arXiv:0811.4598v1 [cond-mat.supr-con] 27 Nov 2008
Two-dimensional magnetism in the pnictide superconductor parent material
SrFeAsF probed by muon-spin relaxation
P. J. Baker,1I. Franke,1T. Lancaster,1S. J. Blundell,1L. Kerslake,2and S. J. Clarke2
1Oxford University Department of Physics, Clarendon Laboratory,
Parks Road, Oxford OX1 3PU, United Kingdom
2Department of Chemistry, University of Oxford, Inorganic Chemistry Laboratory,
South Parks Road, Oxford, OX1 3QR, United Kingdom
(Dated: November 27, 2008)
We report muon-spin relaxation measurements on SrFeAsF, which is the parent compound of a
newly discovered iron-arsenic-fluoride based series of superconducting materials. We find that this
material has very similar magnetic properties to LaFeAsO, such as separated magnetic and structural
transitions (TN = 120 K, Ts = 175 K), contrasting with SrFe2As2 where they are coincident. The
muon oscillation frequencies fall away very sharply at TN, which suggests that the magnetic exchange
between the layers is weaker than in comparable oxypnictide compounds. This is consistent with
our specific heat measurements, which find that the entropy change ∆S = 0.05 Jmol−1K−1largely
occurs at the structural transition and there is no anomaly at TN.
PACS numbers: 76.75.+i, 74.10.+v, 75.30.Fv, 75.50.Ee
Quasi-two-dimensional magnets on square lattices are
the subject of considerable theoretical and experimental
attention.1,2,3This has primarily been due to the success
of models of the spin-1/2 Heisenberg antiferromagnet in
describing the physics of La2CuO4, which is the proto-
typical parent compound of high-Tc cuprate supercon-
ductors. La2CuO4 shows a tetragonal to orthorhombic
structural transition at To≃ 530 K and N´ eel ordering at
TN≃ 325 K. That TNis far smaller than the antiferro-
magnetic exchange constant J ∼ 1500 K demonstrates
that this compound has a remarkably large magnetic
anisotropy, with weak coupling between the CuO2 lay-
ers.2,4The magnetic parent compounds of FeAs-based su-
perconductors such as LaFeAsO1−xFx5have Fe atoms on
a layered square lattice, and it is interesting to note that,
like La2CuO4, these have a tetragonal to orthorhombic
structural distortion followed by antiferromagnetic order-
ing (e.g. Ref. 6). Here we study the magnetic proper-
ties of a newly discovered parent compound to a series
of fluoropnictide superconductors, SrFeAsF,7,8where the
fluoride ions should provide weaker magnetic exchange
pathways between the FeAs layers than for LnFeAsO or
Doped fluoropnictide compounds based on CaFeAsF
and SrFeAsF have recently been found to supercon-
duct,9,10,11,12,13with comparable transition temperatures
to the previously discovered oxypnictide compounds
based on LnFeAsO. These have similar FeAs layers to
the oxypnictides, but divalent metal - fluoride layers re-
place the rare-earth - oxide layers. Fluoropnictides can
be doped on the Fe site, as for CaFe0.9Co0.1AsF (Tc=
22 K),9or the divalent metal site, as for Sr0.5Sm0.5FeAsF
(Tc= 56 K),10and several approaches have already been
explored.9,10,11,12,13The magnetic, electronic, and struc-
tural properties of the parent compounds CaFeAsF and
SrFeAsF, and also EuFeAsF, have already been investi-
gated. All three show transitions evident in resistivity
and dc magnetization measurements, at Ts= 120, 175,
and 155 K respectively.7,8,9,12,13In SrFeAsF the struc-
tural transition has been probed using X-ray diffraction,
and changes in the magnetism using M¨ ossbauer spec-
troscopy.7The structural change varies smoothly below
175 K whereas the M¨ ossbauer spectra became increas-
ingly complicated as the temperature is reduced. It is
also interesting that the sign of the Hall coefficient RH
in SrFeAsF is reported to be positive below Ts, whereas
it is negative in the undoped LaFeAsO and BaFe2As2
parent compounds.8This could result from a different
electronic structure near the Fermi surface, which might
have implications for both the magnetism of the undoped
compound and the superconductivity that emerges when
it is doped.
