Conference Discussln 31

for a slab and I can state exactly how the order parameter energy; that is, it can be shown rigorously to have a lower
varies in the high-held region. What I find, which was free energy than any of the oscillatory solutions in which
rather surprising when I first came upon it, is that the be- the order parameter oscillates around zero in going across
havior does not look like the mixed state or the state in the slab. I might say that even though no flux "vortex" is
which the order parameter oscillates as you go across the present in this solution, the magnetic moment curve looks
—
slab instead, the state has a superconducting core at the just like the type II curve with a first-order transition at a
center of the slab surrounded by nearly normal material. lower critical field and an abrupt drop in the magnetic mo-
For example, if I let kappa equal 2, I find the central core ment. It fa11s ofF to zero at a field approximately equal to
is about 7 or 8 penetration depths wide, nearly completely the usual high-field limit when 8 is equal to kappa. I' ll
superconducting, and then it tails o6 to values less than show the free energy curves and other details of this in my
10 4 which can extend over macroscopic distances. This discussion on Thursday — it will look quite similar to Pro-
solution can be shown to be the one with the lowest free fessor Gorter's results.

Magnetization of High-Field Superconductors

CHARLES P. BEAN
General Electric Re8earch Laboratory, Schenectady, Pew Pork

who, following a suggestion by Landau, developed a

INTRODUCTION
quasimlcroscopic theory of magnetization of alloys in
Magnetization measurements have played a crucial which he showed that the flux entered the specimen
role in the development of the understanding of in quantized superconducting electron current vor-
superconductivity. ' On the one hand, the discovery tices (variously called Quxons, Aux lines, flux threads).
of the Meissner effect (complete Aux expulsion in He showed this penetration to start at a low field
macroscopic specimens) gave the first clue to the (H&p or H„) and be complete at a higher field (H» or
present concept of the soft superconductor as a large H.s). Between these regions of field, the specimen is
scale quantum state. The relative lack of irreversi- presumed to possess a structure of fluxons a mixed —
bility in the transition to the normal state caused by state. While, to date, there has been no direct ex-
an applied field gave assurance that thermodynam- perimental confirmation of this structure, indirect
ical reasoning could be applied with confidence in evidence' (mainly magnetization curves) points to
consideration of the various physical changes at- the essential correctness of this picture. An important
tendant to the transition. On the other hand, meas- feature of this theory is that it calculates states of
urements on alloys' ' indicated a more complicated thermodynamic equilibrium. This predicted lack of
state of affairs. First of all, the Aux penetrated the
specimens over a broad range of field extending the
process to much higher fields than the critical fields
— firmed in recent years. "
hysteresis in homogeneous alloys has been well con-

This paper deals with the theory and experiment

of pure superconductors —
and secondly, there was on hysteretic high-field superconductors. The ideas
observed a large hysteresis in the magnetic behavior. were prompted by an older model of high-field super-
After a brief Hurry of experimental interest in the conductors —the Mendelssohn sponge modeis-
mid 1930's, the field lay dormant. It may be that the wherein one assumes a multiply connected internal
hysteresis noted above discouraged careful experi- structure of high critical field material surrounding a
ment and theory. The present understanding started matrix of material lower (or zero) critical field. The
with a remarkable paper in 1957 by A. A. Abrikosov' treatment, however, is phenomenological with only
occasional contact with the microscopic or quasi-
&
An excellent account of the development of this field is microscopic features of superconductivity. Surpris-
given by E. Mendelsson, Cryogenics 3, 129 (1968)
2 T. C. Keeley, K. Mendelssohn, and J.
R. Moore, Nature ingly enough this simple theory appears to account
134, 778 (1984). for the magnetic properties of inhomogeneous mixed-
3 W. J. de Haas and J. M. Casimir-Jonker, Nature 135, 30
(1985). state superconductors as well as the sponge model
4F. O. A. Tarr and J. O. Wilhelm, Can. J. Res. 12, 265
(1985). &
J. D. Livingston, Phys. Rev. 129, 1948(1968).
For example,
J. N. Rjabinin and L. V. Shubnikov, Physik. Z. Sovjietunion 8 A. Calverley and A. C. Rose-Innes, Proc. Roy. Soc.
'7,122 (1985). (London) A255, 267 (1960).
6A. A. Abrikosov, Zh. Kksperim. i Teor. Fiz. 32, 1442 s K. Mendelssohn,
—
(1957) [English transl. : Soviet Phys. JETP 5, 1174 (1957)]. (1985).
Proc, Roy. Soc. (London) A152, 84
 32 REvIEws oF MoDERN PHYsIcs ' JANUARY 1964

