HAL Id: jpa-00211107
https://hal.archives-ouvertes.fr/jpa-00211107
Submitted on 1 Jan 1989
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Aspects of high temperature superconductivity
Guy Deutscher
To cite this version:
Aspects
of
high
temperature
superconductivity
Guy
DeutscherSchool of
Physics
andAstronomy, Raymond
andBeverly
SacklerFaculty
of Exact Sciences, Tel AvivUniversity,
Ramat Aviv, Tel Aviv, Israel(Reçu
le 17 mai 1989,accepté
le 9juin
1989)
Résumé. 2014
Nous proposons
quelques
remarques sur les aspectsparticuliers
qui
distinguent
laphénoménologie
des nouveauxoxydes
àhaute Tc
de celle dessupraconducteurs
conventionnels. Ils comportent unelargeur
mesurable de larégion critique
et unegrande
sensibilité aux défauts cristallins. Unedescription
cohérente du type LandauGinsburg
estpossible
avec une courtelongueur
de cohérence 03BE ~ 15Å
et uneprofondeur
depénétration
03BB ~ 900Å.
Cette dernière estsensiblement
plus
courte que la valeur couramment admise, etimplique
l’existence d’une bande de conduction assezlarge.
Abstract. 2014 We
present some remarks on
special
features thatdistinguish the phenomenology
of the newhigh Tc
oxides from that of the conventionalsuperconductors. They
include a measurable width of the criticalregion
and ahigh sensitivity
tocrystallographic
defects. A consistent LandauGinsburg interpretation
itpossible,
with a short coherencelength 03BEab ~
15Å
and apenetration
depth 03BBL ~
900Å.
The latter is somewhat smaller than thecurrently accepted
value, andimplies
a broad band scheme.
Classification
Physics
Abstracts 74.30C - 74.40 -74.70V
Introduction.
While the
question
of the mechanism forhigh
temperature
superconductivity
is still very much opentheoretically,
recentexperiments
onsamples
ofimproved quality point
out more and moreclearly
to anambiguous
situation. On the onehand,
there isgrowing
evidence for aBCS
behavior,
with a well defined energy gap and conventional behavior of the Londonpenetration depth
[1].
On the otherhand,
the gap appears to be very sensitive tocrystallographic
defects on the atomic scalejudging
from the anomalousJosephson
behavior atboundaries,
whichclearly
indicate adepressed
orderparameter
[2].
This isquite
contrary
to the behavior of conventional
superconductors,
wherecrystalline
defects such asimpurities,
dislocations,
grain
boundaries do notmodify
thesuperconducting
stateexcept
in as much asthey
reduce the normal state mean freepath,
reduce the coherencelength
and enhance thepenetration depth.
Butthey
do notmodify
the criticaltemperature
and the gap.Accordingly,
all the fundamental data can be obtainedusing polycrystalline,
notparticularly
pure or defect freesamples.
2852
This is not the case for the new
high T,
oxides that must be studied insingle crystal
form for the determination of thepenetration depth,
the lower critical fieldHc1,
the gap, thedepairing
critical currentdensity
etc...This extrinsic behavior can be understood on the basis of a short coherence
length
[3],
by
now well documentedexperimentally
[4].
Extrinsic behavior due to
crystallographic
defects could also beinterpreted
as indications thatsuperconductivity
in thehigh T,
oxides is of an unconventional nature(d
wavesuperconductivity, charge
orspin
transfer modelsetc...).
Early
observation of an anomalous lowtemperature
heatcapacity
in the YBCOcompound
received such aninterpretation
[5].
But later measurements on the Bicompounds
did not show this behavior[6]
which istherefore not intrinsic to
superconductivity
in the oxides[6].
In the absence of direct
proof
of unconventionalsuperconductivity,
we shall take thepoint
of view that theirsensitivity
to defects isonly
due to a short coherencelength,
and that with this caveat theirproperties
can beinterpreted
in the framework of the LandauGinzburg
theory.
We shall discuss in thefollowing
whether the available data can beinterpreted
consistently
within thisframework,
concluding
with some remarks on thephysical significance
of the short coherencelength.
Landau
Ginzburg
parameters
of thehigh
T,
oxides.The
strongly
anisotropic quasi-2D
structure of the oxides as well as their short coherencelength immediately
suggest
that non mean field behavior should beobservable,
in contrastwith conventional
superconducting
metals andalloys.
