INSTITUTE OF PHYSICS PUBLISHING
SUPERCONDUCTOR SCIENCE AND TECHNOLOGY
Supercond. Sci. Technol. 15 (2002) 43–47
PII: S0953-2048(02)25280-X
Magnetic characterization of sintered
MgB2 samples: effect of substitution or
‘doping’ with Li, Al and Si
M R Cimberle1, M Novak2, P Manfrinetti3 and A Palenzona3
1
CNR Dipartimento di Fisica, via Dodecaneso 33, 16146 Genova, Italy
Institut of Physics, V Holesovickach, 18000 Prague 8, Czech Republic
3
INFM and Dipartimento di Chimica, via Dodecaneso 31, 16146 Genova, Italy
2
E-mail: cimberle@fisica.unige.it
Received 21 May 2001, in final form 11 September 2001
Published 11 December 2001
Online at stacks.iop.org/SUST/15/43
Abstract
Powdered and sintered MgB2 samples, both pure and substituted or doped,
have been prepared and characterized through magnetic measurements
performed from T = 5 K up to a few degrees above the transition
temperature of about 39 K. For all the samples, the irreversibility line (IL)
appears much lower than the Hc2–T line and very far from that of both
high-Tc (YBCO) and low-Tc (Nb3Sn) materials indicating the need to
increase the pinning in this material to make it attractive for technological
applications. Moreover, we have verified through different procedures that
the sintered samples behave as well-connected bodies, showing no trace of
granularity; therefore, the critical current density values Jc may be obtained
by applying the ‘critical state model’ in a straightforward way. The
hysteresis loop measurements allowed estimation of Jc both for powders and
sintered samples and confirmed the strong field decrease of Jc, implicit in IL
behaviour. We attempt to introduce defects in the MgB2 structure by
different chemical treatments like substitution of lithium on the magnesium
site and doping of the precursor boron powders with aluminium and silicon.
Jc always increased in doped or substituted samples (up to a factor 3) and
this fact is meaningful, in particular in the light of the small level of
substitution or doping we performed. The best result in terms of Jc is
achieved by silicon doping that, moreover, does not significantly decrease
the transition temperature.
1. Introduction
The recent discovery of the MgB2 phase as an
intermetallic superconductor with an exceptionally hightransition temperature [1] has made this new and relatively
‘easy to do’ superconductor attractive not only from
a theoretical point of view, but also for technological
applications.
Since then, a very fast and extended
characterization of its physical properties has been carried
out all over the world [2–11]. For the coherence length
ξ 0, a value of 5.2 nm, has been reported [2]; such a long
ξ 0 value eliminates problems with weak links, which are
instead typically present in ‘high-Tc’ materials [12], and
suggests the possibility of achieving a remarkably high critical
0953-2048/02/010043+05$30.00 © 2002 IOP Publishing Ltd
current density if proper technological treatments are adopted.
Therefore, the computation of the critical current density Jc,
its temperature and field dependence and the possibility of
increasing it by properly adding pinning centres constitute a
crucial problem to be solved. We have attempted to introduce
defects in the MgB2 structure with two main purposes: first,
to decrease the mean free path of the normal electrons then
decreasing the coherence length ξ and consequently increasing
the Hc2 values; second, to introduce pinning centres in order
to make the irreversibility line steeper. We prepared different
MgB2 samples with Li substitution on Mg site as well as
by doping with Si and Al. The results obtained through the
standard procedure and the variations therein induced by our
modifications are presented and discussed.
Printed in the UK
43
M R Cimberle et al
2. Sample preparation
44
Powder
0.0
X (dimensionless)
The compounds MgB2 and Mg0.9Li0.1B2 were prepared by
direct synthesis from the pure elements: Mg (in form of fine
turnings, 99.999 wt% purity), Li (pieces worked and cut out
under dried organic solvent, 99.8 wt% purity) and crystalline B
(325-mesh powder, 99.7 wt% purity) were well mixed together
and closed by arc welding under pure argon into outgassed Ta
crucibles, which were then sealed in quartz ampoules under
vacuum. The samples were slowly heated up to 950 ◦ C and
maintained at this temperature for one day.