for which it was developed. In fact, I think it is fair into an imhomogeneous mixed-state superconduc-
to say that the hysteresis of high-field superconduc- tor 13 15

tors is at least as well understood as that of ferro- If we assume, as a starting point, that this current
magnetic materials. density is independent of field, the process of mag-
In detail, the program of this paper is to give an netization of a slab of thickness D in a field parallel
exposition of this theory of the static magnetization to its surface is shown in Fig. 1. The field within the
of hard superconductors, compare it to experiment, specimen decreases linearly with distance as a con-
and lastly to discuss the response of hard supercon- sequence of Ampere's law, curl H = 4s-J/10. (In this
ductors to alternating fields with a superimposed equation and all that follow I employ practical units,
steady field. i.e., oersteds, amperes/crn', and volts. ) The config-
uration of fields are shown in Figs. 1(a) and 1(b). In
THEORY OF THE MAGNETIZATION OF
the initial stages of magnetization, the current Qows
HARD SUPERCONDUCTORS
in superficial layers whose thickness 6 is just enough
The basic premise of this theory"" is that there to reduce the internal local field to zero, i.e. ,
exists a limiting macroscopic superconducting cur- 10H/4m J.
. This field-dependent penetration
rent density J.
(H) that a hard superconductor can depth is the central result of the assumptions made
carry; and further, that any electromotive force, above and leads directly to size-dependent mag-
however small, will induce this full current to flow netization curves. At fields of H )~ s. J)/5 — = H*, J

H=2H"
H =(2H

H = H /2 (-4'lt'M, B)

H=0
H=H

H = H"/2
0
0 2H"

(a)
FIG. l. A plot of local fields and current density, as well as magnetization curves, for fields 0, II*/2, H*, and 2H* applied
parallel to the surface of a slab of thickness D. The critical current density J, is assumed independent of field.

locally. On this picture only three states of current currents fiow through the entire volume of the speci-
How are possible with a given axis of magnetic field, men. To calculate the magnetization curves of the
zero current for those regions that have never felt specimen, we have, by definition, that
the magnetic field and full current Qow perpendic-
ular to the field axis, the sense depending on the B= Hdv
sense of the electromotive force that accompanied
the last local change of field. The critical current 4xM = B —H. (1)
may be, on the one hand, an intrinsic property" of In other words, 8 is merely the volume average of
the walls of the sponge in the Mendelssohn model or, the local field while 4+3f is the average field created
on the other hand, a consequence (by Ampere's law) by the currents. By inspection of Fig. 1, one obtains
of the gradient of Aux lines that exists as Aux is driven
~3 Y. B.
Eim, C. F. Hempstead, and A. R. Strnad, Phys.
Rev. Letters 9, 806 (1962); P. W. Anderson, Phys. Rev. Letters
9, 809 (1962).
ro C. P. Bean, Phys. Rev. . Letters 8, 250 (1962). 14 J. Friedel, P. O. Deoennes, and J. Matricon,
Appl. Phys.
C. P. Bean and M. V. Doyle, J. Appl. Phys. 33, 3334 Letters 2, 119 (1968).
(1962). I J. Silcox and R. W. Rollins, Appl. Phys. Letters 2, 231
rs J. Bardeen, Rev. Mod. Phys.
34, 667 (1962). (1968).
 CHARLES P. BEAN 3IIagnetization of Kigh F-ietd Su7ierconductore

8 = H'/2H*, distribution is that shown in Fig. 2(b) wherein each

H ~& m. J,D/5 —
= H* surface is coated with two equal and oppositely di-
—4~& = H —H'/2H*, rected current sheaths. The trapped flux or remanent
flux density 8, is seen from Fig. 2(a) and Eq. (1) to
(slab) (2) be exactly half the flux penetration at Ho, if Ho ~& H*,
8 = H —H*/2, therefore,
II & ~J.D/5 —
= H* 8„=Ho/4H*, Ho & H*
—4n3II = H*/2.
(slab)
These results are sketched in Fig. 1(c). For a cylinder 8, =H*/4, Ho &H*
of radius 8, in an axial field the integrations indi-
cated in Eq. (1) give 8, = Ho/2H* —Ho/4H*', Ho & H* (4:)

