Superconductivity in SrFe_(2-x)Co_xAs_2: Internal Doping of the Iron Arsenide Layers

Superconductivity in SrFe2-xCoa;As2: Internal Doping of the Iron Arsenide Layers

OO
O

o

A. Leithe-Jasper, W. SchncUc, C. Gcibcl, and H. Rosncr
Max-Planck-Institut fiir Chemische Physik fester Staff e, Nothnitzer Str. 40, 01187 Dresden, Germany

In the electron doped compounds SrFe2-2!Co2, As2 superconductivity with Tc up to 20 K is observed
for 0.2 < a; < 0.4. Results of structure determination, magnetic susceptibility, electrical resistivity,
and specific heat are reported. The observation of bulk superconductivity in all thermodynamic
properties - despite strong disorder in the Fe-As layer ~ favors an itinerant picture in contrast to
the cuprates and renders a p- or d-wave scenario unlikely. DFT calculations find that the substitution
of Fe by Co [x > 0.3) leads to the suppression of the magnetic ordering present in SrFe2As2 due to
a rigid down-shift of the Fe-3d^2_y2 related band edge in the density of states.

PACS numbers: 74.10.+V, 74.25.Bt, 74.25. Dw

o ,

Oh:

o

>

m

(N
(N
(N

l>

O
00

o

X

The observation of superconductivity in layered Fe-As
systems has recently attracted considerable interest. The
imdoped compounds, e.g., LaFeAsO [ll, |3| or AFe2As2 {A
~ Ba, Sr) [sl, y, |5|, present an antifcrromagnetic (afm)
transition related to a structural transition at tempera-
tures in the range 140 K to 205 K. Upon doping both the
structural and the magnetic ordering get suppressed and
superconductivity with Tc up to 56 K appears [l|, la, |7| ■
The onset of superconductivity at the disappearance of
an afm ordered state induced by charge doping from a
reservoir layer is reminiscent of the behavior observed in
the cuprates. Since furthermore Fe and Cu are both Sd
elements, and because of similarities in the structure of
the layers, an analogy between the high temperature su-
perconductivity (HTSC) in the cuprates and the layered
Fe-As systems was suggested in a large number of re-
ports. On the other hand, the absence of a Mott-insulator
transition as well as the small ordered Fe moment in the
undoped compounds put this analogy into question, and
suggest that a model based on itinerant and weakly cor-
related 3d electrons might be more appropriate.

One way to get deeper insight into these questions is a
doping experiment on the Fe site. Presently, doping was
generally performed on sites in-between the Fe-As layers,
either on the i?-site or the 0-site in the i?FeAsO systems
or on the A site in the ^Fe2As2 compounds. Such kind
of doping corresponds, both in a localized and in an itin-
erant model, to a simple charge doping and is therefore
not very suitable for discriminating between both mod-
els. On the other hand, replacing a small amount of Fe
by another 3d clement should lead to different effects de-
pending on the model used. In an itinerant model, within
a simple rigid-band approach, substitution on the Fe site
is expected to be quite similar to indirect doping via the
interlayer sites, since only the total number of electrons is
relevant. In a picture with localized 3d electrons, doping
on the Fe site should lead to a very different behavior,
since the correlations in the 3d layers are directly affected
- a few percent Ni or Zn doping on the Cu site in a HTSC
cuprate lead to a drastic reduction of Tc-

Therefore, we investigated thermodynamic proper-
ties and electrical conductivity of the solid solution

SrFe2-a;Coa;As2. While pure SrFe2As2 undergoes a lat-
tice distortion and afm ordering at Tq = 205 K [J|, a
small amount of Co substitution leads to a rapid de-
crease of Tq, followed by the onset of superconductivity
in the concentration range 0.2 < a; < 0.4 with a maxi-
mum Tc of ~ 20 K. The observation of superconductiv-
ity in Co substituted SrFc2As2 provides strong evidence
that an analogy with the HTSC cuprates is not really
justified. We therefore performed band structure calcu-
lations and discuss the electronic structure in view of our
experimental results. The good agreement between cal-
culations and experimental observations confirms that an
itinerant approach is appropriate for these layered Fe-As
systems. The observation of superconductivity despite a
strong in-layer disorder puts also a strong constraint on
the possible superconducting order parameters. While
finalizing our investigation, Sefat et al. [8| and Wang et
al. [9[ reported the observation of superconductivity in
Co doped LaFeAsO with a maximum Tc ~ 10 K. While
their results support our analysis, the higher Tc observed
in SrFe2_3:COa;As2 establishes much stronger evidence.