With regard to both Al- and Si-doped samples (both Al
and Si 99.999 wt% pure), a pre-reaction of B mixed with 0.5
at% of the dopant element was carried out in Ta tubes, closed
by the same procedure as above, and heat treated at 1000 ◦ C for
four days. The resulting doped B was subsequently used for
the synthesis with Mg. These two latter samples are, hereafter,
called MgB2(Al) and MgB2(Si) respectively. The choice of Al
and Si as dopant elements for the first doping attempts was
suggested by the existence of a terminal solid solubility in
B up to few at%. All the final products were grey-blackish
powders, only slightly sintered. They are characterized by
x-ray diffraction using a Guinier camera (Cu Kα radiation
and pure Si as an internal standard). Samples, in prismatic
shape, are then prepared by pressing the powders in a stainless
steel die into a pellet which is then sintered by heat treatment
at 1000 ◦ C for 2 days (again in Ta containers, welded under
argon and closed in quartz tubes under vacuum). The density
of the sintered samples is about 70% of the theoretical sample,
their final dimensions are approximately 2 × 2 × 3 mm.
Both the powders and sintered samples are observed by
SEM microscopy; the SEM images show that the powders
are constituted by agglomerates that show a wide size spread
ranging from a few µm up to clusters that are 150 µm large.
Grains inside the sintered samples have a smaller and more
homogeneous size never exceeding 30 µm.
Magnetic measurements were performed by a commercial
SQUID magnetometer (Quantum Design).
The value of the lattice parameters of the pure MgB2
samples i.e. a = 3.085(1) Å and c = 3.524(1) Å, are in very
good agreement with the literature [13]; those of Mg0.9Li0.1B2
and of both doped samples remain practically unchanged (both
the a and c values lie within ±0.001 Å with respect to that of
MgB2). At the very low level of Si and Al doping (0.5%),
a variation in the x-ray diffraction pattern is not expected
[14, 15].
As has already been noted in [16], Li substitution on Mg
up to a level of 30% does not change the x-ray diffraction
pattern but only the position of some diffraction peaks.
Anyway, due to the closeness in the atomic volume values
of these elements, for substitutions up to 10%, the variation
is very small and below the sensitivity of our measurements.
However, in the Mg0.9Li0.1B2 sample, a slight variation in
the intensity of reflections was observed; a calculated x-ray
pattern, by means of the Lazy–Pulverix programme [13],
agreed with the experimental pattern, confirming the effective
substitution on Mg site by Li.
On the other hand, because no reaction towards the
crucible wall was detected, and no unreacted Mg was observed,
any variation in the physical behaviour, we mean the decrease
0.2
Pure sintered
Reground
powder
Li-substituted
-0.2
Al-doped
-0.4
Si-doped
-0.6
µοH=10 Gauss
-0.8
-1.0
34
35
36
37
38
39
T[K]
Figure 1. Susceptibility versus temperature for powdered and
sintered samples in an applied magnetic field of 10 G.
of the transition temperature or the Jc variation, should give the
best indication that the dopant or substitutive element entered
in the MgB2 structure.
3. Experimental results and discussion
In figure 1 susceptibility measurements versus temperature at
a constant field of 10 G are shown for the following different
samples: pure powders, pure sintered sample, powders from
a reground sintered sample, Al-doped, Si-doped and Lisubstituted sintered samples.
All the measurements have been normalized to the
constant value χ = −1 at low temperatures. The highest
transition temperature (Tc onset = 38.8 K) is observed in the
pure MgB2 samples (both powders and sintered samples).
As may be seen, substitution and doping always decrease
Tc, indicating that the new element effectively entered in
the structure.
The strongest reduction is produced by
Al (Tc onset = 37.2 K) in accordance with data in the
literature [18], while Li and Si produce smaller Tc variations
(less than 0.6 and 0.4 K respectively). The transition
widths are of about 1 K, the steepest transition being
observed in the Si-substituted sample (Tc = 0.75 K).
Although the powders and the pure sintered sample show
the same onset of the transition, in the latter the transition is
worsened (χ = −0.5 is reached at T = 37.7 K in the powders
and at T = 37.3 K in the sintered sample). We point out
that this happens in spite of the potential counter-effect of the
penetration depth, that can decrease the shielded volume due
to the small grain size.
In figure 2 the irreversibility line (IL) and Hc2 for both
powders and some sintered samples are presented together
with IL for YBCO and Nb3Sn reference samples. In the
x-axis the reduced temperature is reported. The IL is measured
by observing the deformation of the response function in
the SQUID magnetometer, while the sample changes from
an irreversible to reversible behaviour as suggested in [19].
