ren1995

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

PHYSICA

ELSEVIER Physica C 251 (1995) 15-26

Damage caused by magnetic pressure at high trapped field


in quasi-permanent magnets composed of melt-textured
Y-Ba-Cu-O superconductor
Y. Ren a,*, R. Weinstein a, j. Liu a, R.P. Sawh a, C. Foster b
a Institute for Beam Particle Dynamics and Texas Center for Superconductivity, University of Houston, Houston, TX 77204-5506, USA
b Indiana University Cyclotron Facility, Bloomington, IN 47405, USA
Received 28 March 1995; revised manuscript received 14 June 1995

Abstract

Quasi-permanent magnets made of melt-textured Y B a 2 C u 3 0 7_ 8 (MT-YBCO) superconductor can now trap multi-tesla
fields, B r The interaction of the trapped field and the critical current causes an outward pressure, proportional to B2, which
can crack the magnet. We have done an experiment to observe such cracking in a mini-magnet fabricated from four
MT-YBCO discs activated at 49 K using an applied field of 14 T. We have compared the results to existing theories which
describe magnetic pressure in a trapped-field magnet (TFM) previously activated. We find that a modification is needed to
describe magnetic pressure during the process of activation. We present the experimental results and the expanded theory,
based on the simple Bean model. Theory and experiment show good agreement. We find that cracking is more likely during
activation, and conclude that 10 T is achievable in TFM's composed of present materials. Cracking is most probable at the
center of a TFM, with the cracks running radially outward.

1. Introduction results are summarized in the books by Montgomery


[1] and by Wilson [2].
It is well known that in wire-wound solenoidal
If a type-II superconductor sample is placed in an
electromagnets, whether superconducting or conven-
externally applied magnetic field Ba, and then the
tional, the Lorentz force interaction between the
applied field is removed, a portion of the field
current and the magnetic field results in stresses
remains trapped inside the sample. This effect can be
within the coil. When the field and the current
employed to make trapped-field magnets (TFM's)
density are both high, these stresses can be very
from bulk superconductor. The trapped field B t is
large, and can burst the coil radially outward. This is
maintained by persistent currents within the super-
one of the problems of major concern in the design
conductor. Therefore, a similar current-field interac-
of wire-wound solenoid superconducting magnets.
tion and similar forces and stresses will be present in
Several authors have calculated the magnetic forces
a TFM as in a solenoid.
and stresses inside a wire-wound solenoid. Their
Our group has been studying TFM's since 1988,
[3-11]. At present we are studying prototype mag-
* Corresponding author. nets using discs of melt-textured YBa2CU3OT_ ~

0921-4534/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved


SSDI 0 9 2 1 - 4 5 3 4 ( 9 5 ) 0 0 3 9 8 - 3
16 Y.Ren et aL /Physica C 251 (1995) 15-26

(MT-YBCO) materials with a diameter ~ 2 cm and


thickness ~ 0.8 cm. After light-ion (p÷ or He ÷÷)
irradiation, these discs can sustain a critical current
density of the order of 3 to 4 × 1 0 4 A / c m e at 77 K
in their self-field. Mini-magnet fabrications, made up
of four such discs have achieved trapped fields of
2.25 T at 77 K, and 5.3 T at 65 K [11]. Most
recently, another four-disc magnet, which will be
discussed elsewhere, has achieved a new record-high
trapped field of 8.33 T at 54 K. These values are
comparable to the fields of conventional supercon-
ducting solenoid magnets. An experimental study Fig. 1. Trapped-fieldprofile at 77 K for the upper surface of a
was conducted to observe cracking due to magnetic MT-YBCO disc. This disc was later used as the top piece of the
stresses for such magnets. The intent of the study four-disc mini-magnet.
was to indicate present limitations on trapped field,
and possible means of improvement for the future.
Sumption and Collings [12] first calculated the the trapped-field profile at 77 K by cooling the
stress distribution inside a cylindrical sample con- sample with liquid nitrogen in an applied field of 1.9
taining maximum trapped magnetic field B t .... • T (field cool). After removal of the applied field, an
Independently, we did the calculations for the axial Hall sensor (F.W. Bell, model BHA 921), 1
same case. When we performed the experiment on mm above the disc surface, was used to scan the
damages due to magnetic stress, it became clear that entire upper surface of the disc. All four discs showed
magnetic pressure d u r i n g the activation process, not a maximum trapped field Bt,ma x ~ 1 T, which corre-
covered by Sumption and Collings or by our earlier sponded to a critical current density J¢ ~ 3 x 1 0 4
work, had to be considered. During activation, both A / c m 2. Fig. 1 shows the trapped-field profile for
the current and the field are changing, and at a one of the discs. The single peak feature is in
certain stage of the activation, higher stresses result, excellent agreement with the field distribution ex-
as will be shown below. Since a TFM must survive pected theoretically for a single grain [5,6].
the activation in order to achieve Bt, max, a n extension The four discs were then glued together with GE
of the theory is required in order to compare to varnish to form a mini-magnet, with a transverse
experiment. In this paper, we will report on our Hall sensor (F.W. Bell model BHT 921) between
experimental and theoretical studies for this problem, two central discs. Also between the discs was a Lake
and will compare our experimental observations with Shore Cernox temperature sensor (model CX-1070-
the extended theoretical calculations. SD), which was used to monitor the temperature of
the mini-magnet. Two standard Pt temperature sen-
sors were also attached, one on the top and one on
2. Experiment the bottom surfaces of the mini-magnet, in order to
observe possible temperature variations over the di-
The TFM studied in this experiment was fabri- mensions of the assembly.
cated from four high-quality single grain MT-YBCO The mini-magnet fabrication was activated using
discs with dimensions ~ 2 0 mm × 8 mm. The discs a 27 T resistive magnet at the National High Mag-
were processed by a melt-texture technique utilizing netic Field Laboratory (NHMFL), Tallahassee,
excess Y, Pt,. and SmBa2Cu307_ 8 seeds [11], with Florida. A variable-temperature cryostat, inserted in
the c-axis parallel to the axis of the disc. Then the the magnet, allowed us to vary the temperature of the
discs were bombarded with 200 MeV protons, to a mini-magnet between 4.2 K and 300 K. For each
fluence of 2.5 × 1 0 1 6 p + / c m 2, at the Indiana Uni- test, the mini-magnet was first held at 100 K (i.e.
versity Cyclotron Facility. For each individual disc, above Tc ~ 92 K) while the resistive magnet was
initial characterization was obtained by measuring energized to a field Ba, ma x . It was then cooled down
Y. Ren et al. / Physica C 251 (1995) 15-26 17

