Ceramics International 29 (2003) 117–137

www.elsevier.com/locate/ceramint

Enhancement of the trapped-fields and mechanical

properties of Nd–Ba–Cu–O bulk superconductors
M. Matsuia,b,*, M. Murakamia
a
Superconductivity Research Laboratory, International Superconductivity Technology Center,
1-16-25 Shibaura, Minato-ku, Tokyo 105-0023, Japan
b
Railway Technical Research Institute, 2-8-38, Hikari-cho, Kokubunji, Tokyo 185-8540, Japan

Received 29 April 2002; received in revised form 5 May 2002; accepted 5 May 2002

Abstract
We have studied the effects of processing conditions and the initial compositions in Nd–Ba–Cu–O bulk superconductors on the
field-trapping capability and the mechanical properties. Sintering the sample prior to the crystal growth was effective for the fab-
rication of a good quality sample with high trapped-fields. The optimization of Nd422 and Ag contents along with processing
conditions led to an increase in both the field-trapping ability and mechanical properties. As a result, the trapped-field of Nd–Ba–
Cu–O 30 mm in diameter reached 1.4 T at liquid nitrogen temperature. The average tensile strength and the Weibull coefficient
exceeded 50 MPa and 13, respectively.
# 2002 Elsevier Science Ltd and Techna S.r.l. All rights reserved.
Keywords: B. Microstructure; C. Superconductivity; Trapped-field; Tensile strength; Bulk; Nd–Ba–Cu–O

1. Introduction was controlled by the top-seeded melt-growth (TSMG)

technique using a c-axis aligned seed crystal with a
After the discovery of high temperature super- higher decomposition temperature.
conductors, intensive research has been carried out to It is known that Y–Ba–Cu–O forms only a stoichio-
seek for the melt-processing methods suitable for metric superconducting 123 phase [7,8]. Unlike Y123,
industrial applications. At an early stage of the research, RE123 (RE: Nd, Sm, Eu, Gd) is known to form a RE-
bulk superconductors were fabricated with a sintering Ba solid solution [8,9]. When RE123 compounds are
method, since the technique is commonly used for the synthesized in air, a large amount of RE elements sub-
synthesis of ceramic materials. However sintered oxide stitute Ba site, leading to the degradation of super-
superconductors suffered from a weak-link problem at conducting properties [8,10]. Recent investigations
the grain boundaries. Melt-growth techniques [1,2] have revealed that superconducting properties were largely
solved such a weak-link problem through grain align- improved when the samples were synthesized under low
ment. An addition of non-superconducting particles was oxygen partial pressure [11]. In particular, Nd–Ba–Cu–
also effective in increasing flux pinning. Later, the O (Nd123/Nd422) superconductors exhibited remark-
refinement of non-superconducting inclusions was able properties such as Tc of 95–96 K and high Jc
achieved through the addition of Pt and CeO2, which accompanied by a large fishtail in a Jc–B curve [11], and
further increased the Jc values [3,4]. Recently, the Jc hence they were considered to be promising materials to
value in a self-field has reached the level of 105 A/cm2 at replace Y–Ba–Cu–O superconductors in various indus-
77 K [5,6]. In order to impart a high field-trapping trial applications.
ability, it is necessary to increase the size of a single In the Nd–Ba–Cu–O system, both Nd1+xBa2
x-
domain besides high Jc. Hence, the crystal orientation Cu3O7-
(Nd123ss) and Nd4-xBa2+xCu2O10 (Nd422ss)
phases have a wide range of Nd–Ba solid solution [8–
* Corresponding author. Tel.: +81-3-3454-9284; fax: +81-3-3454-
10,12]. Therefore, the superconducting properties of
9287. Nd–Ba–Cu–O superconductors are sensitive to the pro-
E-mail address: [email protected] (M. Matsui). cessing conditions such as the growth temperature [12]
0272-8842/03/$22.00 # 2002 Elsevier Science Ltd and Techna S.r.l. All rights reserved.
PII: S0272-8842(02)00094-9
118 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137

Fig. 1. Optical micrographs for the samples (a) without Ag addition and (b) with 10 wt.%Ag2O addition. The initial molar ratio of Nd123:Nd422 is
10:2. The average sizes of Nd422 particles in the matrix were (a) 4.44 and (b) 2.33 mm.

In this work, we fabricated Ag-doped large single-

domain Nd–Ba–Cu–O superconductors with various
initial compositions under different processing condi-
tions with the aim of optimizing the processing con-
ditions and the amount of Nd422 and Ag in regard to
the field-trapping capability. We also studied the effect
of Nd422 and Ag content on the mechanical
properties.

