy, 74.70.-b, 72.15.Jf, 75.40.Cx
Intense investigations of properties of the rare-earth iron
oxypnictides reveal some similarities between their phase
diagram and that of the cuprate superconductors. Namely, in
the cuprates superconductivity emerges when mobile elec-
trons or holes are doped into antiferromagnetic parent
compounds
1,2
and an analogous behavior is observed in the
iron-based superconductors where an electron doping seems
to suppress the spin-density-wave SDW instability allow-
ing superconductivity to appear.
3,4
However, there were
some doubts raised, based on results of neutron-diffraction
studies, whether this behavior is common for all iron
oxypnictides.
5
Clarifying this uncertainty can be important in
determining a role that is played by magnetic interactions in
the mechanism of superconductivity. Since there is no obvi-
ous way to distinguish between itinerant and localized
magnetism
6
we utilize an experimental technique that allows
the study of magnetic ordering of charge carriers. In order to
do it we studied the specific heat and thermoelectric power
TEPmeasured in the magnetic field up to 13 T of doped
and undoped oxypnictides including a NdFeAsO compound
that caused above mentioned doubts.
Polycrystalline samples of NdFeAsO and SmFeAsO were
prepared by conventional solid-state reaction. First, NdAs,
SmAs, and FeAs were synthesized from pure elements in
evacuated silica ampoules at 600 ° C. In the next step stoi-
chiometric amounts of NdAs or SmAs and FeO were
weighed and thoroughly mixed. The raw mixtures were
pressed into pellets, wrapped in Ta foil, and sealed in silica
tubes under reduced pressure of Ar gas. The pellets were
heated at 1160 ° C for 40 h with intermittent regrinding and
pelletizing. A high-density sample of SmFeAsO
0.8
F
0.2
was
prepared by a high-pressure and high-temperature method
using a cubic anvil apparatus.
7
The stoichiometric mixture of
SmAs, FeAs, Fe, Fe
2
O
3
, and SmF
3
was placed in a BN con-
tainer inside a pyrophyllite cube equipped with a graphite
heater. The compound was synthesized at a pressure of 3
GPa and temperature 1350 ° C for 4.5 h. The phase purity of
the obtained samples was checked by means of powder x-ray
diffraction XRD carried out on a STOE diffractometer us-
ing Cu K radiation and a graphite monochromator. The
XRD patterns of NdFeAsO and SmFeAsO samples showed
no detectable amount of impurities while SmFeAsO
0.8
F
0.2
contained some amount of Sm oxyfluoride. The lattice pa-
rameters calculated from XRD data were a =3.965 Å and
c = 8.575 Å for NdFeAsO, a =3.937 Å and c =8.500 Å for
SmFeAsO, and a =3.927 Å and c =8.461 Å for
SmFeAsO
0.8
F
0.2
.
The magnetic field in the magnetothermopower measure-
ment was parallel to the temperature gradient. A sample was
clamped between two spring-loaded copper blocks provided
with heaters and a pair of thermometers Cernox 1050. The
blocks were thermally insulated from the surrounding, there-
fore a thermal difference of any sign might be produced by
the heaters. The voltage difference between blocks was mea-
sured using an A20 EM Electronics low-noise preamplifier.
More details about the method can be found in Ref. 8.
The specific heat C
p
was measured using a heat-flow
calorimeter.
9
In this method the sample is connected with a
heat sink by means of a sensitive heat-flow meter of high
thermal conductance. To sense the heat flux we used a com-
mercial miniature one-stage Peltier cell with sensitivity of
0.45 V/W at room temperature and 0.08 V/W at liquid nitro-
gen temperature. The sample was fixed on the cell top plate
made of 0.5-mm-thick alumina. The bottom of the heat-flow
meter was permanently attached to the heat sink a massive
copper block of controlled temperature. An in-field cali-
brated Pt thermometer was attached to the sink. Such a de-
vice was surrounded by a double passive radiation screen
gold plated. Both screens were in a good thermal contact
with the sink. The whole ensemble was evacuated down to
10
-6
hPa and placed in the gas-flow variable-temperature
insert of an Oxford cryostat with a 13/15 T superconducting
magnet.
First we present data on a temperature dependence of the
specific heat C
p
of the superconducting SmAsFeO
0.8
F
0.2
sample Fig. 1a accompanied by C
p
T dependences mea-
sured for two nonsuperconducting parent compounds:
SmAsFeO Fig. 1b and NdAsFeO Fig. 1c. In the fluo-
rine doped SmAsFeO
0.8
F
0.2
sample, we see a kink at the T
= 53 K that is related to the formation of the superconduct-
ing state. An application of the magnetic field of 13 T almost
completely smears the kink out while a hump near T
SDW
140 K in both parent compounds is resistant to an influ-
ence of the magnetic field. This anomaly can be associated
PHYSICAL REVIEW B 79, 212502 2009
1098-0121/2009/7921/2125024 ©2009 The American Physical Society 212502-1