to the desired experimental temperature T (T < To).


As soon as the temperature goal was reached and j!! ii 'iiiiii
thermal equilibrium was achieved, the applied field
was swept down to zero at a rate of 0.15 T / m i n . The " • .......... ,t... : : :_%.,... .....
::::::::::::::::::::::::::::::::::::::
x ........

magnetic field at the center of the mini-magnet was iilill E ii~iiii?~iiiiiiii~'.'~iiiiiiiii~

recorded every 30 s. Discrepancies among the three :::::::::::::::::::::::::::::::::::::

temperature sensors were less than 0.5 K.


The damage to the mini-magnet occurred as the
applied field was being ramped down, after the
sample had been cooled to 49 K in an applied field a b
of 14 T. Fig. 2 shows the measured magnetic field at Fig. 3. Trapped-field profile at 77 K for the top disc of the
the center of the mini-magnet as a function of time. mini-magnet after the 49 K, 14 T test. (a) 3D representation
showing the multi-peakcharacter; (b) 2D representation showing
When the applied field reached 8.15 T, the magnetic
the positions of the peaks (O) and the valleys(X).
field at the center suddenly dropped from 13.48 T to
8.28 T. After that, the trapped field was observed to
be only 1.91T when the applied field was zero. In
previous activations this same fabrication had used for characterization. Instead of the single-peak
achieved B t --- 8.33 T at 54 K. result we had found before the tests (see Fig. 1), all
In order to check whether the sample had broken, f o u r of the component discs showed multiple peaks.
we immediately conducted a repeat of a test previ- As an example, Fig. 3(a) shows the mapping profile
ously done at 76.5 K with 5 T activation, in which for the top disc of the mini-magnet. It is seen that
the mini-magnet had trapped B t = 2.58 T. In the there are three peaks of roughly the same height. The
repeat test, only 0.56 T was trapped. It appeared that valleys between these peaks are plotted in Fig. 3(b).
the sample had suffered damage. It should be noted We note that the valleys between the peaks meet
that visual inspection of the mini-magnet assembly, near the center of the disc, and approximately follow
and of the individual discs failed to reveal any the direction of the radius. We interpret these results
obvious damage, such as cracks on the surfaces. as cracking of the disc near the center, followed by
We later examined each disc by activating at 77 radial crack propagation. Since we cannot see any
K, and using the same mapping technique originally cracks on the surfaces, these valleys represent cracks
deep inside the sample. What was a unified melt-tex-
tured disc had become three regions, each carrying
15 i , i i I ' i i J I i i i , I i i , i I i i ~ i
high current, but joined by weak links. Of the re-
(8 maining three discs, one cracked into three regions,
I- sured Field as the top one had done, one cracked into two almost
v
-o10 equal regions, with the crack running right through
.m
the center, and the last disc had three cracks, one
LL running through the center.
AppliedFi~eld,,,~ ,~ , , ~~, ,, ,~
"" 5
t-"

3. Theoretical calculations

In order to understand the damage to the TFM


0 20 40 60 80 100 discs, as observed in the experiment, we calculate
Time (min) the stresses induced by the interaction between the
magnetic field and the persistent current inside a
Fig. 2. Magnetic field at the center of the mini-magnet, as a
function of time, when the applied field is ramped down from 14
cylindrical TFM. We assume, as a model:
T to zero at 49 K. The straight line represents the applied field (1) The TFM is a long cylinder with the c-axis of the
which is ramped down linearly. crystal lattice as the axis of symmetry (z-axis).
18 Y. Ren et al. /Physica C 251 (1995) 15-26