2. Experimental

High-Tc single-domain bulk superconductors are

usually synthesized by the TSMG technique. Nd bulks
are also subjected to this method. However, the decom-
position temperature of Nd system is highest among
high-Tc 123 superconductors. Hence, there was not an
Fig. 2. Thermal schedules for the preparation of Nd–Ba–Cu–O bulk
superconductors: (a) sintering, (b) melt-growth and (c) oxygenation. appropriate candidate for a seed crystal with higher
Sample A was not sintered while sample B was sintered before the decomposition temperature. Although Lo et al. [18]
melt-growth. Both samples were annealed in pure oxygen. attempted to fabricate Nd bulk with MgO seed crystal,
it was difficult to control the crystal orientation due to
and the compositional fluctuation at the growth the mismatch of the crystal lattice between MgO and
interfaces. Nd123. This was one of the factors that made it difficult
Furthermore, high Tc superconductors are brittle to fabricate single-domain Nd bulks. Recently, several
ceramic materials, and the field-trapping capability is researchers reported one of Ag functions in a bulk
often restricted by their poor mechanical properties sample was to decrease the decomposition temperature
[13,14]. The presence of normal conducting precipitates of Nd system [19], leading to the easier fabrication of
such as RE211, Nd422ss (Nd422 solid solution) and Ag Nd bulks. Furthermore, Fig. 1 shows the microstructures
particles affect the distribution of residual stress, and for the samples (a) without and (b) with Ag addition. In
thereby mechanical performance of bulk super- Nd system, Ag addition depressed the coarsening of
conductors. Although it was confirmed that Ag addition Nd422 particles unlike other systems, suggesting that
enhanced the mechanical strength of bulk super- these particles play a better role in flux pinning. Accord-
conductors [15–17], the effect of combined addition of ingly, we fabricated Ag-doped Nd bulks 30 mm in dia-
211(422) and Ag particles are still under discussion. meter and 10 mm in thickness using Nd seed crystals
 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137 119

without Ag addition in flowing 1%O2-Ar atmosphere.

The detailed experimental procedures are as follows.
We prepared precursor Nd–Ba–Cu–O powders with
various nominal compositions. After thorough mixing, the
powders were pressed into pellets 40 mm in diameter and
subjected to cold isostatic pressing. First, the pellets with
the nominal composition of Nd123:Nd422=10:2+10
wt.%Ag2O were employed to investigate the effect of sin-
tering prior to melt-processing on the superconducting
properties and the field-trapping capabilities.
Fig. 2 shows the thermal schedules for the preparation
of Nd–Ba–Cu–O bulk superconductors. Sample A was
grown using the TSMG technique with a Nd123 seed
crystal in flowing 1%O2-Ar atmosphere without sinter-
ing before melt-processing. Sample B was sintered at
Fig. 3. Photo of sample A. Sample A was not sintered in pure oxygen
before the crystal growth.
1040  C for 1 h in pure oxygen and then processed by

Fig. 4. Field distributions trapped by samples A and B at liquid nitrogen temperature: (a) A in the remnant state: (b) A in 1T; (c) B in the remnant
state; (d) B in 1T.
120 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137

the TSMG method. The grown crystals were oxyge- surface. The maximum trapped-field values were mea-
nated at 300 for 200 h. The densities of a precursor sured by attaching a Hall sensor directly on the sample
pellet, a sintered pellet and a grown crystal were mea- surfaces. The field dependencies of the maximum trap-
sured to investigate the effect of sintering. The trapped ped-field values for the grown blocks were also mea-
magnetic fields were measured in liquid nitrogen using a sured. Small specimens cut from grown bulks were
Hall sensor, which was scanned 1 mm above the top subjected to DC magnetization measurements using a

Fig. 5. Results of X-ray pole-figure analyses of (103) plane for sample A. The measured positions are presented in the inset. The positions (b) and
(d) are near the area where the flux could easily penetrate, which was presented in Fig. 4(a) and (b).
 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137 121

Fig. 6. Jc–B curves at 77 K for small specimens cut from various relative positions along a- and c-axis in samples (a) A and (b) B. The relative
locations are presented in the inset.
122 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137

superconducting quantum interference device magnet- trapped-field values were also measured. Small speci-
ometer (SQUID) to determine the Jc values. Further- mens were cut from samples D, E and F, and then DC
more, X-ray pole-figure analyses were performed to magnetization measurements were carried out with a
measure the misorientation angles for sample A. SQUID magnetometer to determine the Jc values.
Next, we fabricated a bulk Nd–Ba–Cu–O sample with Microstructural observations at various locations were
the same nominal composition of Nd123:Nd422=10:2 performed with a polarized microscope. Furthermore,
+10 wt.%Ag2O. The sample was also sintered in pure the specimens for the tensile tests were cut from various
oxygen at 1040  C for 1 h before the melt-growth.
However, during the crystal growth, we changed the
slow cooling rate from 1.0 to 0.3 C/h to crystallize this
sample (sample C) to reveal the effect of cooling rate on
the microstructure, the superconducting properties and
the field-trapping capabilities. Except for the cooling
rate, the heat schedules for this sample was the same as
that for sample B (see Fig. 2).
We also prepared 10 wt.%Ag2O-doped Nd–Ba–Cu–O
bulk samples with different Nd422 content,
Nd123:Nd422=10:1.5 (sample D), 10:2.5 (sample E)
and 10:3 (sample F) to study the effect of Nd422 content
on the microstructures, the superconducting properties,
the field-trapping capabilities, and the mechanical
properties. All the bulk samples were fabricated and
oxygenated using the same thermal schedules shown in
Fig. 2. We measured the trapped magnetic fields and
maximum trapped-field values in liquid nitrogen using a Fig. 7. Trapped-field distribution of sample C at liquid nitrogen tem-
Hall sensor. The field dependencies of the maximum perature.