On the reducedtemperature
scale E =(T, -
T)/Tc,
there should be a detectabletemperature
range Ec withlarge
ther-modynamic
fluctuations of the orderparameter
andcorrespondingly
non mean field values of the criticalexponents.
Taking
a 3Daveraged e
of 12Â
and athermodynamic
critical fieldHc
= 12.000G,
theGinzburg
criteriongives Ec
of the same order as that forsuperfluid
He[7].
Fluctuation effects are further enhanced in 2D. The measured7c
should then besignificantly depressed
relativeto its mean field value.
Specific
heat measurements on YBCOsamples
dodisplay
fluctuation effects nearT,.
In addition to the mean fieldjump,
there is a small but definite enhancement of the heatcapacity
both above and belowTc,
that has beeninterpreted
asqualitative
evidence for aÀ transition
[8].
The available data is notyet
of sufficient accuracy to allow a determination of criticalindices,
but is sufficient to establish that the width of the criticalregion
is indeed muchlarger
than for conventionalsuperconductors,
but smaller than that ofsuperfluid
He. This is confirmedby
two additional sets of measurements : that of thepenetration depth
which follows withgreat accuracy
mean field behavior in the entire range of measurements,up to 0.5 K from
T,,
[9] ;
and that of the excessconductivity
aboveT,,
which is ingood
agreement
with the mean field behavior of alayered superconductor
withgab
= 12Â,
e,
= 2Â,
also up to 0.5 K fromTc
[10].
The
Ginzburg
criterion can be written under the form :where A is a numerical
coefficient,
00
the fluxquantum, e,
is the coherencelength along
the c axis andÀab
thepenetration depth
in the(a, b)
plane.
Here we haveexpressed
the condensation energy per coherent volume( Hc/8 2 -u) . Pl using î3 = e ab . 2 ec,
and haveEquation
(1)
can be rewritten as :The numerical factor
(A e c 112)
is in any case smaller than one.Taking
forç c
an upper limit of 5Â,
and Tc
= 92 K(YBCO),
weget
À : 900Â.
This is smaller than the
currently accepted
value À = 1 400Â,
andonly barely compatible
with the lower end of the
experimental
bracket of ±500 Â
[9].
There are indeed some
experimental
difficulties that may lead to an overestimate ofA. For instance it is well known that anomalous low field
penetration
and microwaveabsorption
(discussed below)
arealways
observed in theoxides,
even wheninsingle crystal
form. These effects must be due to the presence of weaksuperconducting regions,
that arenecessarily
the sites of alocally
enhancedpenetration depth.
We also note that
reflectivity
measurements near theplasma
edge
in YBCO lend a value of theplasma
frequency
thatcorresponds
to À =(c / lJ) p)
= 700À
[11].
The
point
that we aremaking
here is that a consistent LandauGinzburg description
of the oxidesrequires
apenetration depth
somewhat smaller than the valuegenerally accepted.
It could of course be the case that such adescription
is notapplicable,
but one would then haveto
explain
the excel-mean field fit of À(T).
A
large
k value indicates alarge
effective mass, aspointed
out in[9],
and is indicative ofstrong
coupling.
Conversely,
a smaller À is consistent with a broad band and weakcoupling.
As is wellknown,
theOrsay
school hasargued
in favor of such a broad bandscheme,
with the Fermi level located near the Van Hovesingularity
of aquasi
2D band[12].
We have shown that a consistent LandauGinzburg description
of the oxides favors this laterpoint
of view. More efforts should be devoted towards a definiteexperimental
determination of thepenetration depth,
and to its calcuIation in the different theoretical schemes.Boundary
effects : low field microwaveabsorption.
Low field microwave
absorption
in thehigh Tc
ceramics[13]
and insingle crystals
[14]
point
out to the existence of
Josephson
junctions
in these materials. Periodicabsorption
lines have been observed in YBCOsingle crystals,
with a fieldperiod
that is a minimum when the field isparallel
to a(110)
direction,
and varies as(cos 0 )
where 0
is theangle
between the field and a(110)
direction. The well known(110)
twin boundaries of YBCO are thus involved. Theproposed interpretation
[14]
is aperiodic penetration
of vortices in onejunction,
or a series ofjunctions, parallel
to one set of twin boundaries. But in view of thehigh density
of twin boundaries in thecrystal,
it is not clear how oneparticular boundary
could dominate theabsorption
behavior.Some further
puzzling
features of theseexperiments
are :i)
the lineperiodicity corresponds
to an effective
junction
width that is muchlarger
than theexpected penetration depth
andii)
the threshold fieldHe1J
for vortexpenetration
has an anomaloustemperature
dependence.