(2) The Bean model for the critical state [13] is valid. ¢Yr+~r dr
In particular, the critical current density arc is inde-
pendent of the magnetic field.
(3) The TFM is activated via the field-cooling proce-
dure. The external field Ba is applied along the
z-direction, so that the persistent current flows in the
a-b plane of the crystal lattice. \ • I

(4) The stress is within the elastic limit, and the \ 1


]
elastic properties (such as Young's modulus) of the dO/
material are isotropic in the a-b plane of the crystal Z

lattice ( x - y plane in the coordinate system).


Fig. 5. Balance of forces inside an element of the cylindrical
3.1. Stress after activation TFM.

We had calculated the situation after the activa-


tion field is completely removed. This is the same with
case calculated earlier by Sumption and Collings.
Ba,max
According to the Bean model, the spatial distribution ,* = R - - - (3)
of the trapped field B(r) and the current density ~oJc
J(r) can be described as follows (see Fig. 4). The solution of the stress distribution for the TFM
Case (1). If the activation field B..... _> B *, with proceeds in the same manner as any stress problem
B * = IZoJcR being the maximum trapped field, involving a distributed body force (such as the stress
in a rotating flywheel). We will go through the
B(r)=lZoJc(R-r)=B*(1-R), (1) details for case (1) only.
A volume element r dr d0 d z of the TFM, as
and
shown in Fig. 5, is subjected to three forces having
J ( r ) =J~ for 0 < r < R . components in the radial direction:
Case (2). If, however, B a.... < B*, we have (1) a body force arising from the Lorentz force
interaction of the current and the field, F ( r ) =
B(r)=Ba ..... J(r) =0 for0<r<r*,
JcB(r)r dr d0 dz;
and (2) a circumferential (or hoop) tension which acts on
each end of the element; and
B(r)=lXoJc(R-r ), J(r)=J c
(3) a radial tension which acts on the inner and outer
for r* < r < R , (2) curved surfaces. Summing up the forces on the
element in the direction of the radius gives the
following equation of equilibrium:
o"o dr dO dz + O'rr dO dz
n t
( d ~r )
- or+ ~ dr ( r + d r ) d O d z
Ba. ma~
x -~
-J~B(r)r dr d0 d z = 0.
Neglecting small quantities of higher order, we have
R 0 R R r* 0 r* FI
do'r
a b or0 -- orr -- r dr - J c B ( r ) r = O " (4)
Fig. 4. Magnetic-field distribution along the radius inside a cylin-
drical TFM, according to the Bean model. (a) Ba,max _> B *, (b) The unknown variables tro(r) and o-~(r) in Eq.
Ba.max < B *. (4) can be expressed in terms of a single variable u,
Y. Ren et al. / Physica C 251 (1995) 15-26 19

the radial displacement, and can be derived from a which is the magnetic pressure acting on the inside
consideration of Poisson's ratio v [14]: surface of a cylinder "containing" a uniform mag-
netic field B* [15].
o0 = -r + 11 (5) In case (2), i.e. the case in which the TFM is not
1
activated to its maximum possible field (this case has
not been considered by Sumption and Collings),
°'r= - ~ 1 dr + v- (6) Ba,m x < B *. In this case we have different expres-
sions for B ( r ) and J ( r ) for the inner (r < r * ) and
where E is the Young modulus of elasticity and v is the outer (r > r *) regions of the sample (see Eq.
Poisson's ratio for the material.
(2)). If we express Ba,max in terms of B*, i.e.,
We now substitute Eqs. (5) (6) and (1) into (4), B~,m,x = aB *, where a is a numerical factor and
and obtain a differential equation with only one a < 1, then the Bean model gives r* = ( 1 - a ) R .
unknown, the displacement u: Note that now there is no body force in the inner
dZu 1 du u 1 -- 1/2 region (r < r*). Therefore we have different equa-
d r 2 + r dr
r2 E
tXoJc2( R r). tions for radial displacements for the two regions.
Following the same procedure as for case (1), and
(7)
taking into account proper boundary conditions, we
The solution to this equation is finally have, for the inner region (r < r * ),
1 1-v z
u ( r ) =Cmr + C 2 --lXoJc2(½Rr2-1r3). 7+5v (1+ v)(1-~)
r E °b'= o'ri= lz°J2R2 24 2
(8)

We note that C z = O, otherwise u I r = 0 will be infin- (l + v)(1--a) 2 (1-- v ) ( 1 - - 0~)3


ity. The constant C a can be determined from the
condition 4 6

err I r=R = 0 (9) + (1-- v ) ( 1 - - '