Fig. 8. Field dependencies of the maximum trapped-fields for samples B and C at liquid nitrogen temperature.
 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137 123

Fig. 9. Average sizes of Nd422 phase at various locations in samples B and C. The inset shows the analyzed locations.

Fig. 10. Field dependencies of Jc values at 77 K for various locations in sample C. The relative locations are presented in the inset shown in Fig. 6(a).
124 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137

Fig. 11. Top views and field distributions trapped by 10 wt.%Ag2O-doped Nd–Ba–Cu–O bulks with different Nd422 content at liquid nitrogen
temperature: (a) Nd123:Nd422=10:1.5 (sample D), (b) 10: 2 (sample B), (c) 10:2.5 (sample E) and (d) 10:3 (sample F). The trapped-field distribution
for sample B presented in Fig. 4(c) was again shown for comparison.
 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137 125

locations in samples B, D, E and F with dimensions of itored. The field dependence of the maximum trapped-
3.5  3.5  12 mm3. The tensile loading direction was field values was also measured. The tensile strength was
perpendicular to (100)/(010) plane of the specimens. The measured like the tensile measurements of samples D,
both edges of the specimens were covered with alumi- B, E and F.
num rods and firmly fixed with epoxy resin [20]. The Finally we fabricated the Nd–Ba–Cu–O bulk sample
rate of displacement for the tensile measurements was with the nominal composition of Nd123:Nd422
0.5 mm/min. =10:1.5+20 wt.%Ag2O (sample H) to study the effect
In order to study the effect of Ag addition, we also of Ag addition on the mechanical strength. The trap-
fabricated the Nd–Ba–Cu–O bulk sample with the ped-field distribution and the field dependence of the
nominal composition of Nd123:Nd422=10:2+20 maximum trapped-field values were measured.
wt.%Ag2O (sample G). The processing conditions for Since the trapped field values are known to depend on
sample G were the same as those for sample B. After the sample thickness, we measured the trapped-field
the trapped magnetic field in liquid nitrogen was mea- distribution and the maximum trapped-field value of the
sured, the maximum trapped-field value was mon- stacked samples by changing the thickness.

Fig. 12. (a–c) Optical micrographs and (d) the average sizes of Nd422 at the relative locations in samples D, B, E and F: (a) sample D
(Nd123:Nd422=10:1.5+10 wt.%Ag2O), sample B (Nd123:Nd422=10:2+10 wt.%Ag2O), (b) sample E (Nd123:Nd422=10:2.5+10 wt.%Ag2O)
and (c) sample F (Nd123:Nd422=10:3+10 wt.%Ag2O). The overall average sizes of Nd422 phase were 2.34, 2.33, 2.19 and 2.06 mm, respectively.
The microstructure for sample B was already shown in Fig. 1(b).
126 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137

Fig. 12. (continued)

3. Results and discussion than that of sample A, 6.2 g/cm3. This indicates that the
pore density in sample A is larger than sample B, sug-
3.1. Optimization of processing conditions gesting that more liquid loss occurred in sample A.
Since a liquid loss during the melt-processing enhances
3.1.1. Pre-sintering effect before crystal growth the compositional shift toward Nd-rich direction, Nd-
In this section, we investigated the effect of sinter- rich Nd1+xBa2-xCu3O7-
has more tendency to form in
ing before melt-processing on the superconducting sample A. However, even in sample B, the density
properties and the field-trapping capabilities. The decreased after the crystal growth presumably due to
precursor powders with the composition of the generation of oxygen gas in a partially melted state,
Nd123:Nd422=10:2+10 wt.%Ag2O were used to leading to the density lower than the theoretical value.
fabricate single-domain Nd–Ba–Cu–O bulks. The Fig. 4 shows the field distributions trapped by samples
density of the precursor pellet without sintering was A and B in the remanent state, and the presence of 1T at
only 4.5 g/cm3, which was about 65% of the theoretical liquid nitrogen temperature. Compared to sample B
value of 6.96 g/cm3. Due to such a low density, the with symmetrical trapped-field distribution, sample A
samples were largely deformed after the melt-growth had the area where the flux could easily penetrate. In
and a dent was formed near the center of the surface as sample A, the flux already reached the center at 1T and
shown in Fig. 3. The density of the pellet was increased the field distribution around the center was sectioned
to 6.5 g/cm3 when sintered at 1040  C for 1 h before the into two parts. This is partly attributed to the mis-
melt-growth. Such a precursor pellet with the high den- orientation caused by the formation of subgrains, which
sity was not deformed during the melt-process. was created by dislocation walls as a result of the
We also measured the densities of melt-grown crystals entrapment of Nd422 particles in the matrix [21].
with and without pre-sintering treatment. The density of Fig. 5 shows the results of X-ray pole figure analyses
sample B with pre-sintering was 6.4 g/cm3 and higher for several positions on sample A. The positions (a) and
 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137 127