8 = H'/H* —H'/3H*', (cylinder)

8, = H*/4, & H* .
4~J.R/10 —
Ho
H ~& = H*
In fact, the entire hysteresis loop may be calculated
4zr3f =—H —H'/H* + H'/3H*', using the principles outlined above. The results are
(cylinder) (3) 8 = HH, /2H* & (H' —H,')/4H*, H & Ho (slab
8 = H —H*/3, 8 = HHo/H* a (H' —Ho')/2Ho
—
= Ho
H & 4irJ.B/10 a (Ho + HHo —H'Ho + H'/3)/4H*',
& Ho . (5)

It-
Ho (cylinder)
The plus signs apply for the course from —Ho to Ho
and the minus sign for the course from H, to Ho.
is interesting to note that this loop is the precise
diamagnetic equivalent of the Rayleigh hysteresis
loop used in ferromagnetism. Lord Rayleigh" intro-
H =
Hp duced this form as the simplest analytic expression
to Qt empirically the magnetization loop of weakly
magnetized ferromagnets.
Since we have the hysteresis loop, the loss of energy
per unit volume per cycle, W., may be immediately
calculated from its area, i.e.,
"c H=0

W. = (1/4~) HdB, (6)

which gives

W. = Ho/6nH*, Ho & H* (slab)

H=H0
W. = Ho/3zrH* —5Ho/16irHo', Ho & H*
(cylinder) . (7)
This gives the Rayleigh cubic dependence of loss on
(b)
maximum field. In the case of Ho &) Ho, the loss ap-
Fro. 2. A plot of local fields and current density in a slab proaches a linear dependence. It is illuminating to
after a field II0 has been applied and removed.
consider the loss in small fields, 'H0 ((
H*, where the
loss is a surface loss W. and both Eqs. (7) become,
Let us now consider what happens if a field Ho is
after multiplying by the volume and dividing by the
applied to a virgin slab and then removed. This state
surface area,
of affairs is indicated in Fig. 2. As the field is re-
moved, the surface feels an emf oppositely directed
— W, = 5Ho/12m' J. ergs/cm'/cycle .
to the one felt as the field was increasing hence the
surface currents reverse. At zero field, the current M Lord Rayleigh, Phil. Nag. 5, 28 (1887).
34 REVIEWS OF MODERN PHYSICS ' JANUARY 1964

A more microscopic picture of the losses may be had of Aux enters that is proportional to H and Ax. For
by considering that within the specimen local elec- a slab geometry, d, B = 2Hhx/D.
tric fields exist during the change of magnetization. By the definition of Ax above, AB
The local Joule heating is the product of these fields 20HDH/4m J.
(H)D. If we define a new penetration
and the local current density. field H* (H) = wJ, (H)D/5 and pass to the differ-
In the paragraphs above I have assumed the cur- ential limit
rent density to be independent of field. The approach
may be generalized by consideration of Fig. 8.
dB/dH = H/H*(H), H « H*(H) (slab) .
In the limit that H* is independent of field, we ob-
tain the parabolic relationship of Eq. (2). If, for in-
stance, J,(H) were assumed to be inversely propor-
tional to field" then the Aux density would increase
with the cube of the applied field. Conversely, H*(H)
"c could be obtained experimentally by measur'ing the
slope of the magnetization curve. In practice this
technique is somewhat dificult to employ since it
requires very accurate measurement of the B,H curve
H