(12)
at the edge of the cylinder, since there is no outside
reinforcement. Having evaluated the constants, we
put the resultant expression for u(r) into Eqs. (5) and for the outer region (r > r * ),
and (6), and get the radial distribution of the stresses
as 7+5v (1 -- v ) ( 1 - - or) 3
7+511 l+_2v/r 1 %o = tXoJ~R 2 24 6
tro = tXoJZR 2
24 3 kR ]
(1- v)(1-a) 4
+
+ --U- ' (lO)

o-r ~ l~oJc2R 2 -( (1-v)(1-a)36 - ( 1 - v ) ( 1 - a ) 8 )4


24 3

+ ---if- - 1+ 2V(R
.
×(R) 2 3 )
These solutions are essentially the same as obtained
by Sumption and Collings, but in a different form.
Since the maximum trapped field B * = l~oJcR, we
1+3v()2]
+ --~ , (13)
see that the magnetic stresses scale with B * 2/21.%,
20 Y. Ren et al. /Physica C 251 (1995) 15-26

Crr° = Ix°J2R2
7+51'
24
(1--1")(1--O:)
6
3
B~llBn I
( 1 - - 1,)(1 -- a ) 4
+
8 B,=B==-B~

(1- v)(1-a (1- v)(1-a


+ 6 - 8 0 r"
a b
Fig. 7. Magnetic-field distribution along the radius inside a cylin-
drical TFM, with decreasing applied field, according to the Bean
(14) model. (a) Ba. . . . _>B* ; (b) B..... < B*.

Here the superscripts i and o denote the inner (r <


r * ) and the outer (r * < r < R) regions, respectively. or near the central region. In addition, if we compare
Obviously, when Ba,max = B * (i.e., a = 1), these o-o and o"r for the same radius r, we find that the
solutions coincide with Eqs. (10) and (11). hoop stress is always larger than or equal to the
Fig. 6 shows the magnetic stresses ob and o"r (in radial stress. Therefore if the sample is uniform and
terms of B* 2/2/z0) as a function of r (in terms of the strength is isotropic, as we have assumed in our
R) for a few typical values of the activation fields, calculation, the cracks tend to occur radially rather
according to Eqs. (10-14). The only material param- than circumferentially, and tend to start at the center.
eter involved in the calculation is Poisson's ratio v, These conclusions agree with the crack patterns
which we take as 0.3, following Goyal et al. [16] We shown in Fig. 3, and with the cracks in the other
see that as a general trend, both o"0 and o-r increase three discs. However, cracking occurred during, not
with the decrease of r. In other words, the stress after activation, and we must now consider that case.
increases as the center is approached. Therefore in
case of sample breakage, it most likely will start in 3.2. Stress during activation

0.8 To evaluate the experimental results, we now


consider the magnetic stresses during the activation
of the TMF's. In order to achieve the maximum
trapped field B *, the activation field Ba, max m u s t be
higher than, or, if we ignore the flux creep, at the
0.4-'= ' \ ~ very least equal to B *. In practice, activation in-
volves an unknown B* and, as a consequence,
13 -B= ~ = 0. 6B~ Ba,max is chosen to exceed the range of possible
values of B * by a sufficient amount. The magnetic
pressure in this case is due to the interaction of the
current with the sum of the self-field and the applied
field. Therefore during the activation, the magnetic
stresses may be, and usually are, higher than those
-0.2' , I , calculated previously with the applied field being
0 0.2 0.4 0.6 0.8 1 completely removed. As illustrated in Fig. 7(a), ini-
r/R tially the persistent current flows only within the
Fig. 6. Radial distribution of magnetic stresses inside a cylindrical outer region of the cylinder. At this stage, the mag-
TFM, according to Eqs. (10-14): solid lines, ob, the hoop stress; netic flux density B in the inner region does not
and dashed lines, o'r, the radial stress. change. With the decrease of the applied field, the
Y. Ren et al. / Physica C 251 (1995) 15-26 21