Fig. 13. Distributions of Jc–B properties at 77 K for the bulk samples with different Nd422 content: (a) sample D (Nd123:Nd422=10:1.5+10
wt.%Ag2O), (b) sample E (Nd123:Nd422=10:2.5+10 wt.%Ag2O) and (c) sample F (Nd123:Nd422=10:3+10 wt.%Ag2O). The relative locations
are presented in the inset shown in Fig. 6(a).
128 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137

Fig. 13. (continued)

(b) are close to the seed, and the positions (c) and (d) are mation during the melt-processing. Sintering a sample
located near the periphery of sample A. Furthermore, prior to melt-processing was found to be effective in
the positions (b) and (d) are near the area where the flux enhancing the superconducting properties and the field-
could easily penetrate. It is evident that the misorientation trapping capability.
angles between grains are larger in (c) and (d) than in (a)
and (b). These results also support the idea that the mis- 3.1.2. Effect of cooling rate during crystal growth
orientation is responsible for the weak-link behavior. Nd–Ba–Cu–O bulk superconductors are sensitive to
Fig. 6 shows Jc–B curves at 77 K for the small speci- the processing conditions because the growth tempera-
mens cut from various positions along a- and c- axes in ture [12] and the composition at the growth interface
samples A and B. It is clear that Jc–B properties for strongly affect the Nd/Ba substitution in Nd123 matrix.
sample A were more dispersive than those for sample B. Hence, the cooling rate during the crystal growth is
Such inhomogeneity of Jc is attributed to a liquid loss closely related to the Nd/Ba substitution. We investi-
caused by the lower density in sample A, which was not gated the effect of the cooling rate on the super-
sintered before the crystal growth. It is also notable that conducting properties and the trapped-field values.
the Jc values were very low at the locations beneath the Fig. 7 shows the trapped-field distribution of sample
seed in both samples A and B. This is attributed to the C at liquid nitrogen temperature. The field distribution
non-steady state solidification that took place at an trapped by sample B was already presented in Fig. 4(c).
early stage of crystallization [22], which may be the Both samples had the trapped-fields higher than 1T. It is
source of depressed superconducting properties caused evident that a decrease in the growth rate from 1.0 to
by the large Nd/Ba substitution in Nd123 phase. 0.3  C/h did not degrade the field-trapping capability in
Considering the results of X-ray pole-figure analyses the remanent state. Fig. 8, however, shows a clear dif-
in Fig. 5 and the fact that the Jc values for sample A ference between sample B and sample C in that the
keep increasing from the center toward the edge, it is maximum trapped-field values are plotted as a function
plausible that heavy deformation produced the area of the external field. The trapped-field values of sample
where the flux could easily penetrate. B are always higher than those of sample C. It is also
For the fabrication of Nd bulk superconductors, it is notable that the secondary peak effect is observed in
very important to avoid a liquid loss and large defor- sample B, while it is absent in sample C.
 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137 129

The trapped-field is correlated with the micro- this, the Nd/Ba chemical fluctuation is also dependent
structures and the average Jc–B properties in the entire on the growth temperature and
T, which also affects
sample. Therefore, homogeneous microstructures and the superconducting properties. Fig. 10 shows the field
Jc–B properties are important for achieving a high field- dependencies of Jc values for various locations in
trapping capability. In sample C, the grain size of sample C. Apart from the region beneath the seed
Nd422 inclusions in the matrix was strongly dependent (region A1) with the lowest Jc values due to non-steady
on the distance from the seed due to the pushing effect solidification discussed in Section 3.1 [22], it is clear that
of Nd422 particles. As shown in Fig. 9, the size sample B contains the regions exhibiting remarkable
decreased with distance from the seed both in the hor- secondary peak effect with the relatively high peak field
izontal and vertical directions. In contrast, the size dis- while some regions are observed in sample C, where the
tribution in sample B was smaller than that in sample C. Jc–B properties are gradually improved with the
These features are closely related to the amount of distance from the seed. This is due to the fact that the
undercooling (
T), where
T is the difference between range of Nd/Ba substitution depends on the under-
Tp and the growth temperature [23]. It is well known cooling, the mass valance in the bulks and the pushing/
that small particles can be trapped by the matrix as
T trapping behavior of Nd422 at the growth interface. In
becomes large. Therefore, in sample C fabricated at the the case of the cooling rate of 0.3  C/h, the crystal began
slower cooling rate of 0.3  C/h, the distribution of to grow under the smaller
T and hence the crystal
Nd422 inclusions is less homogeneous and smaller solidified at the early stage had the smaller Nd/Ba
Nd422 particles were entrapped in a large
T region far substitution in the matrix. This substitution becomes
away from the seed. Such microstructural variation large with increasing
T [12] until it is suppressed by
within the sample causes the location dependence of Jc– the mass valance. Furthermore, the amount of pushed
B properties because the average size of non-super- Nd422 particles is larger, which tends to generate
conducting particles affects the remanent Jc value and Nd-rich Nd123 phase. Accordingly, the area where
furthermore the pushing/trapping behavior of Nd422 Nd123ss phase with the relatively large Nd/Ba
changes the composition at the growth front. Besides substitution was enlarged compared with sample B.

