arXiv:1608.04410v1 [cond-mat.supr-con] 15 Aug 2016
Mn-doping induced ferromagnetism and enhanced supercondu
ctivity in
Bi
4
−
x
Mn
x
O
4
S
3
(
0
.
075
≤
x
≤
0
.
15
)
Zhenjie Feng
1
,
2
,
a
, Xunqing Yin
1
, Yiming Cao
1
, Xianglian Peng
1
, Tian Gao
3
, Chuan
Yu
1
, Jingzhe Chen
1
, Baojuan Kang
1
, Bo Lu
4
, Juan Guo
5
, Qing Li
1
, Wei-Shiuan Tseng
6
,
Zhongquan Ma
1
, Chao Jing
1
, Shixun Cao
1
,
2
,
a
, Jincang Zhang
1
,
7
and N.-C. Yeh
6
,
a1
1
Department of Physics, Shanghai University, Shanghai 2004
44, China
2
Shanghai Key Laboratory of High Temperature Superconducto
rs, Shanghai 200444, China
3
Department of Physics, Shanghai University of Electric Pow
er, Shanghai 201300, China
4
Laboratory for Microstructures, Shanghai University, Sha
nghai 200444, China
5
School of Physical Engineering, Zhengzhou University, Zheng
zhou 450001, China
6
Department of Physics, California Institute of Technology
, Pasadena, CA 91125, USA
7
Materials Genome Institute, Shanghai University, Shangha
i 200444, China
(Dated: August 17, 2016)
We demonstrate that Mn-doping in the layered sulfides Bi
4
O
4
S
3
leads to stable Bi
4
−
x
Mn
x
O
4
S
3
compounds that exhibit both long-range ferromagnetism and
enhanced superconductivity for
0
.
075
≤
x
≤
0
.
15, with a possible record superconducting transition temp
erature (
T
c
)
∼
15 K
amongst all BiS
2
-based superconductors. We conjecture that the coexistenc
e of superconductivity
and ferromagnetism may be attributed to Mn-doping in the spa
cer Bi
2
O
2
layers away from the
superconducting BiS
2
layers, whereas the enhancement of
T
c
may be due to excess electron transfer
to BiS
2
from the Mn
4+
/Mn
3+
-substitutions in Bi
2
O
2
. This notion is empirically corroborated by
the increased electron-carrier densities upon Mn doping, a
nd by further studies of the Bi
4
−
x
A
x
O
4
S
3
compounds (A = Co, Ni; x = 0.1, 0.125), where the
T
c
values remain comparable to that of the
undoped Bi
4
O
4
S
3
system (
∼
4
.
5 K) due to lack of 4+ valences in either Co or Ni ions for excess
electron transfer to the BiS
2
layers. These findings therefore shed new light on feasible p
athways to
enhance the
T
c
values of BiS
2
-based superconductors, although complete elucidation of
the interplay
between superconductivity and ferromagnetism in these ani
sotropic layered compounds awaits the
development of single crystalline materials for further in
vestigation.
a
PACS numbers: 74.70.-b, 74.62.Bf, 74.25.Bt, 74.25.Ha
I. INTRODUCTION
One of the commonalities among the cuprate and iron-
based high-temperature superconductors is their lay-
ered structures.
1,2
Interestingly, even for conventional
superconductors, the highest superconducting transition
temperature (
T
c
) has been found in layered magne-
sium diboride MgB
2
.
3
Recently, superconductivity with
T
c
= 4.5 K was discovered in a new superconductor
Bi
4
O
4
S
3
.
4
This compound has a layered structure com-
posed of two superconducting BiS
2
layers and spacer lay-
ers of Bi
4
O
4
(SO
4
)
1
−
x
, where x indicates the deficiency
of (SO
4
)
2
−
ions at the interlayer sites. Since the discov-
ery of Bi
4
O
4
S
3
, several other BiS
2
-based superconductors
LnO
1
−
x
F
x
BiS
2
(Ln = La, Ce, Pr, Nd) with the highest
T
c
∼
10.6 K have been reported.
5–11
Both experimental
and theoretical studies to date have indicated that the
BiS
2
layers play the role of the superconducting planes in
these sulfide superconductors, similar to the CuO
2
planes
in the cuprate superconductors and the Fe
2
An
2
(An = P,
As, Se, Te) layers in the iron-based superconductors.