persistent current penetrates further into the sample, 1.2 L


and finally reaches the center when B a = B a.... - - B *
After this, the magnetic field inside the sample de-
creases uniformly with the applied field, until B a -- 0.
// Y \
If the activation field Ba, max < B *, the process is
08
similar to the first stage of the previous case, but we
end up with a trapped field B t = Ba,max at the most
(see Fig. 7(b)). •~4~ 0.6
To calculate the magnetic stresses during the
whole process, we have to use different formulas to
express the field and the current distributions for
different stages and for different regions. The calcu-
lations are unavoidably lengthy and tedious, but the 0.2-
physical principles and the mathematical methods
involved are the same as used previously. In particu- o
lar, two conclusions of the previous calculations still 1.25 0.75 0.5 0.25
hold, i.e., Ba/B*
(1) both the radial and the hoop stresses o"r and tr0 Fig. 8. The maximum magnetic stress inside a TFM when the
increase when the center is approached, and applied field drops from Ba,max tO zero, for different activation
(2) for the same radius r, the hoop stress tr0 is fields B~.ma~ = 1.25B *, 1.10B *, 1.00B *, and 0.75B * (from top
higher than, or equal to the radial stress o"r. Thus the to the bottom), respectively.
discs are still most likely to crack near the center,
with the cracks propagating radially.
Our interest is in the mechanical limits of the and for a - l > x > 0
discs during activation. Thus we only need to know
B
2 (' Z2 +[ v ) - ( l + x ) - --3 + v ] . (16)
how the maximum stress, O ' m a x =O'olr= 0 changes °'max = 2/.% 3 4
with the decrease of the applied field. If we express
B ..... and B a in terms of B * , i.e., Ba,m,x = a B * , In case of ot < 1 (Fig. 7(b)), Eq. (15) represents the
and B, = xB *, when we sweep down the applied maximum stress for the whole process of ramping
field from B .... x to 0, x will change from ot to 0. down the applied field from B a. . . . to 0 ( a > x > 0).
Here we give the expressions for O'max directly. The In Fig. 8 we show how the maximum stresses
detailed derivations for these equations are given in change with decreasing applied field during activa-
the Appendix. In case of a > 1 (Fig. 7(a)), we have, tion for a few typical values of a. As we can see
for a > x > a - 1 from the plot, during the activation, for a certain
range of the applied field, the stresses are indeed
B .2 [ 1- v higher than the stresses after the applied field is
trmax= Z / t 0 --'-4--(1--Or+X) 4 removed. Thus, cracking will occur during activa-
tion, and to predict cracking the expanded theory
1-- V must be used. There are significant differences, how-
3 (l+x)(1-a+x)3
ever, between the cases when a_< 1 and a > 1.
When a _< 1, the maximum stress during activation
l+v is slightly higher, but close to the value with B a --- 0.
+--(1- ot + x ) 2
2 For example, when a = 0.75, O'max/o'(Ba = 0) =
1.03 and when a = 1, the ratio is also 1.03. On the
-(l+v)(l+x)(1-a+x)
other hand, when a is even slightly higher than 1,
say a = 1.10, the ratio becomes 1.22; and for a =
+ 2(2+V)(l+x)3 - - - '~3 + v ] (15) 1.25, the ratio is as high as 1.54. Although in
principle the analytical expression of the ratio
22 Y. Ren et al. /Physica C 251 (1995) 15-26

Ormax/tr(B a =0) as a function of a can be found, we 60


instead calculate numerically for each specific value
of a. Of course we should remember that if a < 1, 50
we cannot reach the maximum trapped field B *. t)_

However, it is clear that in order to reach the goal of 40


t/)
a given trapped field, Bt, we should use the lowest
possible activation field, in order to avoid excessive 03 30 ¸
magnetic stresses and consequent cracking. =E
t~
3r
4. Discussions and conclusions

Or , ,
We now compare the experimental results with 14 12 10 8 6 4 2 0
the theoretical calculations. The strength of the oxide
superconductors is much lower than that of low-tem-
Applied Field (Tesla)
perature superconductors. For example, the strength Fig. 9. Calculated maximum magnetic stress for the mini-magnet
is between 800 to 2000 MPa for Nb3Sn compound in as a function of decreasing applied field from 14 T to zero, using
the estimated trapped field Bt = B * = 10.6 T. The radius of the
bronze-processed wire [17]. The experimental values mini-magnet is 1 cm.
of fracture strengths for YBCO depend highly on the
microstructure, ranging from ~ 10 MPa for porous
sintered material [18] up to ~ 100 MPa for silver-
sheathed tapes [19]. The bulk melt-textured materials
as the applied field is reduced, is plotted in Fig. 9.
fall in between, and have fracture strengths which
The stress rises as B~ is reduced. When the applied
vary from 40 MPa without Ag enhancement up to 70
field has been reduced to 8.14 T, where the break
MPa with Ag enhancement [20]. The fracture strength
occurred, the magnetic stress is about 43 MPa,
expected for the discs used in this experiment is
slightly above the fracture strength of 40 MPa found
about 40 MPa. In Fig. 6, we find, at the center of the
by others [20].
TFM, the maximum stress O'max = 0.71(B* 2/2/Zo).
MT-YBCO is not uniform and the strength varies
Thus we conclude that the MT-YBCO can allow a
from sample to sample. Also, we have made several
maximum trapped field of ~ 12 T. However, this
oversimplified assumptions in our calculations. For
value is what the TFM can hold, if we can avoid
example, we have used the Bean model instead of
using excessive activation field.
the B e a n - K i m model for the current, and we have
Now let us estimate the magnetic stress which
neglected creep. Considering this, we judge that the
caused the observed damage during activation. We
experiment and the theory agree very well. We also
have found [7] that to a good approximation, for the
conclude that had we used an activation field of less
temperature range of 20 K to 65 K, the temperature
than 12 T, we would have reached a trapped field
dependence of the trapped field for MT-YBCO can
above 10 T, without damaging the sample.
be fitted to a straight line
The microstructure of the MT-YBCO is not uni-
~-~t = a - bT. (17) form. It contains a significant number of imperfec-
tions, such as grain boundaries, crystalline defects,
Prior to the 14 T, 49 K run, which broke the discs, second-phase inclusions, cavities, etc. These imper-
we have found the same four-disc mini-magnet fections may cause stress concentration in local re-
t r a p p e d B t = 7.03 T at 59 K, and 8.33 T at 54 K. gions, and may initiate small cracks. It is these weak
Using these results we can extrapolate to B t at 49 K. points that limit the strength of the material. Accord-
We find, at 49 K, B t = B * = 10.6 T. Using Eqs. (15) ing to fracture mechanics [21], unstable crack propa-
and (16), with tx = B ..... /B * = 1 4 / 1 0 . 6 = 1.32, the gation will occur when the stress intensity factor
calculated maximum stress inside the mini-magnet, reaches a critical value K c, usually called the frac-
Y. Ren et al. /Physica C 251 (1995) 15-26 23