For a conventionaljunction :
where
where
Àj(r)
is thepenetration depth
in thejunction
andk L (T)
the Londonpenetration
2854
Since
Jc (T)
oc(F, - T )
andÀL(r)
oc(F, -
T)- 1/2,
it follows thatHe1J
(T)
oc(Tc -
T)3/4.
Instead,
theexperiments
ofBlazey give
Hlj (T)
oc(Tc - T)’,
with n = 1.9 ± 0.3[14].
We wish to show here that this can be the behavior of acomposite junction composed
of a stack of twin boundariesseparated by
distances of the order of 100Â,
as is sometimes observed in YBCOgrains
[15].
First,
it has been shownpreviously
that the critical currentacross a
boundary
in ahigh Tc
oxide varies nearT,
as(Te 2013
T)2
rather than(Tc - T).
This is due to the short coherencelength
whichproduces
at theboundary
adepression
of the orderparameter,
ài
oc(Tc - T)
instead of thebulk à,
oc(T, -
T)1/2
[16].
A second modificationoccurs if we are
dealing
with a stack of boundariesseparated by
distances d smallcompared
toAL-
In that caseAL(T)
has to bereplaced by
the effectivescreening
distance(d/2)
inequation
(4)
[17].
With thesemodification,
we obtainHelJ
oc(T, -
T)3/2@
in much betteragreement
withexperiment.
The low field microwave
absorption
in thehigh Tc
oxides appears to be a very useful way tostudy
theirspecial electromagnetic properties.
We haveproposed
that stacks of twins can leadto the observed behavior.
Blazey
notes that the series ofabsorption
lines varies withsamples
ageing
andtemperature
cycling
[14],
and we remark that this is inkeeping
with the knownmobility
of the twin boundaries in the oxides.They
mayeventually
becometrapped
indefectuous
regions
of thecrystal, leading
to stack formation. Thishypothesis
needs of courseto be confirmed
by
direct observation. Short cohérencelength
and Coulomb effects.We have so far
argued
that the available data on the oxides can bereasonably
wellinterpreted
within a LandauGinsburg
framework,
with a rather conventional Londonpenetration depth
and a short coherencelength.
In this lastsection,
we wish topoint
out to someimplications
of thisshort e
concerning
the electron-electron interaction.In its
simplest
form,
theexpression
for the criticaltemperature
of a BCSsuperconductor
is :where
hw ex
is an excitation energy and :N
being
the normal statedensity
of states at the Fermilevel,
Ve,,
the attractivepotential
that results from theexchange
of the excitations oftypical
energyhw ex’
andV c *
the effective screened Coulombrepulsive potential.
Effective
screening
requires
that the distance between the two members of aCooper
pair
atthe time where the excitation created
by
the first one is absorbedby
the second one, be muchlarger
than atypical
distancekp 1 :
This can be rewritten as
EF IÍúJex.
Now,
since the coherencelength eo
isequal
to(V F/ £0 .)
wherew g
is the gapfrequency,
and sinceúJg
« úJex, condition(7)
canonly
be fulfilled if :In a conventional
superconductor,
lùex is aphonon
frequency,
(EFlh£oe,,)
is of the order of a fewhundred,
and the bare Coulombrepulsion
istypically
reducedby
one order ofmagnitude.
As is wellknown,
this is the reasonwhy
the condition forsuperconductivity
Ve , >
V c * is
metin so many metals and
alloys.
It has
long
been advocated that an effective way toget
high
temperature
superconductivity
is to raise the
perfactor
inequation
(5),
i.e. to havehlùex
of the order of the Fermi energy. Butit was also
recognized
that this was not asimple prescription,
since one would then at the same time haveVc* =
Vc
andaccordingly
a small effective NV[19].
There is still considerable
controversy
about the value ofEp
in the oxides(broad
or narrowband) ;
and we do not know at all that ofhwex,
since the mechanism forsuperconductivity
in the oxides is not known. But we do know from critical field measurements andboundary
effects[16]
that ) =
10Â,
and hence that condition[8]
is not well fulfilled and therefore condition[7]
even less.The
high
criticaltemperature
of the oxides and their short coherencelength
areexperimental proofs
that asignificant
attractive electron-electronpotential
can exist eventhough
the classical Thomas Fermiscreening
is weak(this
is also confirmedby
the smallisotope
effect).