1,2,12
a
Nai-Chang Yeh (ncyeh@caltech.edu); Zhenjie Feng (fengzhe
n-
jie@shu.edu.cn); Shixun Cao (sxcao@shu.edu.cn). Corresp
on-
dence should be addressed to N.-C.Y. and requests for materi
als
should be addressed to either Z.-J. F. or N.-C. Y.
A major challenge facing this new class of layered su-
perconductors is to optimize
T
c
by exploring different
spacer layers. Additionally, the effects of doping by ei-
ther non-magnetic or magnetic elements are important
issues for investigation. To date, suppression of super-
conductivity has been observed in the case of Cu and Ag
substitutions for Bi in the Bi
4
O
4
S
3
superconductor,
13,14
whereas coexistence of superconductivity and ferromag-
netism has been reported in the CeO
1
−
x
F
x
BiS
2
15,16
and
Sr
0
.
5
Ce
0
.
5
FBiS
2
17
systems at low temperatures. How-
ever, none of these doping effects are fully understood.
Aiming at addressing the aforementioned issues, we re-
port in this work our studies of 3
d
transition-metal sub-
stitutions for Bi in Bi
4
O
4
S
3
by synthesizing Bi
4
−
x
A
x
O
4
S
3
(A = Mn, Co, Ni; 0
.
075
≤
x
≤
0
.
15) compounds with
conventional solid state reaction. We first focus on the
investigation of Bi
4
−
x
Mn
x
O
4
S
3
because these results are
most interesting and reveal a possible record
T
c
∼
15 K,
and then perform comparative studies on Bi
4
−
x
A
x
O
4
S
3
(A = Co, Ni) in the Discussion section to elucidate
the underlying physics. Based on our empirical find-
ings, we suggest that the coexistence of superconductiv-
ity and long-range ferromagnetism in all Bi
4
−
x
A
x
O
4
S
3
(A = Mn, Co, Ni) compounds may be attributed to
the selective doping of 3
d
transition-metal elements in
the spacer Bi
2
O
2
layers, whereas the enhancement of
T
c
found only in Mn-doped samples may be due to substan-
2
tial electron transfer from Mn
4+
/Mn
3+
-substitutions in
Bi
2
O
2
to the superconducting BiS
2
layers.
II. EXPERIMENTAL
Bulk polycrystalline Bi
4
−
x
Mn
x
O
4
S
3
(x = 0.075, 0.1,
0.125, 0.15) and Bi
4
−
x
A
x
O
4
S
3
(A = Co, Ni; x = 0.1,
0.125) samples were synthesized by conventional solid
state reaction method. For the Bi
4
−
x
Mn
x
O
4
S
3
samples,
high purity Bi (99.99%), Bi
2
O
3
(99.99%), S (99.999%),
MnO
2
(99.99%) were first weighed in stoichiometric ra-
tio and then grounded thoroughly in a glove box under
high purity argon atmosphere. Next, the mixture was
pressed into a pellet shape and sealed in an evacuated
quartz tube (10
−
4
Torr). The pellet was then heated up
to 510
◦
C and kept for 10 hours. After cooling the pellet
to room temperature, the product was well mixed again
by regrinding, pressed into a pellet shape, and then an-
nealed at 510
◦
C for another 10 hours. The samples thus
obtained looked black and were hard. It is important to
note that the sample may not be heated above 550
◦
C.
Otherwise S-O gas would be produced and could result
in explosion of the quartz tube.
4
Similar procedures were
applied to the synthesis of the Bi
4
−
x
A
x
O
4
S
3
(A = Co,
Ni; x = 0.1, 0.125) samples.