Fig. 14. Tensile strengths of the specimens cut from the various locations in the grown samples with different Nd422 content: sample D
(Nd123:Nd422 =10:1.5+10 wt.%Ag2O), sample B (Nd123:Nd422 =10:2+10 wt.%Ag2O), sample E (Nd123:Nd422 =10:2.5+10 wt.%Ag2O) and
sample F (Nd123:Nd422 =10:3+10 wt.%Ag2O).
130 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137

Fig. 15. Weibull plots of tensile strengths for the bulk samples with different Nd422 content: sample D (Nd123:Nd422=10:1.5+10 wt.%Ag2O),
sample B (Nd123:Nd422=10:2+10 wt.%Ag2O), sample E (Nd123:Nd422=10:2.5+10 wt.%Ag2O) and sample F (Nd123:Nd422=10:3+10
wt.%Ag2O). The Weilbull coefficients of samples D, B, E and F are 3.56, 3.68, 8.61 and 1.59, respectively.

Considering the correlation with the trapped-fields 3.2. Effect of Nd422 and Ag content for the field-
already presented in Fig. 7 and the Jc–B properties, the trapping capability and the mechanical properties
average Jc–B properties in the entire samples reflected
the trapped-fields and therefore the peak effect of 3.2.1. Effect of Nd422 content on the field-trapping
sample C in the trapped-field as a function of the capability and the mechanical properties
external field was smeared out in the course of In this section, we investigated the effect of Nd422
integration of Jc values. Consequently, we must pay content on the microstructures, the superconducting
particular attention to the pushing/trapping behavior of properties and the field-trapping capabilities for Nd
Nd422 and the growth temperature, which strongly bulk samples.
affect the Nd/Ba substitution in Nd123 matrix at the Fig. 11 shows the top views and the field distributions
growth interface, in order to achieve a superior field- trapped by 10 wt.%Ag2O-doped Nd bulks with differ-
trapping ability in the Nd–Ba–Cu–O system. ent Nd422 content:Nd123:Nd422=10: 1.5 (sample D),
Furthermore, the present results in this section suggest 10: 2 (sample B presented in Fig. 3), 10:2.5 (sample E)
that for the characterization of large grain materials, and 10:3 (sample F). It is evident that the segregation of
the measurement of the trapped-field in the remanent Nd422 particles at the periphery of a bulk increased
state is not sufficient to fully characterize the materials with increasing Nd422 content, related to the magnitude
properties. In Nd system, although the trapped-field of undercooling. Fig. 12 shows the microstructures and
value was almost the same for large grain materials the average grain sizes of Nd422 particles at relative
crystallized at different cooling rates, both the Jc–B locations in the bulks. The amount of Nd422 particles
properties and the field dependence of the trapped-field trapped in the matrix increased with Nd422 content. An
were different. In this particular system, a relatively excessive addition of Nd422 caused the segregation of
larger growth rate provided better superconducting Nd422, leading to inhomogeneous microstructure in the
properties along with more uniform microstructure. bulk. As the distance from the seed increases, the
 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137 131

amount of Nd422 trapped in the matrix increased properties. In order to confirm this fact, we measured
accompanied by the size reduction of Nd422. Further- the field dependencies of Jc values at various locations
more, the composition at the growth front tends to shift in the bulk samples. Fig. 13 shows the distributions of
toward Nd-rich direction with increasing the amount of Jc–B properties for the samples with different Nd422
Nd422 inclusions pushed away from the growth inter- content. The Jc–B properties for the sample with the
face and hence the Nd/Ba substitution is enhanced in nominal composition of Nd123: Nd422=10:2+10
the matrix, which resulted in poor superconducting wt.%Ag2O (sample B) are already presented in Fig. 6(b).

Fig. 16. The maximum field values trapped by Nd bulk superconductors with different Nd422 content as a function of the external field at liquid nitrogen
temperature: sample D (Nd123:Nd422=10:1.5+10 wt.%Ag2O), sample B (Nd123:Nd422=10:2+10 wt.%Ag2O), sample E (Nd123:Nd422=10:2.5+10
wt.%Ag2O) and sample F (Nd123:Nd422=10:3+10 wt.%Ag2O). Sample D contained some large-size cracks, representing the maximum trapped-field
values as open symbols.