(o)
to obtain accurate derivatives.
Before Leaving this section several comments are
in order concerning the validity of the assumptions
that underlie this development. First, the assump-
tion of an infinitely sharp front on the curve of cur-
rent density against distance is never completely
valid. There is always an exponentially decreasing
tail to the curve. In the case of the inhomogeneous
mixed state this is a London penetration depth in
(bj which the field is reduced from H&r to 0. In the case
of the filamentary superconductor it is a quasi-Lon-
don depth but one which is longer than that of the
"
bulk material. The assumption employed above will
be asymptotically valid as the field-dependent pene-
tration depth 6 greatly exceeds the London penetra-
tion depth. A second point concerns the tacit as-
sumption that the critical current density is inde-
pendent of time. Owing to the possibility of thermal
activation of Qux lines through filaments or across
barriers in the inhomogeneous type II supercon-
ductor" this is not a universally valid assumption.
FIG. 3. A plot of the eQ'ects of a current density J, However, many processes in solid-state physics have
depends on field H. Figure 3(a) shows a schematic curve of
this dependence. Figure 8(b) shows the currents at a plane
that
—
this character for instance the movement of disloca-
surface for an applied field H (solid curve) and an applied tions or the reorientation of ferromagnetic domains.
field H+ hH (dashed curve). Figure 8(c) shows the internal But depending on the temperature and the nature of
fields under the same conditions.
the barriers one can often describe the properties over
a wide frequency range with the assumption of a
Figure 3(a) shows an arbitrary dependence of cur- time-independent yield stress or coercive force.
rent density on field. The consequent, profiles of cur- Whether this be true or not for high-field supercon-
rent density and field are shown in Figs. 8(b) and 3(c). ductors is a subject for experimental investigation.
Consider that a field H has been applied, giving the
field and current distributions shown as solid curves. EXPERIMENTAL MAGNETIZATION CURVES
If an increment of field AH is applied, the curves
are displaced inward by an amount The static magnetization measurements reported
hx
10AH/4w J.(H), where J,(H) is t'he current density
here are made by Ripping the specimen, generaljy
appropriate to the field at the surface. An amount » C. P. Bean (to be published).
 CHARLES P. BEAN Magnetization of High Fi-eld Suyerconductors

0.125 in. in diameter and an inch or so in length, in panded cylinder was then crushed, repressed, and
and ouit of a 2000-turn coil mounted axially in an given a final firing for 1 h at 1500'C in a vacuum of
electromagnet. The integrated voltage signal, pro- 10 ' mm Hg. The experimental points for both the
portional to the magnetization, is displayed on a Aux penetration and Aux retention are seen to be Gt
General Electric Huxmeter. The calibration is made quite closely by Eqs. (3) and (4) with the assumption
by the assumption of perfect diamagnetism at very of a current density of 660 000 A/cm'. A more com-
low Gelds. plete hysteresis loop is given elsewhere. "
The sub-
stantial agreement with theory implies that the
theory may be valid for inhomogeneous type II
4000-
ample. ""
superconductors of which V30a is apparently an ex-
The main disagreement with theory is seen
in the high field point whereat a small amount of Aux
entered the specimen discontinuously. These Aux
3000
jumps" "
which may be almost complete, in con-
trast to the small one shown here, are the principal
limitation to the employment of this technique over
~ 2000— a wide field range with thick, high current density
tX3
specimens.
An experiment on a synthetic filamentary" ma-
1000— terial is shown in Fig. 5. The sample was prepared
by H, . J. Charles" and W. G. Schmidt who pressed
lead into porous Vycor glass at 355'C and 44 000 psi.
0'--'
The Ailing was incomplete in contrast to that of a
2000 4000 specimen reported later in this paper. The figure
H (08) shows the initial Inagnetization and two hysteresis
loops. The solid lines are derived from Kqs. (3) and
1500— (5) with the single assumption of a current density
of 2.7 && 10' A/cm'. The agreement is seen to be sub-
stantial. The lack of detailed agreement may be
mainly a consequence of a surface layer, discussed
1000— later in this paper, as well as the inaccuracy, in de-
tail, of the assumption that the current density is
independent of field.
Eim, Hempstead, and Strnad" have reported an
500— important series of experiments on the static mag-
netization behavior of Nb —Zr, Nb, Sn, and V, Ga.
Usually they employed a thin wall, cylindrical shell
geometry to minimize the difficulty of analysis ow-
JE ing to the dependence of critical current on field. In
0
0 2000 4000 6000 8000 most cases their measurements were made under con-
H (Oe)
ditions of current How through the entire cross sec-
(b) —
tion of the sample a condition they term the "crit-
Fxe. 4. Flux penetration and flux retention in a sample of "
ical state. They found that the magnetization over
sintered VsGa. Figure 4(a) shows the fiux penetration at
4.2'K in a sample 0.38 cm in diameter. The solid curve assumes
a current density of 6.6 X 104 A/cms. Figure 4(b) shows the r9 C. P. Bean and R. W. Schmitt, Science 140, 26 (1963).
remanent flux density after exposure to various fields II. The 20 P. W. Swartz, Phys. Rev. Letters 9, 448 (1962).
solid curve is theoretical. sr B. B. Goodman, Phys. Letters 1, 215 (1962).
22 R. M. Bozorth, A. J. Williams, and P. D. Davis, Phys.
Rev. Letters 5, 148 (1960).
23 C. P. Bean and M. V. Doyle, J. Appl. Phys. 33, 3334
Figure 4(a) shows the penetration of Aux into a
"
sintered sample of Vaoa. The specimen was pre-
(1962).
24 Y. B. Kim, C. F. Hempstead, and A. R. Strnad, Phys.
pared by pressing a stoichiometric mixture at 50 000 Rev. 129, 528 (1963).
» C. P. Bean, M. V. Doyle, and A. O. Pincus, Phys. Rev.
psi and firing it in argon at 1150'C for 3 h. The ex- Letters 9, 93 (1962).
26 R. J. Charles and W. A. Harrison, Phys. Rev. Letters ll,
rs Z. E. Kunzier, Rev. Mod. Phys. 33, 501 (1961). 75 (1963).
 REvIEws or MoDERN PHYBIcs JANUARY 1964