By this procedure, we measure IL for Jc → 0 and therefore
the highest H g irr (by using the terminology of [20]). The
upper critical field Hc2 is identified by the onset of the
Magnetic characterization of sintered MgB2 samples
6
500
T=5K
Si-doped IL
Si-doped Hc2
200
Al-doped Hc2
100
3
4
300
Al-doped IL
Pure sintered IL
3
Pure sintered Hc2
0
-100
Pure sintered
-200
YBCO IL
2
M[emu/cm ]
5
µοH [T]
400
Si-doped
-300
Nb3Sn IL
Powder
1
-400
Reground powder
-500
-4
0
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
T/Tc
Figure 2. Irreversibility line and Hc2 versus T for powdered and
sintered samples. For a comparison the IL for YBCO and Nb3Sn
samples are also shown. IL is indicated by continuous lines, Hc2 by
dotted lines. The same symbol refers to the same sample: full
symbols →IL, open symbols →Hc2.
superconducting transition (1% of the normal state value)
measured at various magnetic fields. No significant variation
in the Hc2 behaviour is observed either in substituted or doped
samples, thus indicating that the introduction of disorder has
not been sufficient to change the normal electron mean free
path and the ξ value. IL exhibits at low fields an upward
curvature followed, at higher fields, by a linear behaviour.
What is clear at a glance is that, for all the samples, it lies well
below that of YBCO and, to a greater extent, below that of the
Nb3Sn sample. This behaviour means a strong field decrease
of Jc (see also the comparison between critical current densities
in MgB2 and Nb3Sn sintered samples reported in [2]), which
makes the need of introducing effective pinning centres in the
MgB2 samples clear.
For all the MgB2 samples, the IL are similar, with the
exception of the Al-substituted one, which stays lower. By
measuring the hysteresis loops at T = 25 K, we have verified
that the closing down of the loops happens at lower fields for
the Al-doped sample, confirming the IL trend and suggesting
a different temperature dependence for the Al-doped MgB2
sample.
In figure 3 the hysteresis loops measured at T = 5 K for the
Si-doped and pure sintered samples are shown together with
the hysteresis loops measured both on the parent powders and
on powders obtained by regrinding a pure sintered sample.
This is done to check if the sintering process can change
the intrinsic properties of the MgB2 grains. For the sake of
clarity the hysteresis loops measured on the other sintered
samples are not presented. Anyway, all the samples show
a similar decrease of the hysteresis loops by increasing the
field: about one order of magnitude from 0 up to 2 tesla. The
substitutions do not change this general trend: their effect is
essentially to enlarge to various extents the amplitude of the
loops with respect to that observed in the pure sintered sample.
At low fields (and for temperatures up to 15 K) many jumps
are observed at the highest magnetization values. This is an
indication of connected samples in which, due to the high value
-3
-2
-1
0
1
2
3
4
µoH [T]
Figure 3. Hysteresis loops measured at T = 5 K for the various
samples as indicated in the legend.
of Jc and to the sample dimensions, a strong field gradient is
established. Also in the starting powders a jump has been
observed. We remember that powders exhibit a wide spread in
the granulometry with clusters that can reach a dimension of
150 µm. The hysteresis loops on the parent powders are of the
same order of magnitude as that of the sintered samples. In the
light of the great difference between the dimension of powders
and sintered samples, this implies that powders present a very
high Jc with respect to the sintered samples.
Before calculating the critical current density from the
hysteresis loops, it is necessary to assess if a sample behaves
like a connected body or some level of granularity is present.
To do this, it is either possible to measure the remanent
magnetization Mrem as a function of the maximum magnetic
field [21] or to check the shape of the hysteresis loop. In
particular, if the sample is totally connected, the derivative
of the virgin curve or of the reverse leg of the magnetization
follows a well-defined analytical behaviour. The results are
well known in the case of a field constant Jc and depend on
the sample shape (slab, cylinder etc) [22], but if Jc is strongly
field dependent it is impossible to define an analytical solution.
Therefore, we have examined the derivative of the reverse leg
of the magnetization at high fields, where the approximation
of constant Jc is better satisfied. Moreover, this analysis allows
determination of the value of the full penetration field µ0Hp at
a certain µ0Hmax value, and therein the Jc(H) value.
In figure 4(a) we show Mrem versus µ0Hmax as obtained
on the pure sintered sample at T = 5 K. The shape of the
curve, that presents a one-step transition, clearly indicates the
absence of granularity. The same was observed at T = 25 K. As
further confirmation of a ‘totally connected body’ behaviour,
in figure 4(b) the derivative of the reverse leg of magnetization
as observed in the same pure sample at T = 25 K and µ0Hmax =
0.6 T is shown. A fit of the experimental data with the
relationship
dM
2 · (Hmax − H ) (Hmax − H )2
−
= −1 +
dH
3Hp
(3Hp )2
(1)
valid for a bulk superconductor with a field constant Jc [22], is
reported as well. The reliability of the fit confirms the absence
45
M R Cimberle et al
a)
b)
Figure 4. (a): Mrem versus µ0Hmax at T = 5 K for the pure sintered MgB2 sample. (b): Derivative of the reverse leg of the magnetization
dM/dH versus µ0 (Hmax −H) as measured at µ0Hmax = 2 T and T = 25 K in the pure sintered sample (triangle: experimental data; circles
with continuous line: fit by the ‘critical state model’ (see text)).