ture toughness of the material. For a small crack in a of bulk superconductor. We have found that both the
large plane with remote tensile stress, radial and the hoop-stresses are proportional to B * 2,
the square of the maximum trapped field. During
Kc = Q ~ (18)
field-cooled activation, the current increases with the
where a is the size of the crack, tr the tensile decrease of the applied field. For activation field
strength, and Q is a configuration factor determined Ba,max > B *, the magnetic stress during activation
by the geometry. The Kc value of YBCO has been will exceed the value of the stress after B a = 0 (i.e.,
reported to be about 1 MPal/m [18,22]. Taking after completion of the activation). Since Ba,max > B *
Q = 1 for simplicity, and t~= 40 MPa, the initial is the usual condition in activation, cracking under
crack size a is estimated to be 200 lxm. However, if magnetic pressure is more likely during activation.
larger cracks exist inside the sample, the strength The observations agree well with the theoretical
will be further reduced and the magnet may break at predictions.
an even lower trapped field. In our calculations of the cracking of the TFM,
It is interesting to note that if the activation field giant flux jumps have not been involved. We inter-
B ..... < B * (non-saturated case), the corresponding pret the damage as due to magnetic pressure without
magnetic stress will be much smaller than the satu- prior giant flux jumps.
rated case Ba.max >_B *. As seen in Fig. 6, if the The mechanical strength sets a limit for the mag-
activation field is 60% of the maximum trapped netic field which a MT-YBCO superconductor can
field, the maximum stress is only 40% of its value in trap. We conclude that present materials can trap
the case of saturation. In practice, there are other over 10 T. However, it is clear that it is necessary to
advantages if we do not push the trapped field to its improve the mechanical strength at the same time
limit. For example, we will have a more uniform when we make efforts to enhance the critical current
field distribution (see Fig. 4b), and a lower creep rate density and the grain size, in order to increase the
of the trapped field [3]. trapped field.
The cracking of all four discs at one time was not
expected, but after the fact, it is not surprising. The
four discs were very similar, and subject to essen- Acknowledgements
tially the same current and fields. Hence, all four
were near the cracking point. When one cracked,
We want to thank the staff of the Indiana Univer-
there was a sudden reduction in its current and field.
sity Cyclotron Facility for providing support for the
As a result there was a large EMF induced in the
YBCO irradiation. We also wish to thank NHMFL at
other discs. By Lenz' law the direction of any addi-
Florida State University, Tallahassee, for making
tional induced field was such that it increased the
available a 27 T magnet and variable temperature
existing field, and consequently increased the mag-
cryostat, for support during our experiment, and for
netic pressure.
its many courtesies. Our work was supported by the
In conclusion, we have observed the cracking
State of Texas, via its Advanced Technology Pro-
under magnetic pressure of a mini-magnet composed
gram and TCSUH, and by NASA-JSC, the NSF
of four 0 2 0 mm × 8 mm MT-YBCO discs. It had
Division of Materials Science, and ARO.
been field cooled to 49 K in a field of 14 T. All four
discs cracked. To the limit permitted by the time
resolution in our measurements (observations every
30 s), all four broke simultaneously. The four discs Appendix
all show faults which carry low, or zero current and
which are consistent with a crack starting at or near Derivation o f formulas for stress during activation
the center of the disc, and propagating radially.
We have calculated the stresses induced by the As pointed out in Section 3.2, the basic physical
interaction between the magnetic field and the persis- principles and mathematical methods for calculating
tent current inside a cylindrical trapped-field magnet the stress during activation are exactly the same as
24 1I..Ren et al. /Physica C 251 (1995) 15-26

those used in Section 3.1 for stress after activation. where r* = (1 - a + x)R. For stage 2,
The only difference here is that the magnetic field
and current distributions inside the superconductor B(r)= (;-) l+x- B*, J(r)=Jc,
are constantly changing with the decreasing applied
field B,, so different equations have to be used for for 0 < r < R. (A.5)
different stages in the activation. So for stage 1 we need to distinguish the inner
In Section 3.1, Eq. (4) describes the force balance (r < r* ) and outer (r > r * ) regions.
for the field profile shown in Fig. 4a, where Jc flows Substituting Eqs. (A.3) and (A.4) into (A.2) leads
uniformly throughout the superconductor. To deal to
with the different field and current distributions oc- d2u i 1 du i ui
curring during activation, we simply replace Jc in + = 0, for r < r* (A.6)
Eq. (4) with the general current distribution, J(r), dr 2 r dr r 2

