CERN-ACC-2020-0021
17/09/2020
CERN – EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH
PERFORMANCE OF M
G
B
2
SUPERCONDUCTOR DEVELOPED FOR
HIGH-EFFICIENCY KLYSTRON APPLICATIONS
H. Tanaka
3
, T.Suzuki
3
, M. Kodama
3
, T. Koga
3
, H. Watanabe
3
, A. Yamamoto
1,2
and S. Michizono
2
1
CERN, Geneva, Switzerland
2
KEK, Tsukuba, Japan
3
Hitachi, Tokyo, Japan
Abstract
An 8-km long MgB
2
wire for a prototype klystron magnet was made and evaluated. The wire was
made by a typical in situ method; it has 10 filaments and 0.67 mm in outer diameter. The homogeneity of
I
c
of this wire was evaluated by several methods. Deviation of I
c
values in short sample wires was very
small. In addition, the current sharing temperature of the MgB
2
magnet (made of two reels of wire 2.9 km
long each) agreed well with the estimated value of the I
c
-B-T properties in short sample wires. Based on
the obtained results, it can be said that the I
c
properties of the entire wire length are quite uniform.
Presented at the Magnet Technology Conference, Vancouver, Canada, 23-27 Sep 2019
Geneva, Switzerland
September 2020
CLIC – Note – 1160
IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 30, NO. 4, JUNE 2020 6200105
Performance of MgB
2
Superconductor Developed for
High-Efficiency Klystron Applications
Hideki Tanaka , Takaaki Suzuki, Motomune Kodama, Tomoyuki Koga, Hiroyuki Watanabe , Akira Yamamoto,
and Shinichiro Michizono
Abstract—An8-kmlongMgB
2
wire for a prototype klystron
magnet was made and evaluated. The wire was made by a typical
in situ method; it has 10 filaments and 0.67 mm in outer diame-
ter. The homogeneity of I
c
of this wire was evaluated by several
methods. Deviation of I
c
values in short sample wires was very
small. In addition, the current sharing temperature of the MgB
2
magnet (made of two reels of wire 2.9 km long each) agreed well with
the estimated value of the I
c
-B-T properties in s hort sample wires.
Based on the obtained results, it can be said that the I
c
properties
of the entire wire length are quite uniform.
Index Terms—Critical current, high energy efficiency,
homogeneity, MgB
2
wire.
I. INTRODUCTION
T
HE critical temperature of magnesium diboride is 39 K [1],
which allows equipment containing MgB
2
wires and mag-
nets t o be made highly energy-efficient and liquid helium-free.
Magnetic resonance imaging (MRI) using MgB
2
wires 1.6 km in
length made by the ex situ method has already been implemented
as the first practical use of MgB
2
[2]. Ex situ MgB
2
wires of over
3 km long were developed for making a magnet [3]. MgB
2
wires
made by in situ method were also reported. A coil made with a
300-m long wire [4] and a coil with a 1.7-km long wire for MRI
use [5] were successfully implemented. The use of MgB
2
is not
limited to magnets; MgB
2
wires have been used as power cables
in superconducting (SC) links at CERN [6] and in the Best Paths
project [7].
For klystron use, we need DC solenoid magnets for focus-
ing the electron beam. The power efficiency of the magnet is
important because in an design option of the Compact Linear
Collider 380 GeV (CLIC-380 GeV), the number of klystron
magnets reaches 4,000–5,000 [8]. The power consumption of
a typical copper magnet for klystron applications is 20 kW for
Manuscript received September 20, 2019; accepted January 20, 2020. Date
of publication January 30, 2020; date of current version February 14, 2020.
(Corresponding author: Hideki Tanaka.)
H. Tanaka, T. Suzuki, and M. Kodama are with the Research and Develop-
ment Group, Hitachi Ltd., Hitachi 319-1292, Japan (e-mail: hideki.tanaka.cj@
hitachi.com).
T. Koga and H. Watanabe are with the Hitachi Works, Hitachi Ltd., Hitachi
317-8511, Japan.
A. Yamamoto is with the High Energy Accelerator Research Organization
(KEK), Tsukuba 305-0801, Japan, and is also with European Organization for
Nuclear Research (CERN), Geneva 1211, Switzerland.