Table 1. Jc values (in A cm−2) valued by the magnetic measurements at various fields and temperatures.
T=5K
Pure sintered
Si-doped
Al-doped
Li substituted
Powder
Reground powder
µ0H = 2 T
µ0H = 6 T
µ0H = 0 T
5 × 10
1.5 × 105
1 × 105
7 × 104
2 × 106
1 × 106
5 × 10
1.5 × 104
1 × 104
7 × 103
2 × 105
1 × 105
2 × 10
6 × 104
4 × 104
5 × 103
1.5 × 104
8 × 103
4
of granularity. The Jc calculation may be, therefore, performed
by the classical formulas of the ‘critical state model’. In
particular, taking into account the parallelepiped shape of our
samples, we have modelled the magnetization profile inside
the sample like an ‘inverted roof’ [23] with the same slope in
all directions (that means no anisotropy of the critical current
density). In table 1, the Jc values calculated for the various
samples at T = 5 K and T = 25 K are shown. We point out
that in any case Jc increases by increasing the doping, ranging
from 5 × 104 A cm−2 for the pure sintered samples up to
1.5 × 105 A cm−2 for the Si-doped sample at T = 5 K and
zero field. Going to µ0Hext = 2 T, Jc decreases by about
a factor 10. In table 1 the Jc values relative both for the
parent powders and for the reground powders are reported.
Because of the large spread in the powder sizes, to estimate
Jc we attributed the highest µ0Hp value, as obtained from the
derivative of the reverse leg of magnetization to the largest
observed cluster dimension (150 µm for parent powders and
70 µm for the reground ones). The obtained values are much
higher than for the sintered samples, but the same strong field
dependence is observed. This is consistent with the estimation
of the activation energy U0 obtained by the magnetization
decay measurements performed at µ0Hext = 0.2 T and 2 T and
T = 10 K and analysed with the standard Anderson creep
theory [24]. The U0 values decrease by about one order of
magnitude: we found them to be 0.8 and 0.067 eV respectively.
A decrease of a factor 2 in Jc in the reground powders may
46
T = 25 K
µ0H = 0 T
3
4
Figure 5. Magnetization versus temperature for different applied
magnetic fields in Al-doped sample.
be attributed to a variation of intrinsic properties of the MgB2
grains that have been pressed, sintered and reground.
In the light of the presented data the decrease in Jc of
about one order of magnitude observed in the sintered sample
with respect to the powder, does not appear as a ‘weak link
effect’, as in high-Tc superconductors. The grain boundaries
have the same superconducting properties of the grains and
the sintered samples behave, from the magnetic point of view,
Magnetic characterization of sintered MgB2 samples
like well-connected bodies. This appears reasonable due to
the small surface contact between the grains that, for their
hardness, are difficult to compact. Moreover, the sintering
procedure can change the intrinsic superconducting properties
of the material.
Finally, the temperature variation of the critical current
density is measured for the Al-doped samples. The results
are presented in figure 5 as M (H = 0 T, 0.6 and 1 T)
versus temperature. A linear decrease is observed both in the
remanent magnetization and in the higher field magnetization.
4. Conclusions
Powdered and sintered samples of the MgB2 pure, Lisubstituted and doped with Si and Al, have been characterized
by a set of magnetic measurements. The ILs indicate
unambiguously a wide region below Hc2 where the ‘flux flow’
regime exists and dissipation takes places. Therein follows
the need to increase pinning and in particular its strength
at increasing magnetic fields to make this material more
appealing for technological applications. The very strong field
dependence of the critical current density is also indicated
by the values of the activation energy U0 calculated by the
magnetization decay measurements on powders at T = 10 K
and at 0.2 and 2 T.
An attempt to introduce disorder has not produced a
significant change either in Hc2 or in the IL at least at the
low-level of substitution and doping we have carried out.
Nevertheless, the hysteresis loops measured at low temperature
indicate an improvement of the critical current density for all
the chemical treatments, the best result being achieved by the
Si doping (about a factor 3 in the Jc values). So, Si or Al act as
the pinning centres at least at low temperature, suggesting their
presence as clusters. By increasing the temperature from 5 K
to 30 K the critical current decreases only by a factor of 6. The
presence of magnetic granularity has been tested by different
procedures, but no trace is found in our sintered samples.