a wide range of 6eld H could be derived from a cur- have components of the voltage at harmonics of the
rent density, J(H) = n/(H+ + H). In this expression impressed frequency. Obser vation of these har-
0. and H+ are constants. The fact that B+ was found monics can, in principle, give a measure of the minor
to be = 5 kOe accounts for the good fit we obtain in hysteresis loop and hence of the critical current
our low-field measurements using the assumption density. %e have tacitly assumed that these critical
that the current density is independent of field. currents are independent of the velocity of the mo-
—
tion of Qux an assumption that is unjustified for the
entire frequency range of the electromagnetic spec-
4r M {GAUSS)
trum. The frequency dependence of these voltages
4000-- will indicate to what extent this assumption is valid.
To derive the magnitude of the harmonics of volt-
age we must perform a Fourier analysis of the voltage
H p 5490
2000-- waveform. First we note that the instantaneous Qux
density, from Eq. (5), is

B = B(H) + hhp/H*(H) ~ (h,' —h', )/2H*(H)

(cylinder) hp «HP (9)
with the plus sign for the course from to +hp. —hp
We assume that h = hp cos &pt = hp cos (2prft), which
-X00--
gives

-40X-- B = B(H) + [hp/2H*(H)][(2 cos pet & sin' &pt)] . (10)

If we assume an expansion of the form
FIG. 5. Magnetization curve and hysteresis loops for lead
in porous glass at 4.2'E. The points are experimental while
the solid curves are theoretical with an assumed current
density of 2 7 X 104 A/cms.
B = B(H) + eg n„i sn(nut) + n~l
1
gp cos (ntpt),

(11)
THE RESPONSE OF HIGH-FIELD we may evaluate the coefficients n. and p. by the
SUPERCONDUCTORS TO ALTERNATING usual technique of multiplying B by sin(mppt) or
MAGNETIC FIELDS cos(mtpt) and integrating from cot = 0 to tet = 27r.
The limitations of static magnetization measure- The results are"
ments may be circumvented by a technique that n. = 0, (n even);
measures the response of the superconductor to small
alternating fields that are superimposed on a col-
n = —[hp/H*(H)] [4/(n —2) (n) (n + 2)], (n odd)
linear steady field. In this case superconducting cur-
rents are induced to flow in a thin layer (the thick-
Pi = hp/H*(H); P. = 0 (n ) 1) . (12)
The voltage output for a secondary coil of X turns is
ness is proportional to the amplitude of the alter-
'
nating field) at the surface of the sample. Further-
more, these currents are Qowing in fields that are
identical to the static applied field within an error of
plus and minus the amplitude of the alternating field.
V

which
= prR'NB && 10 =
+ Vs s cos (3&et) + Vpth
gives in conjunction
Vr. cos
& (&vt

cos (5cvt)
—y)

with the definition

+, (13)
of
Lastly, a Qux jump will allow Qux to enter the bulk H*(H) for a cylindrical specimen
of the specimen but, hopefully, will not affect the —— 1.088 Vs, d
V1.t
response to alternating fields once the catastrophic
voltage pulse has occurred. Vs.s = —[4hpfXR/J. (H)]10 ' volts
If at some steady field H a collinear alternating V5th —0.238 V3,d
field of amplitude A0 is applied, the magnetization
V7th = 0.1 1 1 VartI
traverses a minor hysteresis loop. If the alternating
field were a pure sine wave, the dependence of B on V. = [5/(n —2)(n+ 2)]Vs,e.
~h (14)
time would consequently be nonsinusoidal. A sec- By application of Eqs. (14), J,(H) may be de-
ondary coil wound on the sample would have a non-
sinusoidal voltage induced across it and thus would
sr D. B. DeHaan, Nouoelles taMes d'integrales definies (Leide,
Pa~is, 1897), p. 97.
 CHARLES P. BEAN Magnetization of High-Field 8uyerconductore 37