Fig. 17. Top view and trapped-field distribution of sample G (Nd123:Nd422=10:2+20wt.%Ag2O) at liquid nitrogen temperature.
132 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137

The more homogeneous Jc–B properties could be Nd123:Nd422=10:2.5. A further Nd422 addition
obtained with decreasing Nd422 content presumably increased the scattering of the tensile strengths and thus
because the amount of Nd422 particles pushed from the decreased the average tensile strength. Accordingly, the
growth front decreased. Furthermore, the prominent peak main contribution of Nd422 particles to the mechanical
effect was observed in sample D with the smallest Nd422 properties was an increase in the lowest tensile strength,
content. This is understandable by considering the fact leading to the homogeneous distribution of the mechan-
that the peak effect originates from the chemical variation ical properties. In our experiments, the overall average
in the matrix and therefore the sample with the largest sizes of Nd422 particles for all the samples were larger
volume of Nd123 matrix exhibited the remarkable peak than 2 mm as already presented in Fig. 12. Furthermore, a
effect. As a result, the trapped-field distributions of Nd bulk superconductor has crystallographic features, e.g.
bulks with the larger amount of Nd422 particles were less microcracks and porosity that strongly affect the
symmetrical. Unfortunately, single-domain bulk sample mechanical properties. Therefore, Nd422 particles dis-
with the composition of Nd123:Nd422=10:1.5+10 persed in the matrix did not contribute to particulate
wt.%Ag2O contained large-size cracks even though we reinforcement but suppressed the propagation of crack-
could fabricate single-domain samples for all the compo- ing. For engineering applications, the reliability of
sitions, suggesting that Nd422 particles affected the mechanical strength is very important. The Weibull dis-
mechanical properties of a bulk sample. Fig. 14 shows the tribution function F() is often used to evaluate this relia-
tensile strengths of the rectangular specimens cut from the bility. F() is expressed as follows [24]:
various locations in the grown samples with different   m 
Nd422 content. A gradual increase of the average tensile 
Fð Þ ¼ 1
exp
ð1Þ
strength was observed, and the sample with the initial 0
composition of Nd123:Nd422=10:2.5 exhibited the aver-
age tensile strength of 56.8 MPa. It is notable that the where  is the tensile strength, m is the Weibull coefficient
minimum tensile strength was largely improved with and  0 is a constant. Generally, F() is evaluated in the
increasing Nd422 content up to the composition of following logarithmic form:

Fig. 18. Field dependencies of the maximum trapped-fields for sample B (Nd123:Nd422=10:2+10 wt.%Ag2O) and sample G
(Nd123:Nd422=10:2+20 wt.%Ag2O) at liquid nitrogen temperature.
 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137 133
 
1 sample without large-size cracks is fabricated in the
lnln ¼ mln
mln0 ð2Þ
1
Fð Þ initial composition of Nd123:Nd422=10:1.5. According
to previous reports [15–17], Ag addition improved the
For convenience, F(i) is approximated by mechanical properties, however, the function of Ag
particles was not discussed in detail. Hence, we investi-
ði
0:3Þ gated the effect of Ag addition on the microstructures,
FðiÞ ¼ ð3Þ
ðn þ 0:4Þ the superconducting properties, the field-trapping cap-
ability and the mechanical properties.
where i is the order of destruction and n is the number
of samples. Fig. 15 shows the Weibull plots of tensile 3.2.2. Effect of Ag addition on the field-trapping
strengths for the bulk samples with different Nd422 capability and the mechanical properties
content. The Weibull coefficient was largely enhanced In this section, we investigated the effect of Ag2O
with increasing Nd422 content up to ratio of 2.5. It content on the field-trapping capabilities and the
should be noted that the sample with the composition of mechanical properties. Fig. 17 shows the top view and
Nd123: Nd422=10: 2.5 exhibited very high Weibull the trapped-field distribution of sample G
coefficient exceeding 8. The Weibull coefficient (Nd123:Nd422=10:2+20 wt.%Ag2O). One can see that
decreased with further addition of Nd422. This result a single grain can be grown even with this amount of Ag.
also suggests that appropriate Nd422 addition is Fig. 18 shows the field dependencies of the maximum
required to make the more uniform distribution of trapped-fields for samples B and G. Their field depen-
mechanical strength. Fig. 16 shows the maximum field dencies were very similar because there was no appreci-
values trapped by Nd bulk superconductors with differ- able difference in the microstructures and the
ent Nd422 content as a function of the external field. superconducting properties for these samples. Thus, it is
Except for sample D containing large-size cracks, since concluded that an increase in Ag content from 10 to 20
the higher trapped-field values were achieved in the wt.% did not affect the field-trapping ability.
sample with large secondary peak effect, it is probable However, the distribution of mechanical properties
that the higher trapped-field value can be obtained if the was quite different. Fig. 19 shows the tensile strengths of

Fig. 19. Tensile strengths of the specimens cut from sample G (Nd123:Nd422=10:2+20 wt.%Ag2O). The tensile strengths for the specimens cut
from sample B (Nd123:Nd422=10:2+10 wt.%Ag2O) were also depicted for comparison.
134 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137