The crystal structures of all samples were character-
ized by X-ray powder diffraction (XRD, 18 kW D/MAX
2550) using the Cu-
K
α
radiation. The lattice constants
were calculated from the 2
θ
values and the Miller in-
dices by using the Jade 6.5 software. After XRD stud-
ies, these polycrystalline samples were cut into rectan-
gular shape and polished for electrical resistivity mea-
surements. The electrical resistivity was measured with
a standard four-terminal method covering temperature
range from 3 to 300 K in a Physical Property Measure-
ment System (PPMS-9, Quantum Design, Inc.). Typi-
cal current densities used for the resistive measurements
were
∼
100 A
/
m
2
. No apparent dependence on the cur-
rent density was found up to
∼
2000 A
/
m
2
, whereas re-
sistive signals became difficult to resolve for current den-
sities significantly smaller than 100 A
/
m
2
. The magne-
tization and specific heat measurements were conducted
using the same PPMS with Vibrating Sample Magne-
tometer (VSM) and specific heat options. The zero-field-
cooling (ZFC) and field-cooling (FC) of the magnetic sus-
ceptibility measurements of the samples were performed
in the warming process. Additionally, the carrier densi-
ties of bulk polycrystalline samples were determined from
their normal-state Hall coefficients at 300 K by means of
the van der Pauw method and with the use of the Hall
Effect Measurement System CVM200 made by the East
Changing Company.
III. RESULTS AND ANALYSIS
A. Structural characterization
The XRD patterns of different Mn-doped samples
and the corresponding crystalline structure are shown
in Fig. 1(a)-(b). These data suggest that all samples
acquired the expected tetragonal phase (space group
I4/mmm) with minor rhombohedra Bi
2
S
3
and Bi im-
purities, the latter being common occurrences in the
Bi
4
O
4
S
3
system
4,13,14,18
and exhibiting no superconduc-
tivity above 3 K.
19–21
The nominal Mn-doping (x) de-
pendence of the lattice constants
a
and
c
are illustrated
in the inset of Fig. 1(a). The general trend of decreasing
lattice parameters with increasing Mn-doping is reason-
able because the ionic radii of Mn are much smaller than
that of Bi
3+
.
22
This trend is also indicative of success-
ful incorporation of Mn-ions into the Bi
4
O
4
S
3
unit cells.
Similarly, XRD studies of Bi
4
−
x
A
x
O
4
S
3
(A = Co, Ni)
also indicate that the lattice constants of Bi
1
−
x
A
x
O
4
S
3
were all reduced relative to those of Bi
4
O
4
S
3
, as shown
in Fig. 1(c). Moreover, the lattice constants for different
dopants followed the descending order of Co, Ni and Mn,
as explicitly shown in the inset of Fig. 1(c).
B.
M
-vs.-
T
and
ρ
-vs.-
T
studies of
Bi
4
−
x
Mn
x
O
4
S
3
Temperature (
T
) dependent magnetization (
M
) of
Bi
4
−
x
Mn
x
O
4
S
3
with x = 0.075, 0.10, 0.125 and 0.15 was
studied under both zero-field-cool (ZFC) and field-cool
(FC) conditions from 3 to 300 K and with an external
field
H
= 100 Oe, as illustrated in Fig. 2. For each mag-
netization curve, three characteristic temperatures are
noteworthy: The N ́eel temperature (
T
N
) near
∼
125 K
for x = 0.125 and 0.15, below which
M
decreased due
to the onset of antiferromagnetism; the Curie tempera-
ture (
T
Curie
) near
∼
50 K for all samples, below which a
rapid upturn followed by saturation in the FC magneti-
zation curves appeared, suggesting the formation of long
range ferromagnetism; and the temperature
T
c,M
∼
4
.
5
K below which rapid decrease in magnetization occurred
as the result of supercurrent-induced diamagnetism. Ad-
ditionally, we note the dramatic contrasts between the
ZFC and FC magnetization curves for
T
c,M
< T < T
Curie
in all samples: The ZFC curves all exhibited an initial
upturn of magnetization, signaling the onset of ferromag-
netism, which was followed by gradual decrease and then
a sharp downturn in magnetization. Interestingly, both
the diamagnetic contribution in the ZFC curve and the
magnitude of ferromagnetism in the FC curve increased
with increasing x.