Samples were prepared by a sintering technique in
glassy-carbon crucibles which were welded into tanta-
lum containers and sealed into evacuated quartz tubes
for heat treatment at 900 °C for 16 h followed by two re-
grinding and compaction steps. First precursors SrAs,
C02AS and Fe2As were synthesized from elemental pow-
ders sintered at 600 °C for 48 h (Fe, Co powder, 99.9
wt.%; As sublimed lumps 99.999 wt.%; Sr 99.99 wt.%.
These cducts were then powdered, blended in stoichio-
metric ratios, compacted, and heat treated. All steps
were carried out within an argon- filled glove box (O2,
H2O < 1 ppm). This mode of synthesis helps to prevent
possible contamination by toxic arsenic [lO[. Samples
were obtained in the form of sintered pellets.

Metallographic preparations and microstructure inves-
tigation on polished surfaces with optical microscopy
were performed in an argon-filled glove box. Electron-
probe microanalysis (EPMA) with energy dispersive
analysis was accomplished in a Philips XL30 scanning
electron microscope. Powder X-ray diffraction was per-
formed using Co Kal radiation (A = 1.789007 A) apply-

ing the Guinier Huber technique with LaBg as internal
standard (a = 4.15692 A). Crystallographic calculations
were done with the WinCSD program package 11 1 .

Evaluation of the lattice parameters and EPMA in-
vestigations of selected samples unambiguously reveals
the substitution of Fe by Co which is accompanied by a
decrease of the c axis length of the unit cell (see Table
m. For the sample SrFe1.8Coo.2As2 which was essentially
single phase (minor impurity SrAs) the crystal structure
was refined by Rietfeld methods (BaAU-type of structure
[12| . space group M/mmm, Sr in 2a (0, 0, 0), Fe/Co in
4d (0, 1/2, 1/4), As in 4e (0, 0, 0.3613(1)); Ri = 7.4, Rp
= 10.1). Other samples of this investigation contained as
minor impurity phases Fei_a;Coa;As.

Magnetization was measured from 1.8-400 K in a
SQUID magnetometer in various fields both after zero-
field cooling (zfc) and during cooling in field (fc). Heat
capacity was determined by a relaxation-type method on
a commercial measurement system (1.9-lOOK, 0-9T).
Electrical dc resistivity data p{T) (3.8-320 K) were col-
lected with a current density of < 1 x 10~^ Amm~^ in a
four contact arrangement. Due to the sample geometry
the inaccuracy in p is estimated to be ±20%.

To investigate the influence of electron doping on the
electronic structure of SrFc2As2 on a microscopic level,
we performed density functional band structure calcu-
lations within the local (spin) density approximation
(L(S)DA). Using the experimental structural parameters
Jl, we applied the full-potential local-orbital code FFLO
13| (version 7.00-28) with the Perdew-Wang exchange
correlation potential [1J| and a carefully converged k-
mesh of 24'^ points for the Brillouin zone. The substitu-
tion of Fe by Co and Sr by La was simulated within the
virtual crystal approximation (VCA).

TABLE I: Lattice parameters a,c of SrFe2-a:Coa;As2 (nom-
inal compositions) and superconducting transition tempera-
ture T'^^'s (defined as the crossing of the tangent to the steep-
est slope of the fc transition with x = 0) , xo is the high-field
susceptibility at r=0. Data for x=0 are from Ref. [J].

X

a

c

^mag

Xo

(A)

(A)

(K)

(10"'^emu/mol)

0.00

3.924(3)

12.34(1)

-

-

0.10

3.9291(1)

12.3321(7)

-

n.a.

0.15

3.9272(1)

12.3123(5)

<1.8

+780

0.20

3.9278(2)

12.3026(2)

19.2

-h600

0.25

3.9296(2)

12.2925(9)

18.1

(-h850)

0.30

3.9291(2)

12.2704(8)

13.2

-f460

0.40

3.9293(1)

12.2711(7)

12.9

+360

0.50

3.9287(2)

12.2187(9)

<1.8

n.a.