The magnetism of LnFeAsO compounds has already
been intensively investigated by a wide range of tech-
niques. Neutron diffraction measurements have been
carried out on some of the undoped oxypnictides:
The results in each case indicate similar structural
transitions at around Ts = 150 K, followed by long
range, three dimensional antiferromagnetic ordering of
the iron spins with significantly reduced moments <
1 µB/Fe at TN, around 20 K below Ts, confirmed
by other techniques.17,18These features move to lower
temperature with increasing doping and are absent in
the superconducting phase for LaFeAsO1−xFx19and
CeFeAsO1−xFx,6although magnetism and superconduc-
tivity seem to coexist over a small doping range in
SmFeAsO1−xFx.20In constrast, SrFe2As2has coincident
magnetic and structural ordering occurring in a first-
order phase transition at To = 205 K.21It seems that
in general AFe2As2 materials have more closely related
structural and magnetic phase transitions, and more
three-dimensional magnetism than the single layer FeAs
materials. With the discovery of new fluoroarsenide par-
ent materials it is important to compare the magnetic
structures and the separation between Tsand TNin the
oxide-arsenide and fluoro-arsenidematerials. Here we ad-
dress these comparisons in SrFeAsF using the techniques
of muon-spin relaxation, which is a local probe of the
magnetic fields inside the sample and their dynamics,
and also specific heat measurements which examine the
changes in entropy at the transitions.
The SrFeAsF sample was synthesized in a two step
process similar to that described in Ref. 7. Stoichiomet-
ric quantities of sublimed strontium metal (Alfa 99.9 %),
strontium fluoride powder (Alfa 99.9 %), iron powder
(Alfa, 99.998 %), and arsenic pieces (Alfa, 99.9999 %;
ground into powder) were ground together and sealed in a
9 mm diameter niobium tube. This was heated at 1◦/min
to 500◦C and this temperature was maintained for 12
hours to ensure complete reaction of the volatile compo-
nents before heating at 1◦/min to 900◦C. After 40 hours
at 900◦C the product was removed from the Nb tube,
ground to a fine powder, pressed into a pellet, and placed
into an alumina crucible which was then sealed in a pre-
dried evacuated silica tube. This was heated at 1◦C/min
to 1000◦C for 48 hours and then cooled at the natural
rate of the furnace to room temperature. All manipula-
tion was carried out in an argon-filled glove box. Analy-
sis of the product by laboratory X-ray powder diffraction
(PANAlytical X-pert PRO) [Fig 1(a)] revealed that the
sample consisted of about 97 % by mass SrFeAsF; SrF2
was identified as a crystalline impurity phase, but no
other crystalline binary or ternary impurity phases were
identified. The refined room temperature lattice parame-
ters of SrFeAsF were a = 4.00059(3)˚ A, c = 8.9647(1)˚ A,
V = 143.478(4)˚ A3consistent with other reports.7Mea-
surement of the dc susceptibility was carried out in a
Quantum Design MPMS5 instrument [Fig. 1(b)]. The
magnetization of the sample as a function of field at 300
K showed no significant level of ferromagnetic impurity.