and
do"r
°'o- ~r-- r dr - J ( r)B( r)r = 0. (A.1) d2u ° 1 du ° u°
--+
dr 2 r dr r2
Similar to Section 3.1, we substitute o"0 and o-r with
Eqs. (5) and (6) to get the differential equation for
- - - - l + x -
the displacement u, } Jz0R
d2u 1 du u 1 - v2 for r* < r < R. (A.7)
d r 2 + r dr r2 E J(r)B(r). (A.2) Here the superscripts i and o denote the inner (r <
r * ) and outer (r > r * ) regions, as in Section 3.1.
As described in Section 3.2, there are two cases in
The solutions to Eqs. (A.6) and (A.7) are
which B a , m a x > B * and B ..... < B * respectively.
1
For the first case there are two distinct stages sepa-
u i = C 1r + C 2 - , (A.8)
rated by B a = Ba.r~ax - B *, which corresponds to the r
middle field profile in Fig. 7(a). The supercurrent
only flows in the outer region of the superconductor
in stage 1 when B a > Ba,max - B *, and fully pene-
trates into the superconductor when B a < B a , m a x - B *
and

u ° = C3r + C 4
1
r
1 - v z B__.2_
E I~oR 3
(
_~ l + X r e _
-~"
r3)
in stage 2. Fig. 7(b) indicates that the field profile for (A.9)
Ba.max < B * is similar to stage 1 of B .... x > B * ,
thus we only need to consider the c a s e Ba,max > B * To keep U i finite at r - 0, we must have C 2 = 0.
in the following, and the result for stage 1 can be Both the displacement u and the stresses %(r) and
applied to B . . . . . ( B * o'r(r) should be continuous at r = r *, and from Eqs.
Following the notations in Section 3.2, we ex- (5) and (6) this requires d u / d r to be continuous at
press the magnetic fields in terms of B *( = / x 0 Jc R), r = r *. Also o"r = 0 at r = R, for the same reason as
i.e., B a , m a x = aB*, and n a = xn *. Then a > x > ot explained in Section 3.1. So we have the following
- 1 corresponds to stage 1 and a - 1 > x > 0 corre- boundary conditions from which C 1, C 3 and C a are
sponds to stage 2. From Fig. 7(a), we have, for stage determined:
1,
"Ui]r=r • = U ° l r = r * ,
n(r) ~'n ...... J(r) =0, for0<r<r*,
(A.3) d u i r=r* duO r=r
= dr .' (A.10)

B(r)=
(I+X--R
r) B*, J(r)=Jc,
- du ° uo ]
- d- +r v - - r 1 r=R = 0 .
for r* < r < R, (A.4) Substituting Eqs. (A.8) and (A.9) into Eq. (A.10),
Y. Ren et al. / Physica C 251 (1995) 1 5 - 2 6 25

and solving the resulting linear equations, we obtain


the coefficients C l, C 3, and Ca: -- n*2
o( - 1 I 2-(1-/- 4*/}) - °e + x)4

1 1-v2B*2[(1-v)
CI= l+v _ ( 1 - - v)(1 + x ) ( 1 _ a + x ) 3
E 21. 0 - - (1 - a q'-x)4
3
( 1 - v)(1 + x ) l+v
3 (1 -- Ot +X)3 + -T-(1
l+v - ( 1 + v)(1 + x ) ( 1 - a+x)
+--(1--~+X) 2
2
2(2 + v)(1 + x ) 3+v] (A.12)
- ( 1 + v)(1 + x ) ( 1 - a + x )
+ 3 4 "
+ 2(2+v)(l+x)3 3+v]4 '
This is Eq. (15) given in Section 3.2.
For stage 2 (0 < x < a - 1), we substitute Eq.
1 1-,, (A.5) into (A.2) and reach the following equation for
= ( 1 - - C~+X) 4 u:
C3 1 + v ~-- 2/x o
(1- v)(1 +x) d2u ° 1 du ° u°
-- ( 1 -- O / + X ) 3
--+
3 dr 2 r dr r2
+ 2(2+v)(l+x)3 3+v]4 '
- --- l+x-
bY /xoR

1-v2___B*Z[l(l_a+x)4 for0<r<R (A.13)


C4 = E 2/*0 The solution to this equation is

l + x ( 1 3 -- a + X ) 3 ] " 1 l-v2B*2(l+x r3 )
U = Csr + C 6 r2 - --
r -E -~,~-R 3 8R"
(A.11) (A.14)
As pointed out in Section 3.2, stresses increase when Again the boundary condition at r = 0 requires C 6 =
the center is approached no matter what the field 0, and from the condition
profile and current distributions are. Here in stage 1
the field profile is similar to that of Fig. 4(b), which
E [du u] =0~
is the unsaturated case (Ba,max < B *) discussed in <lr=R 1 ~-1]2 -~r q- 7 r=R