S. Michizono is with the KEK, Tsukuba 305-0801, Japan.
Color versions of one or more of the figures in this article are available online
at https://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TASC.2020.2970391
cooling the Joule heat of the magnet [9], and in the case of a
Nb-Ti superconducting magnet without liquid helium, the AC
plug power is 6 kW, as shown in a previous study [10]. To achieve
high efficiency magnets for klystrons (klystron magnets), we
developed an MgB
2
wire and a magnet that can be operated at
high temperatures and have low power consumption.
In the design of the MgB
2
prototype solenoid magnet, to
achieve high efficiency, it is important to reduce heat penetration
from room temperature to low temperature (superconducting
coils) and to make the current leads of the magnet finer [11].
We chose the outer diameter of the MgB
2
wire to be 0.67 mm,
thinking about workability of the coil winding. In addition, as
the magnetic field at the magnet center is 0.7–0.8 T, two reels
of 2.9-km long MgB
2
wires were needed. The homogeneity
of MgB
2
wires is a very important characteristic, but there are
not so many previous studies evaluating km-class MgB
2
wires.
A 1.7-km long MgB
2
wire was wound as a coil and the good
homogeneity of this wire was shown [5].
We made and cut an 8-km long wire to obtain two wires of
2.9 km length. We measured the I
c
values of short wires sampled
from the ends of these 2.9 km lengths i.e., at the 2250 m and
5150 m positions in the 8 km length, and then evaluated the I
c
homogeneity. The data of I
c
vs. the longitudinal position in the
MgB
2
wire, in which a unit length is several km long, has not
been reported in previous studies. These data can contribute to
further research and practical applications of MgB
2
wires and
magnets. A MgB
2
klystron magnet made by the Wind&React
method was successfully operated, and the I
c
homogeneity of
the wire as a magnet was evaluated.
II. E
XPERIMENTAL DETAILS
A. 8-km Long MgB
2
Wire
The MgB
2
wire for the klystron magnet, with a unit length
of 8 km, was made by the in situ method as follows. Magne-
sium powder (>99.8%) and boron powder (>98.5%, <250 nm;
Pavezyum nano Boron) were mixed at a molar ratio of Mg: B =
1: 2 without a carbon dopant, because at high temperatures and
in low magnetic fields a pure MgB
2
wire has higher J
c
than a
carbon-doped MgB
2
wire [12]. Ten filaments with a Fe barrier
sheath and a Cu bar as a stabilizer were embedded into a Monel
sheath, then cold-worked to be 0.67 mm in outer diameter. The
unit length reached 8085 m at 0.67 mm diameter, and the wire
was cut at the 150 m, 2250 m, 5150 m, and 8050 m positions as
shown in Fig. 1.
1051-8223 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: Hitachi Ltd.. Downloaded on February 26,2020 at 02:51:06 UTC from IEEE Xplore. Restrictions apply.
6200105 IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 30, NO. 4, JUNE 2020
Fig. 1. Cross-sections of the MgB
2
wire. A-, B-, C-, D-, and E- ends are
sampled from 0 m, 2250 m, 5150 m, 8050 m, and 8085 m positions respectively.
The lengths of the five resulting MgB
2
wires were 150 m
as Wire-A, 2100 m as Wire-B, 2900 m as Wire-C, 2900 m as
Wire-D, and 35 m as Wire-E. Two 2900-m long wires, Wire-C
and Wire-D, were insulated with a glass braid and served for
making the klystron magnet by the Wind&React method. The
thickness of the insulation is typically 80 μm. If you need thinner
insulation, Al
2
O
3
layer can be used as the insulation as described
in [13]. The ends of Wire-C are identified as B-end (2250 m)
and C-end (5150 m), and those of Wire-D are C-end (5150 m)
and D-end (8050 m).
Fig. 1 s hows cross-sections at each end; almost the same cross-
sections were obtained from 0 m to 8085 m of the wire. The
MgB
2
filling factors of cross-sections obtained from B-, C-,
and D- ends were 29.2%, 28.7%, and 28.9%, respectively. The
sintering condition of this wire was 600°C for 6 h.