termined or alternately this technique may be used lead was forced in at 855'C and 60000 psi filling
to generate harmonics. completely the 81% void space of the glass. Electron
EXPERIMENTAL ARRANGEMENT
micrographs show the filled pores to 110 A 20 A ~
in diameter while the pore junctions can be 200 A or
The steady field is produced by a Nbe. »Zre. » super- greater. For later reference we note that the pore
conducting coil that is powered by Trygon power
supply model M86 —
1000—
25-ov. The alternating field is Pb IN POROUS GLASS
Oe AMPLITUDE 5 kc FIELD

produced by a primary coil of 541 turns of 8-mil

wire wound on a plastic form of 0.185-in. diameter. +—880e

The primary coil is powered by a 60-W McIntosh FACTOR

audio amplifier driven by an audio oscillator. The UJ

CO
IOO
a —440e OF FOUR

alternating field is monitored by a Ballantine vacuum cC cA

o~+
—
I

tube voltmeter that senses the voltage across a 10-0 —O

I

—
a 22 Oe
noninductive resistor that is in series with the pri- O

mary circuit. The secondary circuit consists of 101

turns of 4-mil wire that is wound directly on the
0.125-in. rod of porous glass impregnated with lead. o —Oe
II
I I

The output of this coil is measured by a Hewlett— 6 00 STEADY FIELD-00 00 30 000

Packard harmonic wave analyzer, model 300' that Fre. 7. Third harmonic output from a sample of lead in
has a Keithley model 102A decade preamplifier con- porous glass exposed to a 5-kc/sec alternating field of various
amplitudes in a steady field II. The sample is that of Fig. 6.
nected before it in the circuit. A series LC filter is
used in the primary circuit to remove extraneous
harmonics and a small mutual inductor is connected structure of a thin surface layer of the porous Vycor
between the primary and secondary circuits to limit is coarser than that of the bulk.
the output to the wave analyzer of voltage of the '
EXPERIMENTAL OBSERVATIONS
fundamental frequency.
To test the equations developed above, we meas-
Pb IN POROUS GLASS ured the harmonic output at 4.2'K in a steady field
of ll 000 Oe. The results are shown in Fig. 6. The
I000
observations cover a range of 500 cps to 5 kc/sec.
(We are limited on the upper end by the 16-kc/sec
limit of our analyzer. ) The alternating fields ranged
from 10 to almost 200 Oe in amplitude. The voltages
were proportional to frequency over the entire fre-
+X
100
—
quency range indicating no measurable viscosity in
o~ flux motion. In another experiment with a 100-kc/sec
I 0 tuned amplifier we extended this conclusion to 88.8
kc/sec. The voltages are closely proportional to the
10
square of the alternating field at larger values of the
field with a noticeable fall off at lower fields as ex-
emplified by the points at the lower left-hand corner
I.O
IO 100 l000 of the figure. The solid line marked "3rd harmonic"
hp f X IO (OE SEC } is that calculated from Eq. (14) with a critical cur-
FD". 6. Harmonic output from a sample of lead in porous rent density of 2. 1 X 10' A/cm' referred to the total
glass in an alternating field. The ordinate indicates the voltage volume of the specimen. The solid lines marked 5th
output of a 101-turn coil wound on the 0.32-cm-diam specimen.
The abcissa is the product of the square of the amplitude of and 7th are derived from Eqs. (14) using the same
the alternating Geld hp and the frequency. The experimental current density. %hile those for the fifth harmonic
oints plot the output at odd harmonics of the fundamental
~