Measurements as a function of temperature in an applied
field of 1000 Oe revealed very similar behaviour to that
reported previously.7A broad feature at around 175 K is
consistent with the closely associated antiferromagnetic
ordering and structural phase transitions which occur in
related compounds.14,22,23Heat capacity measurements
were carried out using a Quantum Design Physical Prop-
erties Measurement System (PPMS) using a standard re-
laxation time approach. A small part of the sample used
for µSR measurements was attached to the sample plat-
form using Apiezon N-grease. Measurements were cor-
rected for the heat capacity of the sample platform and
grease. Muon-spin rotation (µSR) experiments24were
performed using the General Purpose Surface-Muon In-
strument (GPS) at the Swiss Muon Source (Paul Scherrer
Institute, Switzerland). The measured parameter is the
time-dependent muon decay asymmetry, A(t), recorded
in positron detectors on opposite sides of the sample.
Our sample was a pressed powder pellet of 1 cm diam-
eter mounted inside a silver packet on a silver backing
plate. This arrangement gives a time and temperature
independent background to the signal which is straight-
forward to subtract.
.................................................. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..................................................................................
. . . . . ..................
FIG. 1: (Color online.) (a) Rietveld refinement against pow-
der X-ray diffraction data, χ2= 1.98, wRp = 0.044. The inset
shows the structure. (b) DC magnetization measurements vs.
temperature in a field of 0.1 T. The small feature at about 50K
is probably due to a small amount of adsorbed O2 apparent
because of the small sample moment. (Inset) Magnetization
vs. field at 300 K. (c) Heat capacity C(T) showing the peak
at Ts which is highlighted in the inset. The line shows the
lattice heat capacity fit discussed in the text.
The heat capacity measurements shown in Figure 1(c)
show a clear feature at the structural transition and no
anomalies or effects due to latent heat were evident at
any other temperatures. Our data are in good agreement
with those reported on this compound by Tegel et al.7.
To separate the lattice and magnetic contributions to the
heat capacity, we estimated the lattice background using
C(T) = γT + ADCD(T,θD) + AECE(T,θE), (1)
where γ is the Sommerfeld coefficient, and CDand CEare
Debye and Einstein terms respectively. This was found
SrFeAsF showing the spin precession signal evident at low-
temperature, the greater damping of the oscillations close to
TN, and the paramagnetic signal at 140 K. The data are fitted
to Eq. 2 with the parameters shown in Figure 3.
(Color online.)Muon asymmetry data for
to be an effective model for oxypnictides in Ref. 25. The
parameters extracted from this fit (excluding data be-
tween 100 and 185 K) were γ = 3.44(7) mJmol−1K−2,
AD = 56.6(5) Jmol−1K−1, θD = 237(1) K, AE =
52.2(4) Jmol−1K−1, and θE= 407(3) K. These are com-
parable with the values determined for oxypnictide ma-
terials without rare-earth magnetic moments.17,25The
magnetic contribution is plotted in the inset to Fig-
ure 1(c) showing that zero-field and 10 T measurements
were effectively identical, and the integrated magnetic
entropy is 0.5 Jmol−1K−1.
tropy change, it is twice the value observed in LaFeAsO,
where features at both the structural and magnetic tran-
sitions are evident.17SrFe2As2 has a far larger entropy
change at the combined first-order structural and mag-
netic transition, ∼ 1 Jmol−1K−1. The majority of the
entropy change in SrFeAsF occurs close to Ts= 175 K,
but it appears that another much broader feature at lower
temperature also contributes. Since we find long-range
magnetic ordering at TN= 120 K using µSR (described
below), it seems that the broad feature is likely to have
a magnetic origin. The lack of a distinct anomaly in the
specific heat (or in magnetization or resistivity data),7
suggests that the residue below the structural transition
comes from the build up of 2D correlations within the
FeAs planes. Gaining a rough estimate of the in-plane
exchange constant J ∼ 250− 300 K from the position of
the hump, and knowing TN = 120 K, we can estimate
the out-of-plane exchange constant J⊥∼ 0.05J, consis-
tent with the lack of any observed anomaly at TN.3,4
This is a similar situation to that in La2CuO4,26though
with a lower anisotropy in the exchange constants and a
smaller separation between the structural and magnetic
In Figure 2 we present muon decay asymmetry data
at temperatures of 10, 116, and 140 K. At low temper-
atures, up to around 75 K, two oscillations are clearly
While this is a small en-
raw asymmetry data using Eq. 2 described in the text. (a)
Oscillation frequencies ν1and ν2with lines drawn showing the
power law function described in the text. It is noticeable that
the sharp drop-off in the frequencies near to the transition is
poorly described by this function. (b) Linewidths of the two
oscillating components, λ1 and λ2.