What we don’t know is how this is achieved.Or,
in otherwords,
electron correlation effects cannot beignored
in the oxides. Aninteresting question
is whether thehigh 7c
model based on the 2D Van Hovesingularity
[12],
which seems to fit well the LandauGinzburg description
of theoxides,
can be made morequantitative
by including
such correlationeffects,
and stillgive
ahigh T,.
Conclusions.
Superconductivity
in the oxidespresents
many unusual features. Because of the short coherencelength,
critical behavior is observable.Experimentally
the width of the criticalregion
issignificantly
smaller than insuperfluid
He ;
a coherent LandauGinzburg description
thenrequires
apenetration
depth
smaller than thecurrently
accepted
experimental
determination. Theappropriate
value would be close to thatgiven by
the Londonexpression
with no mass enhancement. This would be consistent with a broad band scheme. The short coherencelength
also favorizesjunction
formation at extendedcrystallographic
defects. Thismight
lead to thesuggested
overestimate of theexperimentally
determinedpenetration
depth.
The low field microwaveabsorption
features seen in twinned YBCOcrystals
can beexplained
if one assumes the existence ofcomposite
junctions
made of stacks of twin boundaries.Finally,
the short coherencelength implies
the existence of a mechanism that can reduceeffectively
the Coulombrepulsion, beyond
the usual Thomas Fermiscreening.
Acknowledgements.
I wish to thank K.
Blazey
forstimulating
discussions on low field microwaveabsorption,
and forcommunicating
his resultsprior
topublication.
This article has been
prepared
in honor of ProfessorJacques
Friedel. 1 wish to express tohim here my
deepest gratitude
for the marvelous time that 1 havespent
atOrsay
as one of hisstudents,
discovering
the richness andbeauty
of Solid StatePhysics
under hisleadership.
References
2856
[2]
MANNHART J., CHAUDARI P., DINGER T. R., TSUEI C. C. and MCGUIRE T. R.,Phys.
Rev. Lett. 69(1988)
2416.[3]
DEUTSCHER G. and MÜLLER A.,Phys.
Rev. Lett. 59(1987)
1745.[4]
DEUTSCHER G., Proc. of the M2HTSCConference,
Interlaken 1988, Eds. J. Müller and J. L.Olsen, Physica
C153-155(1988)
15.[5]
ANDERSON P. W., Proc. of the M2HTSC Conference Interlaken 1988, Eds. J. Müller and J. L.Olsen, Physica
C153-155(1988)
527.[6]
FISCHER R. A., GORDON J. E. and PHILLIPS N. E., J.Superconductivity
1(1988)
231.[7]
DEUTSCHER,
G., NovelSuperconductivity,
Eds. S. A. Wolf and V. Z. Kresin(Plenum,
NewYork)
1987, p. 293.[8]
FOSSHEIM K., NES O. M., LAGREID T., DARLINGTON C. N. W., CONNOR D. A., and GOUGH C. E., Proc. of the Adriatico Research Conference, to bepublished
in Int. J. Mod.Phys.
B.[9]
KRUSIN-ELBAUM L., GREENE R. L., HOLZBERG F. and MALOZE-MOFF A. P.,Phys.
Rev. Lett. 62(1989)
217.[10]
KITAOKA Y., ISHIDA K., AZAYAMA K., KATAYAMA-YOSHIDA H., OKABE Y. and TAKAHASHI T.,preprint.
[11]
BONTEMPS N., FOURNIER D., BOCCARA A. C., MONOD P., ALLOUL H., ARABSKI J. and DEUTSCHER G.,preprint.
[12]
FRIEDEL J., Proc. of the M2HTSC Conference, Interlaken, Eds. J. Müller and J. L. Olsen,Physica
C153-155
(1988)
1610.[13]
BLAZEY K. W., MÜLLER K. A., BEDNORZ J. G., BERLINGER W., AMORETTI G., BULUGGIN E., VERA A. and MATTACOTTA F. C.,Phys.
Rev. 36(1987)
7241.[14]
BLAZEY K. W., to appear in the Proc. of the 9th General Conference of the Condensed MatterDivision of