1.00

3.9618(1)

11.6378(6)

-

n.a.

The magnetic susceptibility in a nominal field poH

120 160 200
T (K)

FIG. 1: (Color online) Top: magnetic susceptibility x{T) of
SrFe2-2!Co2,As2 samples in a nominal field of hqH = 2mT.
Bottom: electrical resistivity of the same samples.

2mT (Fig. [T] top) displays strong diamagnctic signals in
zfc due to superconducting transitions in the a;=:0.2-0.4
samples with onset temperatures T™^^ up to 19.2 K for
^=0.2 (see Tabic |l]). The compounds with a;=0.1 and
x—0.5 show no traces of superconductivity. Even for
this low external field the transitions are slightly rounded
which indicates that the samples are not fully homoge-
neous. While the shielding signal (zfc) corresponds to the
whole sample volume (considering approximate demag-
netizing factors), the Meissner effect (fc) is much smaller,
which is probably due to strong pinning.

The normal-state susceptibilities of the samples a:;=:0.2,
0.3, and 0.4 arc field-independent for T > 200 K. There,
the compounds display a paramagnetism which can be
described by x(T) = Xo + X2T^- The extrapolated xo
values are given in TablelH Diamagnetic core contribu-
tions are comparatively small (« —50 x 10~^ emu/mol).
Due to the current sample quality some caution is neces-
sary in interpreting xo j but the order of magnitude and
the temperature dependence is typical for an enhanced
Pauli paramagnet. Comparing the xo values with the
electronic density of states at the Fermi level obtained
from band structure calculations (see below) suggests an
enhanced Sommerfeld- Wilson ratio, which decreases with
increasing Co content for x >0.15, in accordance with a
magnetic instability for x <0.15. SrCo2As2 is a Curie-
Weiss paramagnet {pcS = 2.06/^B/f-u., 6'cw = — 29K)
and does not show magnetic ordering above 1.8 K.

The temperature dependence of the resistivity p{T)

(Fig. [T] bottom) at high temperatures is generally that of
a bad metal. p{3QQK) decreases gradually with electron
doping from 5.5-8.0 ^f^m for a;=0.10"0.15 to 2.0/iSlm
for a;=0.50, as also observed for K-substituted samples. [3[
For a;=0.10 a stepwise increase of p{T) is observed below
«130K which shifts to «90K for x=0.15. This anomaly
can be assigned to the afm ordering and the related lat-
tice distortion observed at ro=205K in SrFe2As2 [i, Q.
Thus, Co substitution (electron doping) leads to the sup-
pression of the afm order in a similar way as reported for
the K-substitution (hole doping) [15[. For a;=0.20 no
such anomaly is seen and the compound shows a super-
conducting transition at 19.4 K.

FIG. 2: (Color online) Molar isobaric specific heat Cp/T^ of
SrFe2-i:Coi:As2 samples (x=0.2; x=0.3 shifted up by 3 units)
for different magnetic fields. For better distinction the data
for hqH — are connected by a line.

In Fig. [2] the specific heat Cp{T) for two selected sam-
ples is presented in a Cp{T)/T^ versus T plot. Strongly
rounded anomalies with onset at Tc as determined by
p{T) and x{T) data are clearly visible confirming that the
superconductivity is a bulk phenomen in these samples.
The idealized step heights Acp and the transition temper-
atures T^^^ were evaluated by a fit with a phononic back-
ground and an electronic contribution according to the
BCS theory or the phenomcnological two-liquid model
which applies well to the thermodynamic properties of
some strong coupling superconductors. The model is con-
voluted with a Gaussian to simulate the broadening due
to the chemical inhomogcneitics.

T^ Debye lattice term, i.e. Cp{T) = j'T+f3T^. While /3 is
field-independent and corresponds to a lattice term with
initial Debye temperature 8d(0) = 255(2) K for both
a;=0.2 and 0.3, the linear term 7' increases nearly lin-
early with field. For a;=0.2 the values of 7' range from
12.6 mJ mol-i at p.aH = OT to 17.5 at 9T. Whether
the residual 7' is induced by defects (see e.g. Ref. [16|) or
whether it is an intrinsic contribution due to ungapped
parts of the Fermi surface [17| has to be elucidated by
further experiments. While we cannot make a definite
statement on the superconducting coupling strength the
better fit is obtained for a;=0.2 and the two-fluid model
while for a;=0.3 the BCS fit is superior. This might indi-
cate a variation of the coupling strength with Co content.