Fig. 20. Weibull plots of tensile strengths for the bulk samples with different Ag2O content: sample B (Nd123:Nd422=10:2+10 wt.%Ag2O) and
sample G (Nd123:Nd422=10:2+20 wt.%Ag2O). The Weibull coefficient increased from 3.68 to 13.18 with further Ag2O addition.

the specimens cut from sample G. Here, the tensile ficial effect of Ag particles on mechanical properties.
strength values for sample B are also plotted in Fig. 19 Fig. 23 shows the field dependence of the trapped field
in comparison with the values for sample G. Besides for sample H. A remarkable secondary peak effect is
some improvement of the average tensile strength, the observed. It is also notable that the maximum trapped-
scattering of the tensile strength values in sample G field value reached 1.4T in the remanent state, and even
drastically decreased and also the Weibull coefficient a larger value of 1.6T in the presence of 1T at 77 K.
exceeded 13 as shown in Fig. 20. For practical applica-
tions, the materials should have the Weibull coefficient
higher than 5 [20]. Hence sample G is a desirable engi-
neering material from viewpoints of the mechanical
properties. Furthermore, Ag particles functioned as
barriers against cracking like Nd422 inclusions. This is
deduced by the fact that the average tensile strength
value increased with Ag addition. One can also observe
that large-size cracks introduced from pores are blun-
ted by Ag particles as shown in Fig. 21. Some crack
propagation also stopped at Ag particles. This also
supports the idea that Ag particles contribute to crack
blunting.
Fig. 22 shows the top view and the trapped-field dis-
tribution of the sample H with the nominal composition
of Nd123:Nd422=10:1.5+20 wt.%Ag2O. One can see
that a single-domain with no macrocracks could be Fig. 21. Large-size cracks introduced from pores are blunted or stop-
ped by Ag particles. The observed position was close to the fractured
obtained, which contrasts with the results of sample D surface of a specimen. Large black regions are pores. Large white and
with 10 wt.%Ag2O which contained several large-size gray particles are Ag particles. Small white particles in the matrix are
cracks (see Fig. 11). It is evident this is due to the bene- Nd422 particles.
 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137 135

Fig. 22. Top view and the field distribution trapped by sample H (Nd123:Nd422=10:1.5+20 wt.%Ag2O) at liquid nitrogen temperature.

Fig. 23. Field dependence of the maximum trapped-field for sample H (Nd123:Nd422=10:1.5+20 wt.%Ag2O) at liquid nitrogen temperature.

These results show that the optimization of processing in practical dimensions. Hence, for fair comparison it is
conditions and the content of Nd422 and Ag2O lead to necessary to minimize the demagnetizing effect by
the production of a good quality Nd–Ba–Cu–O large- increasing the thickness of the sample. The effective
grain sample with the trapped-field exceeding 1T at 77K thickness of a bulk superconductor can be increased
along with a high reliability of the mechanical strength. simply by stacking different blocks.
Fig. 24 shows the remnant trapped-field distribution
3.3. Higher trapped-field for future engineering when the three samples are stacked. The data of 1.4T
plotted at 10 mm thickness is the one for sample H.
The demagnetizing effect has always a negative influ- Then we stacked the sample about 5 mm in thickness,
ence on the trapped-field value of bulk superconductors and the trapped-field was increased to 1.6T. This is due
136 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137

Fig. 24. (a) The maximum trapped-field values as a function of the thickness of stacked samples at liquid nitrogen temperature. The effective
thickness of samples was changed by stacking the different bulks. The top sample was sample H with the trapped-field of 1.4T in the remanent field
at 77 K. (b) The field distribution trapped by stacked samples 30 mm in total thickness in the remanent field.

to a reduction in the demagnetizing effect. The trapped- 4. Conclusions

field value was slightly enhanced to 1.67T at 77 K where
th>e total thickness was 30 mm, but an increase was Nd–Ba–Cu–O bulk superconductors were fabricated
very small. It has been reported that the trapped-field with different initial compositions under different pro-
can be increased by increasing the thickness, but satu- cessing conditions in 1%O2-Ar atmosphere in order to
rates at a certain thickness [25]. Our results can be improve the field-trapping capability and the mechan-
understood in terms of the saturation of the trapped- ical properties. Low density samples suffered from a
field. These results show that this sample has an ability large deformation during processing and the areas
to trap 1.67T at 77 K. where the flux could easily penetrate. Such a problem
 M. Matsui, M. Murakami / Ceramics International 29 (2003) 117–137 137