(Color online.) Parameters extracted from fitting
resolved but as we approach TNthe broadening of each
of the oscillations grows until they are both overdamped.
This overdamped behavior is seen in the 116 K data
set. Immediately above the magnetic ordering transition
the muon decay asymmetry takes the exponential form
expected for a paramagnet with electronic fluctuations
faster than the characteristic time of the measurement.
The data set at 140 K shown in Figure 2 is very similar
to all those taken above the magnetic ordering tempera-
ture, and we saw no change in the relaxation signal when
passing through Ts= 175 K.
Observing two precession frequencies in the magneti-
cally ordered phase and finding that the temperature de-
pendent relaxation is Gaussian (suggesting that the fluc-
tuations of electronic moments are motionally narrowed),
we were able to describe the raw asymmetry data using
the fitting function:
The first two terms describe two damped oscillations, the
third term describes the Gaussian relaxation for muon
spins with their direction along that of the local field
at their stopping site, which are depolarized by a ran-
dom distribution of nuclear moments, and the final term
describes the weak temperature-independent depolariza-
tion observed for muons stopping outside the sample.
Above the magnetic ordering transition there is no oscil-
latory signal and we set A1= A2= 0. In many fluorine
containing magnets a characteristic signal due to the for-
mation of a bound state between a positive muon and
one or more fluoride ions is observed above the magnetic
ordering transition.27,28No such signal is observed in Sr-
FeAsF, probably because the magnetic ordering transi-
tion is at too high a temperature for the muons to be
sufficiently well bound.
The parameters derived from fitting Eq. 2 to the raw
data are shown in Figure 3. The two precession frequen-
cies plotted in Figure 3(a) are well defined and at low-
temperature appear to follow a conventional power law.
Fitting the upper precession frequency to the function
ν(T) = ν(0)(1−(T/TN)α)βleads to TN = 120.6(3) K,
α = 3.1(3), and β = 0.20(2). This is a much sharper
magnetic transition than in LaFeAsO29,30and this would
suggest that the magnetism is more two-dimensional in
this fluoropnictide. Also, β is between the values ex-
pected for 2D Ising and 2D XY order parameters, though
the sharp drop in the frequencies near to TNmay mean
that this fitting function is less effective in estimating
the true critical parameters. The higher precession fre-
quency tends to ν1(0) = 22.22(5) MHz and, assuming the
same power law, the lower precession frequency tends
to ν2(0) = 1.9(1) MHz. These frequencies are a little
lower than in LaFeAsO29,30but in a similar proportion.
This suggests the magnetic structure is very similar to
LaFeAsO and the ordered Fe moments µFe∼ 0.3 µB.14,29
Seeing the lower frequency signal persisting all the way
to the magnetic ordering transition as Carlo et al.30did
in LaFeAsO suggests that this minority oscillation sig-
nal is intrinsic to the sample, and reflects the antifer-
romagnetic structure being sampled at a different site
within the structure. It had previously been suggested
that the some magnetic signals in these pnictide mate-
rials originated in FeAs impurities (e.g. Ref. 17) but we
can discount this possibility for our SrFeAsF sample on
the basis of µSR measurements on FeAs and FeAs2, both
of which give significantly different signals.31In the or-
dered phase the higher frequency oscillation accounts for
about 85 % of the oscillating amplitude. This ampli-
tude ratio for the two oscillating components is similar
to the situation in LaFeAsO, as is the lower frequency
signal becoming overdamped close to the magnetic or-
dering transition.29,30The linewidths λ1and λ2[shown
in Figure 3(b)] are both much smaller than the respective
precession frequencies at low temperatures, giving rise to
the clear oscillations seen in the 10 K data in Figure 2,
and then grow towards the ordering transition giving the
overdamped oscillations seen in the 116 K data.