B. Measurement of I
c
and Its Deviation in Short Sample Wires
The operating temperature of the klystron magnet was esti-
mated as 20 K or higher, so the I
c
-B-T properties of short samples
picked from near the B-, C-, and D- ends were measured at
20 K, 25 K, and 30 K. The I
c
values were measured by the
typical four-probe method. The length of all short sample wires
was 35 mm; the temperature of the wires was controlled using
helium gas cooling and a heater. The criterion of the electric field
was 1 μV/cm, and the distance between voltage taps was 5 mm.
I
c
values obtained from these three samples show the long-range
homogeneity.
The short-range uniformity was evaluated by the I
c
distri-
bution as follows. To evaluate the I
c
distribution of the wire,
a lot of I
c
data are required, so the distribution was obtained
by I
c
measurement with liquid helium bath cooling. The I
c
values were measured by the typical four-probe method. The
length of all short sample wires was 50 mm. The criterion of the
electric field was 1 μV/cm, and the distance between voltage
taps was 5 mm. The number of samples was twenty-one, and
they were sampled from near the D-end. From the I
c
values of
these samples, the standard deviation σ was calculated, and the
I
c
distribution was evaluated with the normal distribution as the
short-range homogeneity.
Fig. 2. (a) J
e
(J
c
)-B-T properties obtained from three short sample wires.
(b) Index-n obtained from V-I curves on I
c
measurement.
C. Homogeneity Evaluation of the MgB
2
Wire in a Magnet
Unlike in high-temperature coated conductors, the I
c
values of
the MgB
2
wire cannot be measured along the wire’s longitudinal
direction with liquid nitrogen bath cooling. Instead, we can
evaluate the homogeneity of the wire by making a magnet and
measuring the current sharing temperature T
cs
of the magnet.
Here, T
cs
means the temperature at which some resistive voltage
is generated in a part of the coil winding.
The method of measuring T
cs
is written in the reference [11]
and can be summarized as follows. First, achieve a steady state
by bringing the coil winding to a certain temperature at a certain
operating current I
op
. Here, the coil winding was conduction-
cooled by a cold head and the temperature distribution in the
coil winding was estimated to be smaller than 1 K. Next, raise
the temperature slowly and obtain the temperature at which
some voltage generates resistivity of the normal zone in the
coil winding. Then, the I
op
dependence of T
cs
(T
cs
-I
op
) can
be obtained by changing I
op
. We can also estimate T
cs
-I
op
from
the I
c
-B-T of the short sample wires. If the measured values
agree with the estimated ones, it means that almost all the wires
wound as a magnet have the same I
c
-B-T properties.
Main specifications of the magnet are as follows. The inner
diameter of the coil: 337 mm; the coil length: 136.6 mm; the
number of turns in a coil: 2432, and the number of coils: 2.
Details of these specifications are also written i n reference [11].
III. R
ESULTS AND DISCUSSION
A. I
c
Values Measured at 20-30 K
Fig. 2(a) shows the results of I
c
measurement obtained at
20–30 K. Distances from B-end to C-end and from C-end to
Authorized licensed use limited to: Hitachi Ltd.. Downloaded on February 26,2020 at 02:51:06 UTC from IEEE Xplore. Restrictions apply.
TANAKA et al. : PERFORMANCE OF MGB
2
SUPERCONDUCTOR DEVELOPED FOR HIGH-EFFICIENCY KLYSTRON APPLICATIONS 6200105
Fig. 3. I
c
distribution measured with short wire samples from near D-end of
the 8-km MgB
2
wire. I
c˙min
was assumed as I
c
_
ave
–4.5σ.
D-end were both 2900 m. The J
e
and J
c
were calculated from
the cross-sectional area of the wire, filling factors of MgB
2
and
I
c
values. Deviation in J
e
-B-T in these three samples at 20 K
has been registered, but at 25 K and 30 K the properties were
almost the same. These results show the possibility of superior
homogeneity at practical temperatures across approximately 6
km of the wire length.
Fig. 2(b) shows the n values obtained from V-I curves for the I
c
measurements shown in Fig. 2(a). There are some fluctuation on
B dependences of n-values, but we can roughly summarise that
the value n is 30 or higher in the condition of J
c
= 1 kA/mm
2
.