To better understand the interplay of magnetism and
superconductivity, we conducted measurements of resis-
tivity (
ρ
) vs.
T
on Bi
4
−
x
Mn
x
O
4
S
3
. As shown in Fig. 3(a),
all samples reached zero resistance at low temperatures.
On the other hand, the resistivity of samples with lower
Mn-doping levels (x = 0.075, 0.10) exhibited monotonic
3
FIG. 1. (Color online) Structural properties of Bi
4
−
x
A
x
O
4
S
3
(A = Mn, Co, Ni) : (a) X-ray diffraction (XRD) patterns
of Bi
4
−
x
Mn
x
O
4
S
3
(0
.
075
≤
x
≤
0
.
15). The inset shows the doping dependent variations of the i
n-plane and c-axis lattice
parameters
a
and
c
. (b) Schematics of the layered structure of Bi
4
−
x
Mn
x
O
4
S
3
. (c) X-ray diffraction (XRD) spectral studies of
Bi
4
−
x
Co
x
O
4
S
3
and Bi
4
−
x
Ni
x
O
4
S
3
for x = 0.125. The XRD spectra indicated that the lattice cons
tants after 3
d
transition-metal
doping were all reduced relative to those of Bi
4
O
4
S
3
, and the values for different dopants followed the descendin
g order of Co,
Ni and Mn, as shown in the inset of (c).
0
50
100
150
200
250
300
0.00
0.01
0.02
0.03
T
Curie
T
c,M
T
N
Magnetization (emu/g)
Temperature (K)
0.075 ZFC
0.075 FC
0.1 ZFC
0.1 FC
0.125 ZFC
0.125 FC
0.15 ZFC
0.15 FC
H=100 Oe
50
100
150
0.000
0.004
0.008
T
Curie
Magnetization (emu/g)
Temperature (K)
T
N
FIG. 2. (Color online) Temperature dependent magnetizatio
n
of Bi
4
−
x
Mn
x
O
4
S
3
(x = 0.075, 0.10, 0.125, 0.15): Zero-field-
cool (ZFC) and field-cool (FC) magnetization as a function of
T
is shown in the main panel from 3 to 300 K with an external
field
H
= 100 Oe. For each magnetization curve, three charac-
teristic temperatures are noteworthy: The N ́eel temperatu
re
(
T
N
) near
∼
125 K for x = 0.075, 0.10 and 0.125; the Curie
temperature (
T
Curie
) near
∼
50 K; and the magnetization-
determined superconducting transition temperature (
T
c,M
).
The inset shows an expansion of the main panel over the
temperature range where strong contrasts appear between th
e
ZFC and FC curves.
temperature dependence up to 150 K, whereas a resis-
tive upturn
23
appeared at
∼
55 K and
∼
23 K for higher
Mn-doping levels x = 0.125 and 0.15, respectively. Given
the highly anisotropic, layered nature of these BiS
2
-based
compounds, the physical origin for this doping dependent
resistive upturn cannot be fully uncovered without the
availability of single crystalline materials. Nonetheless, a
feasible mechanism that contributes to the resistive up-
turn is the occurrence of Kondo resonance at
T < T
K
,
where
T
K
denotes the Kondo temperature. In this sce-
nario, a lower
T
K
for a sample with a higher Mn-doping
level would be consistent with stronger ferromagnetism
and a sharper Kondo resonance of a linewidth
∼
T
K
.
23
Moreover, the formation of Kondo clouds below
T
K
could
help screen localized magnetic moments and so would be
important to the appearance of singlet superconductiv-
ity at
T
c
< T
K
.
24
However, the onset temperature
T
c,ρ
for rapid decrease in resistivity did not exhibit strong
doping dependence (Fig. 3(a)-(b)), and the
T
c,ρ
values
for all doping levels were generally higher than
T
c,M
, the
onset temperature for rapid ZFC and FC magnetization
decrease, although both
T
c,ρ
and
T
c,M
followed a simi-
lar non-monotonic trend (Fig. 3(c)-(d)). Given the com-
plex conduction paths and magnetic domain structures
in typical polycrystalline samples, the doping dependent
resistive upturn above
T
c
may be in part but cannot be
entirely attributed to the occurrence of Kondo resonance.