Our results have shown that long ranged, three-
dimensional antiferromagnetic ordering in SrFeAsF oc-
curs, but with a greater separation between the structural
and magnetic ordering transitions (Ts−TN∼ 50 K) than
in comparable oxypnictide compounds (e.g. LaFeAsO).
While the µSR measurements show that the magnetic en-
vironment within the FeAs planes is very similar to that
in oxypnictide compounds, we note that the magnetic
ordering transition is not as clear in the magnetization
and heat capacity measurements. The heat capacity and
µSR measurements, in particular the lack of a heat ca-
pacity anomaly at TNand the low value of β = 0.2, both
suggest far more two-dimensional magnetic interactions
than in oxypnictide compounds, consistent with the in-
creased separation Ts− TN. This is also consistent with
the expectation that the interplanar exchange mediated
by a fluoride layer will be weaker than that mediated by
an oxide layer.
Part of this work was performed at the Swiss Muon
Source, Paul Scherrer Institute, Villigen, CH. We are
grateful to Alex Amato for experimental assistance and
to the EPSRC (UK) for financial support.
1E. Manousakis, Rev. Mod. Phys. 73, 1 (1991).
2M. A. Kastner et al., Rev. Mod. Phys. 70, 897 (1998).
3P. Sengupta et al., Phys. Rev. B 68, 094423 (2003).
4C. Yasuda et al., Phys. Rev. Lett. 94, 217201 (2005).
5Y. Kamihara et al., J. Am. Chem. Soc. 130, 3296 (2008).
6J. Zhao et al., Nature Materials 7, 953 (2008).
7M. Tegel et al., arXiv:0810.2120 (unpublished).
8F. Han et al., Phys. Rev. B 78, 180503(R) (2008).
9S. Matsuishi et al., J. Am. Chem. Soc. 130, 14428 (2008).
10G. Wu et al., arXiv:0811.0761 (unpublished).
11S. Matsuishi et al., arXiv:0811.1147 (2008).
12S. Matsuishi et al., J. Phys. Soc. Jpn. 77, 113709 (2008).
13X. Zhu et al., arXiv:0810.2531 (unpublished).
14C. de la Cruz et al., Nature 453, 899 (2008).
15Y. Chen et al., Phys. Rev. B 78, 064515 (2008).
16S. A. J. Kimber et al., Phys. Rev. B 78, 140503(R) (2008).
17M. A. McGuire et al., Phys. Rev. B 78, 094517 (2008).
18M. A. McGuire et al., arXiv:0811.0589 (unpublished).
19H. Luetkens et al., arXiv:0806.3533 (2008).
20A. J. Drew et al., arXiv:0807.4876 (unpublished).
21C. Krellner et al., Phys. Rev. B 78, 100504(R) (2008).
22M. Rotter et al., Phys. Rev. Lett. 101, 107006 (2008).
23K. Sasmal et al., Phys. Rev. Lett. 101, 107007 (2008).
24S. J. Blundell, Contemp. Phys. 40, 175 (1999).
25P. J. Baker et al., arXiv:0811.2494 (unpublished).
26K. Sun et al., Phys. Rev. B 43, 239 (1991).
27J. H. Brewer et al., Phys. Rev. B 33, 7813 (1986).
28T. Lancaster et al., Phys. Rev. Lett. 99, 267601 (2007).
29H.-H. Klauss et al., Phys. Rev. Lett. 101, 077005 (2008).
30J. P. Carlo et al., arXiv:0805.2186 (unpublished).
31P. J. Baker et al., Phys. Rev. B (accepted) arXiv:0809.2522