B. I
c
Distribution Evaluated at 4.2 K
Fig. 3 shows the I
c
distribution at 4.2 K and 7.0 T measured
with twenty-one short wires sampled every 70 mm from near
the D-end. The averaged value of I
c
(I
c
_
ave
) was 20.1 A and the
standard deviation (σ) was 0.72 A, which was 3.6% of I
c
_
ave
.If
the minimum value of I
c
(I
c
_
min
) is defined as I
c
_
ave
–4.5σ,the
I
c
_
min
is 16.9 A and 84% of I
c
_
ave
. According to the definition
of the normal distribution, the probability that the sampled wires
have lower I
c
than I
c
_
min
is 0.00035%. Based on these results,
we can assume that the I
c
_
min
of the entire 8-km wire is 84% of
I
c
_
ave
.
C. I
c
Homogeneity of the MgB
2
Wire Evaluated in a Magnet
According to the design of the klystron magnet using Wire-C
and Wire-D, the rated operational current is 57.1 A, and the
maximum magnetic field of the coil winding is 1.06 T [11].
Fig. 4 shows a comparison between the coil load line of the
magnet and the I
c
-B-T property obtained from a short sample
wire at the C-end. From this graph, we can estimate that the T
cs
of the magnet at I
op
= 57.1 A is 29 K. To estimate T
cs
with
high accuracy, the B values of the short sample wire (from the
C-end) at I
c
= 57.1 A and T = 20, 25, and 30 K were read out
fromFig.2.Thenthe(B, T) relation at I
c
= 57.1 A was plotted
in Fig. 5. The dashed line shows the fitting curve of the (B, T)
relation at I
c
= 57.1 A. Here, the maximum magnetic field in
the coil winding in the case of I
op
= 57.1 A is 1.06 T, so we
can calculate from t he fitting function written in Fig. 5 that T
cs
at I
op
= 57.1 A is 29.5 K. We can also estimate the T
cs
values
at other operating currents using the same method.
Fig. 4. I
c
− B − T properties measured from short sample wire (C-end) and
coil load line of the klystron magnet. Operational current is 57.1 A and maximum
magnet field is 1.06 T.
Fig. 5. B vs. T at I
c
= 57.1 A obtained from I
c
-B-T properties of short sample
wire (C-end). The dashed line shows a fitting quadratic fitting function.
Fig. 6. Measurement results of temperature dependence of resistivity of short
sample wires.
AsshowninFig.5,T
cs
at 0 T is expected to be 35.9 K.
This estimation was confirmed by additional measurements as
follows. The temperature dependence of the resistivity of a
short sample wire was measured by the AC four-probe method
(16 Hz, 10 mA) with the Quantum Design Physical Property
Measurement System. Fig. 6 shows the results of measurements
of the samples from the B-end and D-end whose lengths were
both 10 mm. The vertical axis shows resistivity normalized by
the value at 300 K. As shown in Fig. 6, 36.0 K is the threshold
temperature for a measurable non-zero resistivity, and this value
agrees well with the expected value from the fitting function
shown in Fig. 5. Therefore, we can also estimate the T
cs
values
in the magnetic field range of 1 T or less.
Authorized licensed use limited to: Hitachi Ltd.. Downloaded on February 26,2020 at 02:51:06 UTC from IEEE Xplore. Restrictions apply.
6200105 IEEE TRANSACTIONS ON APPLIED SUPERCONDUCTIVITY, VOL. 30, NO. 4, JUNE 2020
Fig. 7. Comparison of measured and estimated values of T
cs
.
Fig. 7 shows I
op
dependences of estimated and measured
T
cs
values (T
cs
-I
op
). The amplitudes of the error bars on the
estimated values reflect the deviations of I
c
-B-T in the three
samples shown in Fig. 2. The T
cs
were measured under a few
values I
op
. The largest I
op
at T
cs
measurement was 63 A, which
was larger than the rated operational current (57.1 A), and the
smallest one was 10 mA. All measured T
cs
values agreed well
with the estimated values within 1 K. This means that there is
no I
c
degradation in the coil winding, and the I
c
-B-T properties
of this wire are quite uniform. The reason for the diff erence
between the measured T
cs
values and the estimated ones could
be the delay in temperature measurements.
As shown in Fig. 7, there is a 9 K or larger temperature margin
at the rated I
op
= 57.1 A, if the magnet is operated at 20 K as
the t emperature of the coil winding. According to the results
of the quench tests, this temperature margin may be reduced
[11]. Therefore, the magnet can be used at higher operational
temperature, e.g., 25 K.