~ ~
requency. The solid lines show the theoretical predictions for
~
fall on the theoretical curve, the two measured sev-
a current density of 2. 1 )& 105 A/cm .
~

enth harmonic points are 50% low. Figure 7 shows

the variation of the third harmonic of a 5-kc/sec
The sample was of porous Vycor glass that had fundamental as the steady Geld is swept over a wide
been filled with 10% hydrofluoric acid and left under range. The amplitude of the alternating Geld is
mineral oil for 108 h. Subsequently, reagent-grade changed by a factor of two for each run. It, is seen
38 REVIEWS OF MODERN PHYSICS JANUARY 1964

that the upper three curves are exactly a factor of nates at 17.5 kOe because the Nbp. »Zr&. »coil carries
four apart, as required by the theory, above about smaller superconducting current, s at lower tempera-
4000 Oe while the lower curves progressively deviate tures. The current density is seen to decrease
in the sense of exhibiting a smaller voltage than that linearly with increasing Geld. Another run, not shown
predicted by the theory. The even harmonic voltages here, carried the 4.2'K run to 28 kOe. The points
are buried in the noise in this range above 4000 Oe, followed the extrapolated line very closely with a
while the noise itself is a strong function of steady slight concavity at high fields. The extrapolated val-
field and amplitude of oscillating field. Apparently, ues are H&(4. 2'K) = 26.0 kOe and H„(8.0'K)
the noise, which is quite flat with frequency, arises 32.4 kOe while the current densities extrapolated to
from the fact that the hysteresis loop is not, in de- zero field are Jo(4.2'K) = 3.5 && 10' A/cm' and
tail, a smooth curve as we envisioned it but rather is Jo(3.0'K) = 6.0 X 10' A/cm'.
a series of tiny flux jumps as the Aux breaks through
DISCUSSION
the individual threads of the matrix. If this be so it
is a rather close analog to the Barkhausen noise in While it is somewhat risky to discuss the experi-
ferromagnets that arises from discontinuous motion mental results on one specimen, several points can
of domain walls. Lastly, as was anticipated, gross be made. First, the fact that for low driving fields
lux jumps have no eGect on the voltages other than the harmonics are not present to the expected
to give a momentary blast over the entire spectrum. amount may be an indication that for low enough
The lower end of Fig. 7 with its peak in voltage driving fields the magnetization becomes reversible,
appears, at first sight, to be very interesting. In- i.e. , the Geld merely varies with the quasi-London
itially I thought that this indicated a nonmonotonic penetration mentioned earlier. Figure 7 suggests that
course of critical current as a function of steady Geld. the critical field for this process is a few oersteds. A.
This is not true inasmuch as the scaling relationships second point is the surprisingly simple dependence
do not apply in this region. These peaks are insensi- —
of critical current on field a dependence that divers
tive to frequency and broaden at lower temperatures. sharply from the substantially hyperbolic depend-
Their cause is, I believe, simply the coarse surface ence observed in many high-field superconductors. "
layer noted earlier. If this layer were presumed to The observed dependence suggests a very simple
have a critical field of 3000 or 4000 Oe and a thick- model for the course of the current density as a func-
ness of 20 p or so, the main elements of this behavior tion of Geld. We assume that resistance appears
would be simulated. " whenever the local current density in a Glament at
any point rises to a critical value. The current density
5.0— is the vectorial sum of two components, the usual
Pb IN POROUS GLASS
London currents and the circulating currents in-
4, 0—
duced by motion of flux through the filament. If the
circulating current is perpendicular to the applied
~ 3, 0
field then these currents add and the total critical
2. 0
current is equal to the sum of the field-induced cur-
rent and the circulating current. This simple picture
is presented with some difFidence inasmuch as the
question of critical currents and fields is not a simple
IO 000 15
I

000
FIELD (Oe)
20 000 25 000
I

30 000
"
one. The actual value of the critical current density
within the lead filaments is not measured precisely
FIe. 8. Macroscopic critical current density plotted as a
function of the applied Geld II. The points are deduced from
J. by t, hese experiments but a lower bound may be es-
the data of Fig. 7. timated. At 3.0'K and zero Geld, the macroscopic
current density is inferred to be 6 )& 10' A/cm'. Since
Figure 8 shows the current densities derived from the lead forms 31% of the volume and, for a random
the data of Fig. 7 and another run at 3.0'K. The array, the current resolved in one direction is half
data are plotted only for those regions of steady field that flowing in the individual filaments, a lower
where Eqs. (14) are satis6ed. The 3.0'K run termi- bound for the microscopic critical current density is
~8 We have abraded
0.006 in. from the diameter of the about 4 )& 10' A./cm'. The actual value is probably
specimen and retested it. As predicted, the low-field voltage closer to 10' A/cm' owing. to the limiting effect of
peaks disappeared but, in addition, the voltages at no field
follow precisely the predicted h0 dependence. This may indi- constrictions. This value is of the same order of mag-
cate substantial damage in depth by the abrasion. nitude as that predicted theoretically but the ob-
 CHARLE8 P. Bm.N Magnetization of IIigh;Fietd Superconductors 39