Therefore, it is reasonable to attribute the strong decrease of
the critical current density of the sintered samples with respect
to the powder to the low density of the sintered samples
(small connection area between grains) and also to intrinsic
variations due to the sintering procedure (e.g., relaxation of
defects during the 2-day long heat treatment). Anyway, the
lack of ‘weak links’ suggests that potentially high Jc values
are obtainable; for this purpose, both the sample density and
the pinning strength must be increased.
Acknowledgments
Two of us (A P and P M) thank the Italian Ministero della
Ricerca Scientifica e Tecnologica (MURST) for financial
support. This work is a part of the National Research Program
‘Alloys and Intermetallic Compounds: Thermodynamics,
Physical Properties, Reactivity’.
References
[1] Nagamatsu J, Nakagawa N, Muranaka T, Zenitani Y and
Akimitzu J 2001 Nature 410 63
[2] Finnemore D K, Ostenson J E, Bud’ko S L, Lapertot G and
Canfield P C 2001 Phys. Rev. Lett. 86 2420
[3] Bugoslavsky Y, Perkins G K, Cohen L F and Caplin A D 2001
Nature 410 563
[4] Kambara M, Hari Babu N, Sadki E S, Cooper J R, Minami H,
Cardwell D A, Campbell A M and Inoue I H 2001
Supercond. Sci. and Technol. 14 L5
[5] Lorentz B, Meng L, Xue Y Y and Chu C W 2001 Preprint
cond-mat/0104041
[6] Kang W N, Jung C U, Kijoon H P, Park Min-Seok, Lee S Y,
Kim Hyeong-Jin, Choi Eun-Mi, Kim Kyung Hee, Kim
Mun-Seog and Lee Sung-Ik 2001 Preprint
cond-mat/0102313
[7] Larbalestier D C et al 2001 Nature 410 186
[8] Dhalle’ M, Toulemonde P, Beneduce C, Musolino N and
Decroux M 2001 Preprint cond-mat/0104395
[9] Bud’ko S L, Petrovic C, Lapertot G, Cunningham C E,
Canfield P C, Jung M-H and Lacerda A H 2001 Phys. Rev.
B 63 220503 (R)
[10] Bud’ko S L, Lapertot G, Petrovic C, Cunningham C E,
Anderson N and Canfield P C 2001 Phys. Rev. Lett. 86
1877
[11] Fuchs G, Müller K-H, Handstein A, Nenkov K, Narozhnyi
V N, Eckert D, Wolf M and Schultz L 2001 Solid State
Commun. 118 497
[12] Cimberle M R, Ferdeghini C, Flükiger R, Giannini E, Grasso
G, Marré D, Putti M and Siri A S 1995 Physica C 251 61
[13] Naslain M M R, Guette A and Barret M 1973 J. Solid State
Chem. 8 68
[14] Li J Q, Li L, Liu F M, Dong C, Xiang J Y and Zhao Z X 2001
Preprint cond-mat/0104320
[15] Xiang J Y, Zheng D N, Li J Q, Li L, Lang P L, Chen H, Dong
C, Che G C, Ren Z A, Qi H H, Tian H Y, Ni Y M and Zhao
Z X 2001 Preprint cond-mat/0104366
[16] Zhao Y G, Zhang X P, Qiao P T, Zhang H T, Jia S L, Cao B S,
Zhu M H, Han Z H, Wang X L and Gu B L 2001 Preprint
cond-mat/0103077
[17] Yvon K, Jeitschko W and Parthé E 1977 J. Appl. Crystallogr.
10 73
[18] Slusky J S, Rogado N, Regan K A, Hayward M A, Khalifah P,
He T, Inumaru K, Loureiro S, Haas M K, Zandbergen H W
and Cava R J 2001 Nature 410 343
[19] Suenaga M, Welch D O and Budhani R 1992 Supercond. Sci.
Technol. 5 S1
[20] Wen H H, Li S L, Zhao Z W, Ni Y M, Ren Z A, Che G C and
Zhao Z X 2001 Preprint cond-mat/0103521
[21] Müller K-H, Andrikidis C, Liu H K and Dou S X 1994 Phys.
Rev. B 50 10218
[22] Bean C P 1964 Rev. Mod. Phys. 36 31
[23] Gyorgy E M, van Dover R B, Jackson K A, Schneemeyer L F
and Waszczk J V 1989 Appl. Phys. Lett. 55 283
[24] Anderson P W 1962 Phys. Rev. Lett. 9 309
47