could be overcome by sintering a sample before the References

crystal growth. With this treatment, we could also
reduce a liquid loss causing the composition shift [1] K. Salama, V. Selvamanickam, L. Gao, K. Sun, Appl. Phys. Lett.
54 (1989) 2352–2354.
toward the Nd-rich direction in Nd123 phase. Fur-
[2] M. Murakami, M. Morita, K. Doi, K. Miyamoto, Jpn. J. Appl.
thermore, the Nd/Ba substitution ratio in Nd123 phase Phys. 28 (1989) 1189–1194.
is closely related to the growth temperature, the com- [3] M. Matsui, N. Sakai, S. Nariki, S.J. Seo, M. Murakami, Super-
position at the growth interface and the mass valance cond. Sci. Technol. 13 (2000) 660–664.
in a bulk. A relatively large cooling rate during the [4] I. Monot, J. Wang, M.P. Delamare, S. Marinel, M. Hervieu,
crystal growth and smaller Nd422 addition suppressed J. Provost, G. Desgardin, Physica C 282–287 (1997) 507–508.
[5] S. Nariki, N. Sakai, M. Murakami, Physica C 357 (2001) 811–
the volume in which Nd substitutes on Ba site, result- 813.
ing in a more symmetrical trapped-field distribution [6] M. Muralidhar, M.R. Koblischca, P. Diko, M. Murakami, Appl.
with the enhancement of the maximum trapped-field Phys. Lett. 76 (2000) 91–93.
value. [7] D.G. Hinks, L. Soderholm, D.W. Capone II, J.D. Jorgensen,
For the achievement of homogeneous mechanical I.K. Schuller, C.U. Segre, K. Zhang, J.D. Grace, Appl. Phys.
Lett. 50 (1987) 1688–1690.
properties, an addition of appropriate amount of Nd422 [8] M. Murakami, Appl. Supercond. 6 (1998) 51–59.
and Ag was required. Nd422 and Ag particles could [9] S.I. Yoo, R.W. McCallum, Physica C 210 (1993) 147–156.
suppress crack propagation, leading to the improvement [10] K. Takita, H. Katoh, H. Akinaga, M. Nishino, T. Ishigaki,
of the minimum mechanical strength. These normal H. Asano, Jpn. J. Appl. Phys. 27 (1988) 57–60.
[11] S.I. Yoo, N. Sakai, H. Takaichi, T. Higuchi, M. Murakami,
conducting inclusions however did not contribute to
Appl. Phys. Lett. 65 (1994) 633–635.
particulate reinforcement of a bulk sample because their [12] K. Osamura, W. Zhang, Z. Metallkd 84 (1993) 522–529.
average size was too large. [13] Y. Ren, R. Weinstein, J. Liu, R.P. Sawh, C. Foster, Physica C
As a result of optimized processing conditions, we 251 (1995) 15–26.
could obtain the bulk sample exhibiting a trapped-field [14] G. Fuchs, G. Krabbes, P. Schatzle, S. Gruss, P. Stoye, T. Staiger,
of 1.4T in the remnant state at 77 K. The same sample K.H. Muller, J. Fink, Schultz, Appl. Phys. Lett. 70 (1997) 117–
119.
could trap 1.6T in the presence of 1T due to the promi- [15] T. Miyamoto, K. Nagashima, N. Sakai, M. Murakami, Super-
nent fishtail effect. The trapped field in the remanent cond. Sci. Technol. 13 (2000) 816–819.
state increased to 1.67T when the field was monitored [16] D. Lee, K. Salama, J. J. Appl. Phys. 29 (1990) 2017–2019.
on the surface of stacked samples 30 mm in thickness. [17] J. Joo, S.B. Jung, W. Nah, J.Y. Kim, T.S. Kim, Cryogenics 39
Furthermore, optimally processed bulk samples showed (1999) 107–113.
[18] W. Lo, N.H. Babu, D.A. Cardwell, Y. Shi, D.M. Astill, J. Mater.
a high reliability in the mechanical strength, where the Res. 15 (2000) 33–39.
Weibull coefficient and the average tensile strength [19] S. Goshima, N. Sakai, M. Takahashi, T. Higuchi, T. Murayama,
exceeded 13 and 50 MPa. K. Iida, S.I. Yoo, M. Murakami, Advances in Superconductivity
8 (1996) 739–742.
[20] N. Sakai, A. Mase, H. Ikuta, S.J. Seo, U. Mizutani,
M. Murakami, Supercond. Sci. Technol. 13 (2000) 770–773.
Acknowledgements [21] P. Diko, S. Takebayashi, M. Murakami, Physica C 297 (1998)
216–222.
The author is grateful to Dr. Sakai, Dr. Nariki and [22] M. Kambara, K. Miyake, K. Murata, T. Izumi, Y. Shiohara,
Dr. Ogasawara for fruitful discussions. This work was T. Umeda, Physica C 330 (2000) 191–202.
supported by the New Energy and Industrial Technology [23] A. Endo, H.S. Chauhan, E. Egi, Y. Shiohara, J. Mater. Res. 11
(1996) 795–803.
Development Organization (NEDO) as Collaborative [24] Japanese Industrial Standards (JIS) R 1625.
Research and Development of Fundamental Technolo- [25] H. Fukai, M. Tomita, M. Murakami, T. Nagatomo, Supercond.
gies for Superconductivity Applications. Sci. Technol. 15 (2002) 1–4.