Section 3.1. The only difference is that r * now is a C 5 is determined to be


function of B a and is moving toward the center as B~
1 1 - - V 2 B .2
is being ramped down. Therefore the stress versus r
Cs=--
curves here should be similar to the two lower 1+ v E go
curves in Fig. 6, i.e., both o"0 and o"r reach and stay
at the same maximum for r < r *. Then to calculate ×[(2+v)(l+x) 8 " (A.15)
the maximum stress O'ma~, we only need to use Eq.
(5) or (6) and substitute u with u i here: The maximum stress here is at r = 0. Therefore,
E du u
~rmax
1- v2 +Vdr 1 _ 12 2 u--+--r dr O'max 1 - v2 + v r=0
E
1 -- U2 (1 + v ) C 1 l_v2 +v r=0
26 Y. Ren et aL /Physica C 251 (1995) 15-26

E [7] I.-G. Chen, J. Liu, Y. Ren, R. Weinstein, G. Kozlowski and


- 1-v 2(1+v)C5 C.E. Oberly, Appl. Phys. Lett. 62 (1993) 3366.
[8] Y. Ren, J. Liu, R. Weinstein, I.-G. Chen, D. Parks, J. Xu, V.
Obot and C. Foster, J. Appl. Phys. 74 (1993) 718.
B*2 [2(2+ v)(1 +x) 3+v I (A.16) [9] R. Weinstein, Y. Ren, J. Liu, I.-G. Chen, R. Sawh, V. Obot
2/z 0 3 4 and C. Foster, Advances in Superconductivity IV, eds. T.
Fujita and Y. Shiohara (Springer, Tokyo, 1994) p.855.
We have reached Eq. (16). [10] R. Weinstein, J. Liu, Y. Ren, I.-G. Chen, V. Obot, R.P.
The s e c o n d case w h e r e B .... x <B*(o~ ,(1) is Sawh, C. Foster and A. Crapo, Proc. Int. Workshop on
mathematically equivalent to stage 1 of the first case, Superconductivity, Kyoto, Japan, June 1994 to be published.
so Eq. (15) is directly applicable for the entire [11] R. Weinstein, Y. Ren, J. Liu, R. Sawh, D. Parks, C. Foster,
V. Obot, G.D. Arndt and A. Crapo, Proc. Fourth World
a c t i v a t i o n p r o c e s s (0 < B a < B . . . . . or 0 < x ,~ og).
Congress on Superconductivity, Orlando, Florida, June 1994,
to be published.
[12] M.D. Sumption and E.W. Collings, ibid., Ref. [11].
References [13] C.P. Bean, Phys. Rev. Lett. 8 (1962) 250.
[14] S.S. Timoshinko, Strength of Materials, 3rd ed. Part I, §15
[1] D.B. Montgomery, Solenoid Magnet Design, Ch. 5 (Wiley- (Van Nostrand, New York, 1956).
lnterscience, New York, 1969). [15] F.C. Moon, Magneto-Solid Mechanics (Wiley, New York,
[2] M.N. Wilson, Superconducting Magnets, Ch. 4 (Oxford Uni- 1984) p.5.
versity, Oxford, 1983). [16] A. Goyal, W.C. Oliver, P.D. Funkenbusch, D.M. Kroeger
[3] R. Weinstein, I.-G. Chen, J. Liu, D. Parks, V. Selvaman- and S.J. Bums, Physica C 183 (1991) 221.
ickam and K. Salama, Appl. Phys. Lett. 56 (1990) 1475. [17] S. Ochiai, T. Uehara and K. Osamura, J. Mater. Sci. 21
[4] R. Weinstein, I.-G. Chen, J. Liu, R. Narayanan, Y. Ren, J. (1986) 1020.
Xu, V. Obot and J. Wu, in: Proc. Third World Congress on [18] T. Shoji and Y. Tazawa, Mater. Sci. Engin. A 143 (1991)
Superconductivity, Munich, Germany, September 1992, eds. 241.
K. Krishen and C.G. Burnham (Pergamon, New York, 1992) [19] S. Ochiai, K. Hayashi and K. Osamura, Cryogenics 32
p.1145. (1992) 799.
[5] I.-G. Chen, J. Liu, R. Weinstein and K. Lau, J. Appl. Phys. [20] D. Lee and K. Salama, Jpn. J. Appl. Phys. 29 (1990) L2017.
72 (1992) 1013. [21] See, for example, A.P. Paker, The Mechanics of Fracture and
[6] J. Liu, I.-G. Chen, R. Weinstein and J. Xu, J. Appl. Phys. 73 Fatigue, An Introduction (Spon, London, 1981).
(1993) 6350. [22] F. Yeh and K.W. White, J. Appl. Phys. 70 (1991) 4989.

You might also like