D. Next Steps
The I
c
variation of the MgB
2
wire was small enough, and
the magnet performance represented by T
cs
was nearly equal to
the estimated value. It can be said that the MgB
2
wire has ideal
homogeneity for the Wind&React method.
The ideal process for making MgB
2
magnets is said to be the
React&Wind method, so we should develop MgB
2
wires which
can be used for the React&Wind method without I
c
degradation.
In the previous study, the bending tolerance of the MgB
2
wire at
room temperature was the same as the one reported in this paper;
the reversible bending radius of the wire sintered at 600°C for
6 h was 137 mm [14]. On the other hand, the inner radius of
the coil for the klystron magnet is 168 mm. This indicates that
the klystron magnet can be made with the MgB
2
wire using the
React&Wind method.
In general, superconducting wires are expected to be used in
magnetic fields higher than 1.0 T. For example, according to the
previous studies, the maximum magnetic fields of 1.5 T MgB
2
MRI are 3.0 T [15] and 2.7 T [16]. For using MgB
2
wire in such
intermediate magnetic fields, a carbon additive should be mixed
into MgB
2
filaments. The J
c
value at 20 K of the MgB
2
wire
with 3% carbon additive made by Hitachi Ltd. is higher in the
range of external magnetic fields of 2.3 T or more [12].
Fig. 8 shows the J
e
properties of several MgB
2
wires at 4.2 K.
Here, FF means the filling factor of MgB
2
in the cross-section
of each wire, and Pure or Dope means non-carbon doped or
Fig. 8. Engineering J
c
of MgB
2
wires measured at 4.2 K. Wire for klystron
is the wire reported in this paper. J
e
values of IN-30, IMD-19 and EX-37 were
shown in a previous study [17]. The C-doped 1.5 mm wire was made by the
in situ method and reported in our past study [4].
carbon-doped MgB
2
wire. IN-30, IMD-19 and Ex-37 are the
typical MgB
2
wires made by in situ, IMD and ex situ method,
respectively [17]. The wire for klystron magnets has higher J
e
compared to other typical wires in 6.5 T or lower magnetic field.
In addition, C-doped 1.5 mm wire was made by Hitachi Ltd. [4]
and it has the highest J
e
between 6 T to 7 T. It is expected that
J
e
values of the C-doped 1.5 mm wire are higher than those of
the wire for klystron applications in 3 T or lower magnetic field
mainly due to the carbon additive.
In this study, the MgB
2
klystron magnet was made by the
Wind&React method with non-carbon doped MgB
2
wire. I n the
next step, we will make a magnet by the React&Wind method.
If a higher magnetic field is needed, the carbon-doped MgB
2
wire will be selected.
IV. S
UMMARY
The homogeneity of the 8-km long MgB
2
wire was investi-
gated by three measurements: (i) I
c
-B properties at 20–30 K of
three short wires sampled from every 2900 m, (ii) I
c
distribution
at 4.2 K of twenty-three short wires sampled from every 70 mm,
and (iii) T
cs
-I
op
at the klystron magnet made with two reels of
2.9-km long wires. The results of evaluation of the homogeneity
is summarized as the following four points:
1) Short wires sampled every 2.9 km have almost the same
I
c
-B-T properties at 20–30 K. This shows possibility of
superior homogeneity across the wire of approximately 6
km in length.
2) Standard deviation of I
c
values at 4.2 K was 3.6% of
the average value of I
c
, I
c
_
ave
. The minimum value of I
c
assumed as I
c
_
ave
–4.5σ is expected to be 84% of I
c
_
ave
.
3) The T
cs
values measured for the klystron magnet agreed
well with the estimated values from I
c
-B-T properties of
the short sample wires.
4) It can be said that the performance of the 8-km long wire
was uniform enough to make MgB
2
coils and magnets.
A
CKNOWLEDGMENT
The authors heartily thank the Cryogenic Station, Research
Network, and Facility Services Division, National Institute for
Materials Science, Japan, for the support with I
c
measurement.
Authorized licensed use limited to: Hitachi Ltd.. Downloaded on February 26,2020 at 02:51:06 UTC from IEEE Xplore. Restrictions apply.