— undoped

— Co doped

— La doped

Acp/T,

cal

ilO mJ mol-i K~2 ^j. 3,^0.2 and «13 mJ

mol ^ K ^ for x=0.3. Interestingly, well below Tc the
specific heat can still be described with a linear plus a

-1

Energy (eV)

FIG. 3; (Color online) Total electronic densities of states
for SrFe2As2 (black), SrFe1.7Coo.3As2 (red) and the fictitious
Sro.7Lao.3Fe2As2 (blue) in the vicinity of the Fermi level. In-
set: dependence of the magnetic moment on the number of
additionally doped electrons per formula unit for doping at
the Fe site (red) and at the Sr site (blue), respectively.

For a microscopic study of the influence of doping with
electrons on the electronic structure in the vicinity of the
Fermi level ep, we simulated two different kinds of par-
tial constituent exchange: (i) Co substitution at the Fe
site and (ii) a fictitious La substitution at the Sr site
to study possible differences for doping (i) within and
(ii) outside the FcAs layers. Throughout all calculations
we used the experimental lattice parameters [J] of the
undoped SrFe2As2 to separate the influence of geome-
try changes and pure electronic doping effects. Since the
changes of the lattice parameters upon doping are very
small (see Table|l| this approach is well justified. The re-
sulting total density of states (DOS) for the Fe-3d domi-
nated regions of the valence band for SrFe2-a;Coa;As2 and
Sri_a;La3;Fe2As2, respectively, is shown in Fig. [3] for an
exemplary doping level of a;=0.3.

The two kinds of doping result in a rather different be-
havior. Whereas (i) Co doping on the iron site results in
an almost perfectly rigid shift of the DOS for all studied

doping levels between x=0 and x=Q.5, (ii) exchange of Sr
by La can't be described at all in a rigid band picture in
the vicinity of ep (see Fig. [3]). In the case of Co substitu-
tion, a detailed analysis of the orbital-resolved DOS, the
corresponding bands and band characters shows that the
rigid band picture holds even for the individual orbitals
close to ep. In contrast, the substitution (ii) of Sr by La
shifts down the energy and changes the dispersion of a
band with Fe-As-Sr hybrid character in a way that it gets
partially occupied, whereas other bands remain basically
unchanged. In consequence, this leads to the formation
of a quite pronounced peak at ep and a sizable increase
of the DOS (sec Fig.[3|). In general, from the experimen-
tal results for the i?FeAsO and the AFe2As2 compound
family it seems that the appearance of superconductivity
at low temperatures is intrinsically related to the prior
destruction of the spin density wave. Therefore, we stud-
ied the instability towards magnetism depending on the
doping level for both doping scenarios (i) and (ii). In Fig.
|3]we plot the size of the ordered Fe moment correspond-
ing to the minimum in total energy in the VCA-LSDA
calculations as a function of Co or La substitution, (ii)
While La substitution stabilizes the magnetic state, elec-
tron doping by (i) Co leads to the disappearance of the
magnetic ground state at a;=0.3.

The disappearance of the magnetic instability is inti-
mately connected with the occupation of the Fc-3d^2_y2
related band along the F-Z direction as previously
demonstrated [J] for the undoped compound (depending
on the As z position). For .t=0.3 (see Fig. [3]) the related
band edge is situated right at Ef, leading to a reduced
DOS and thus to the destruction of the magnetic state.
In contrast, the tendency towards magnetism would be
increased for (ii) La substitution. The DOS enhancement
upon La substitution might lead to an instability of the
Sri_2.La2;Fe2As2 phase, which could be the reason why
such compositions have not yet been reported.

The non-magnetic ground state for x < 0.3 in
SrFe2-a;Coa;As2 in the VCA-LSDA calculations is in sur-
prisingly good agreement with the experimentally ob-
served suppression of the magnetic state and the onset
of superconductivity for .t=0.2. In fact, the used VCA
should result in an overestimate of the stability of a mag-
netic ground state since it does not include the direct in-
fluence of disorder in the Fe-As layer. Most likely, the
doped Co impurities are in a non-magnetic state, in-
dicated by calculations for SrCo2As2 that yield a non
magnetic ground state in agreement with our measure-
ments. Thus, a more sophisticated treatment of the Co
impurities should lead to a sizable reduction of the crit-
ical concentration x for the suppression of the magnetic
instability, in line with our experimental results.