Generally speaking, the
T
c,M
values determined from
the onset of rapid FC magnetization decrease could not
be representative of the intrinsic superconducting transi-
tion, because the coexistence of ferromagnetism and su-
perconductivity would obscure the onset of the Meiss-
ner effect. Similarly, the polycrystalline nature of our
Bi
4
−
x
Mn
x
O
4
S
3
samples could significantly reduce the
T
c,ρ
values below the intrinsic superconducting transition
temperature
T
c
because of the inter-granular weak-link
effects.
25
Hence, additional thermodynamic measure-
ments of
M
-vs.-
H
at
T < T
Curie
and specific heat (
C
)-
4
FIG. 3. (Color online) Characterization of the superconduc
ting transition temperatures of Bi
4
−
x
Mn
x
O
4
S
3
: (a) Resistivity
(
ρ
) vs. temperature (
T
) behavior of Bi
4
−
x
Mn
x
O
4
S
3
. The inset is the enlargement of the lower temperature regim
e for x =
0.125 and 0.15, showing Kondo-like resistive upturn at
T
∼
55 K for x = 0.125 and at
T
∼
23 K for x = 0.15. (b) Detailed
ρ
-vs.-
T
curves of all samples near the onset of resistive supercondu
cting transition (
T
c,ρ
), where
T
c,ρ
(in units of K) exhibits
slight decrease with increasing x. (c) ZFC-magnetization v
s.
T
behavior under
H
= 100 Oe and near
T
c,ρ
(in units of K). (d)
Comparison of the Mn-doping level dependence of
T
c,ρ
from resistive data and
T
c,M
from magnetization measurements.
vs.-
T
studies in both zero and finite magnetic fields were
necessary to unravel the true
T
c
values of Bi
4
−
x
Mn
x
O
4
S
3
.
C.
M
-vs.-
H
studies of
Bi
4
−
x
A
x
O
4
S
3
In Fig. 4(a) and (b) we show the hysteretic
M
-vs.-
H
loops for x = 0.125 at
low
and
high
temperatures,
respectively. Specifically, the
low
-temperature behavior
in Fig. 4(a) with
T
= 3, 7, 12 and 14 K refers to the
appearance of anomalous features associated with each
magnetic hysteresis loop. These features diminished with
increasing
T
. In contrast, the
high
-temperature behav-
ior as manifested in Fig. 4(b) for the
M
-vs.-
H
loop at
T
= 20 K reveals a standard magnetic hysteresis loop
for a ferromagnetic material. We attribute the difference
between the low- and high-temperature behaviors to the
onset of superconductivity in the former. Specifically,
we consider a standard although much smaller supercon-
ducting magnetization loop
26
superposed on top of the
ferromagnetic hysteretic loop. Both the isothermal as-
cending and descending branches of the
M
-vs.-
H
loop at
T < T
c
would deviate from the typical ferromagnetic hys-
teresis loop due to the presence of supercurrents. Hence,
by considering the derivative
dM/dH
of either the as-
cending or descending branch of the
M
-vs.-
H
curve at a
constant
T
, we expect one peak associated with the in-
flection point of a standard ferromagnetic
M
-vs.-
H
curve
at
T
c
< T < T
Curie
. In contrast, an additional peak in the
dM/dH
-vs.-
H
curve is expected near
H
= 0 for
T < T
c
because of the appearance of supercurrents,
26
which is
indeed confirmed by the data shown in Fig. 4(c).
We may define the magnetic field difference between
the two peaks in
dM/dH
as ∆
H
∗
(
T
), which is a mea-
sure of the supercurrent.
26
Therefore, we expect ∆
H
∗
(
T
)
to decrease with increasing
T
and vanish at
T >
∼
T
c
,
which is consistent with the empirical finding shown in
Fig. 4(d), where ∆
H
∗
(
T
) approaches 0 at
T
= (16
±
2)
K for x = 0.125. Similar behavior has also been con-
firmed for x = 0.10 and 0.15, as shown in Fig. 5(a)-(d).