served temperature dependence of the critical cur- ACKNOWLEDGMENTS

rent is about twice that which has been predicted. " G. E. Nichols carried out the experiments reported
in this paper with great skill and energy. We wish to
thank W. DeSorbo, P. E. Lawrence, D. 8. Rodbell,
CONCLUSION
and C. H. Rosner for loaning us various pieces of
The main conclusion of this work is that the mag- experimental equipment. R. J. Charles and W. D.
netization of hysteretic high field superconductors Schmidt prepared the porous glass specimens and
can be well understood in terms of one phenomeno- C. E. Van Buren made the sample of VsGa. Lastly,
logical parameter, the macroscopic critical current I wish to acknowledge many helpful conversations
density. A main problem remaining is the determina- with T. Buchhold, H. R. Hart, Jr. , J. D. Livingston,
tion of the relationship between this parameter and and P. 8. Swartz on the subjects covered in this
the microstructure of hard superconductors. paper.

Hard Superconductivity: Theory

of the Motion of Abrikosov Flux Lines
P. W. ANDERSON and Y. B. KIM
Bell Telephone Laboratories, Inc. , Murray Hill, New Jersey

The central concept in the theory of what one might centers, without going into unnecessary detail on the
call the "critical phenomena" of hard supercon- —
nature of the pinning centers whether they are dis-
—
ductors critical currents, critical fields, decay of —
locations, cavities, precipitates, etc. or the precise
persistent currents, "excess" voltages, etc. is — internal structure of the superconductor. Our task,
clearly Abrikosov's notion of the quantized Aux line. ' —
then, is to study the process presumably thermally
This is made almost obvious by the remark that, be- activated barrier penetration — by which flux lines
cause of the now universally accepted validity of the move.
quantization of flux through superconductors, the Let us then suppose that we have a supercon-
smallest possible breakdown of superconductivity is ductor penetrated by a magnetic field H and carry-
the motion of a single quantum of magnetic Aux ing a bulk current, for simplicity J H, J
through the wire or ring. Thus, in all cases so far cV x H/4z. The magnetic field will penetrate in
conceived, the lowest activation energy for any crit- the form of Abrikosov lines; their density is clearly
—
ical breakdown is that for the motion and creation, not uniform because of J, and we expect their ar-
if necessary — of single Abrikosov Aux lines. This rangement is to some extent irregular. The magnetic
statement is independent of whether the mechanism energy per unit volume is H'/8 wee can think of this
for hard superconductivity is the GLAG one or the as a magnetic pressure exerted by the Qux lines on
Mendelssohn sponge theory, although we assume the each other, and in the absence of pinning centers this
former to be valid in most cases. Even the decay of pressure would have to be equalized by a rearrange
currents in true soft superconductors under o.-par- ment of the lines, leading to J
= 0. Examination of
ticle bombardment' is probably best explained by Abrikosov's theory shows that actually at all but low
the threading of Abrikosov lines through normal fields the internal and external fields are nearly the
holes punched by the n particles. ' same, so that we usually assume 8 = H, a minor
The purpose of this paper is to see how many of simplification of which Friedel et at. have considered
the phenomena of hard superconductivity we can the errors. 4
understand qualitatively in terms of the thermally In finding the rate of the activation process we
activated motion of Abrikosov lines past pinning need to know two things: the driving force exerted
by the magnetic pressure, and the nature of the
~ A. A.
—
Abrikosov, Zh. Eksperim i Teor. Fiz. 32, 1442
(1957) [English transl. : Soviet Phys. JETP 5, 1174 (1957)]. barriers. The former is more available to us theoret-
z P. de Feo and G. Sacerdoti, Phys. Letters 2, 264 (1962).
s N. Cabibbo and S. Doniach, Phys. Letters 4, 29 (1963). 4 J. Friedel, P. G. de Gennes, and J. Matricon, Appl. Phys.
We have proposed a slightly different mechanism. Letters 2, 119 (1963).