In summary, we synthesized polycrystalline
SrFe2-a;Coa;As2 samples and investigated how elec-
tron doping within the Fe-As layers affects the magnetic
and electronic properties of this kind of compounds. We

found that Co substitution suppresses the afm transition
and the associated lattice distortion quite rapidly, from
To = 205 K at x=0 to Tq w 90K at 1-0.15. For j:=0.2
afm order is destroyed and bulk superconductivity
with a Tc « 20 K appears. Further increase of the Co
content leads to a reduction of Tc and finally to the
disappearance of superconductivity for x < 0.5. Thus
electron doping within the Fe-As layer by substituting
Co for Fe has a similar effect as hole doping in-between
the layers by substituting K for Sr, or as electron doping
by substituting F for O in LaFeAsO.

The observation of superconductivity upon substitu-
tion on the 3d site in the layered Fe-As systems is in
strong contrast to the behavior in cuprates, where substi-
tution on the Cu site lead to a fast suppression of super-
conductivity. This indicates that the suggested analogy
between HTSC in cuprates and in Fe-As systems is not
appropriate. Instead we observe a rigid-band like shift
of the DOS to lower energy upon Co substitution. The
suppression of the afm state, which occurs in the cal-
culations at a;=0.3, is connected with the d^2_y2 states
being pushed below the Fermi level. In contrast electron
doping by substituting La for Sr leads in the calculation
to a change in the dispersion and energy of a Fe-As-Sr
hybrid band. The resulting increase of the DOS at ep
stabilizes the afm state and might induce an instability
of the SrFe2 As2 phase. The nice correspondence between
the experiments and VCA-LSDA band structure calcu-
lations indicates that an approach based on itinerant,
weakly correlated 3d electrons is more appropriate for
the description of the Fe-As systems than an approach
based on localized 3d moments within a Hubbard model.

We thank U. Burkhardt, T. Vogel, R. Koban, K.
Kreutzigcr, Yu. Prots, and R. Gumeniuk for assistance.

[1]

[2]
[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]
[12]

Y. Kamihara, T. Watanabe, M. Hirano, and H. Hosono,

J. Am. Chem. Soc. 130, 3296 (2008).

C. de la Cruz et al.. Nature 453, 899 (2008).

M. Rotter, M. Tegel, I. Schellenberg, W. Hermes,

R. Pottgen, and D. Johrendt (2008), arXiv:Q8Q5.4021.

C. Krellner, N. Caroca-Canales, A. Jesche, H. Rosner,

A. Ormeci, and C. Geibel (2008), arXiv:Q8Q6.1209vl.

M. Tegel, M. Rotter, V. Weiss, F. M. Schappacher,

R. Pottgen, and D. Johrendt (2008), arXiv:0806.4782vl.

M. Rotter, M. Tegel, and D. Johrendt (2008),

arXiv:0805.4630vl.

C. Wang et al. (2008), arXiv:0804.4290.

A. S. Sefat et al. (2008), arXiv:0807.0823vl.
C. Wang et al. (2008), arXiv;0807.1304vl.

J. O. Nriagu, Arsenic in the environment, human health

and ecosystem effects, part II (Wiley, 1994).

L. G. Akselrud, P. Y. Zavalii, Yu. Grin, V. K. Pecharsky,

B. Baumgartner, and E. Wolfel, Mater. Sci. Forum 133-
136, 335 (1993).

K. R. Andress and E. Alberti, Z. Metallkde. 27, 126
(1935).

[13] K. Koepernik and H. Eschrig, Phys. Rev. B 59, 1743 [16] G. Triscone and A. Junod, in Bismuth-based High-

(1999). Temperature Superconductors, edited by H. Maeda and

[14] J. P. Perdew and Y. Wang, Phys. Rev. B 45, 13244 K. Togano (Marcel Dekker, New York, 1996), pp. 33-74.

(1992). [17] S. L. Drechsler at al., Physica B 329, 1352 (2003).

[15] G. F. Chen et al. (2008), arXiv:0806.1209vl.