HAL Id: jpa-00247310
https://hal.archives-ouvertes.fr/jpa-00247310
Submitted on 1 Jan 1996
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.
On a Possible Reason of Tc Suppression with Oxygen Doping in Tl2Ba2CuO6+x
I. Schegolev, N. Kolesnikov, V. Kopylov, T. Togonidze, O. Vyaselev
To cite this version:
I. Schegolev, N. Kolesnikov, V. Kopylov, T. Togonidze, O. Vyaselev. On a Possible Reason of Tc Suppression with Oxygen Doping in Tl2Ba2CuO6+x. Journal de Physique I, EDP Sciences, 1996, 6 (12), pp.2265-2273. �10.1051/jp1:1996216�. �jpa-00247310�
On a Possible Reason of Tc Suppression witl~ Oxygen Doping in
T12Ba2Cu06+x
1.F. Schegolev, N-N- Kolesnikov, V.N. Kopylov, T.G. Togonidze
and O.M. Vyaselev (*)
Institute Of Solid State Physics, Russian Academy Of Sciences, Chemogolovka. Moscow District,
142432 Russia
(Receii.ed 26 April 1996, revised 23 Julj.1996, accepted 19 August 1996)
PACS.74.25.Ha Magnetic Properties
PACS.74.15.Nf Response to electromagnetic fields (nuclear magnetic resonance, surface impedance, etc.)
PACS.74.62.Dh Elfects of crystals defects, doping and substitution
Abstract. Magnetic susceptibility and ~~Cu and ~°~Tl nuclear spin-lattice relaxation rates bave been measured in T12Ba2Cu06+x high-temperature superconductor to establish the mech-
amsm of decreasing the transition temperature with oxygen doping. The experiments have revealed the presence of localized paramagnetic centers (LPC) whose concentration grows with
mcreasing oxygen content. These centers con give rise to magnetic scattenng of the Cooper pairs and thus be responsible for the T~ suppression. This assumption is supported by the
pair-breaking elfects observed in the NMR Knight shift and relaxation rate behavior m the
superconducting state.
1. Introduction
It is commonly accepted that trie Tl:2201 system, T12Ba2Cu06+z, is an overdoped une smce
its transition temperature bas been repeatedly proved il, 2j to decrease considerably, clown to absolute zero, with increasing trie oxygen content. In fact, it is one of two HTS systems
(another one is LaSrCUO (3j) that provides an example of existing of a state what is often referred to as a fully overdoped, 1.e. metallic but non-superconducting une. At trie same time, trie question on the real noie concentration m this system and, m particular, on its relation tu trie oxygen content is not quite clear. Figures for the oxygen index at a given T~ reported by diiferent authors ailler from one another considerably making trie calculation of trie noie number rather unrehable. The more so that some non-stoichiometry charactenstic of this system consisting in a partial substitution of Cu for Tl in Tl-O planes (4j is tu be taken into
account.
Increasmg conductivity [si and trie normal state spm part of trie Knight shift for 6~Cu and 2°5Tl [6j with decreasing T~ is usually regarded as a sign of corresponding mcrease m trie noie concentration. However, heat capacity measurements (7j reveal no diiference in trie density of states in Tl:2201 samples with diiferent T~'s. Similarly, m a recent publication [8j Bazhenov (*) Author for correspondence je-mail: vyasel@issp.ac.ru)
© Les Éditions de Physique 1996
2266 JOURNAL DE PHYSIQUE I N°12
and Rezchikov bave found that trie plasma frequency, of trie order of1 eV, remains unchanged
m going from a T12Ba2Cu06+« single crystal with ~ = 0 and T~ = 100 K to that with x
= 0.18
and T~ = 30 K. They bave inferred that if trie mass of carriers is not specifically adjusted to trie change in their concentration, trie noie number remains constant within some 5 to 7i~.
In this paper we present trie results of susceptibility measurements and NMR investigation of Tl:2201 system which show that increasing oxygen content is accompanied here by increasing
amount of localized paramagnetic centers. Their existence shows up m a Curie-like contribution to trie magnetic susceptibility and in a strong field dependence of both 2°5Tl and ~~CU nuclear
spin-lattice relaxation rates at low temperatures combined with an essentially non-exponential
m time initial stage of trie CU nuclear magnetization recovery after an inversion puise.
This fact atone may explain depressing Tc ~vith increasing trie oxygen content, and observed
m this process diminishing trie ratio /l/T~ [6] is a testimony in favor of this explanation. At trie same time, weakening, with oxygen doping, electron 8pin correlations which are supposed
sometimes [9] to provide a pairing mechanism in trie high T~ superconductors may also be inferred front our expenments as evidenced by simultaneous slightly diminishing trie normal state nuclear spin-lattice relaxation rate and increasmg (lie spin part of trie normal state Knight
shift.
2. Experimental
Susceptibility measurements were carried ont on two groups of T12Ba2Cu06+~ single crystals
with T~ = 94 and 26 K. Susceptibility of trie crystals bave been measured by means of a home-made vibrating sample magnetometer (VSM) in a field of 0.î T.
Tire main part of trie VSM is a piezoceramic vibrator which consists of a piezoceramic tube and a plate-form bronze spring. Another piezoceramic element attached to trie other end of trie quartz tube holder bas been used in a feedback loop for an amplitude and phase control.
Trie crystals were attached at the bottom of a quartz tube holder. Trie vibrating frequency
determined by mechanical resonance of trie whole assembly was of trie order of 35 Hz. Trie usual vibration amplitude was about 3-4 mm. Signal from trie pickup coil is detected at trie second harmonic, which provided trie increase of trie magnetometer sensitivity up to 10~/
emu.
Diamagnetic response of superconducting samples in Meissner phase was used for trie absolute calibration of trie VSM. We bave used results of detailed X-ray analysis (2] to relate trie oxygen index with T~. According to this, it is equal to à-o in trie samples with T~ near 100 K and 6.18
m those with T~ near 30 K.
For trie NMR experiments, trie oriented polycrystalline samples with T~ = 100 and 32 K bave been used. Trie pellets were prepared by crushing about là perfect single crystals of trie corresponding phase into thin powder, which afterwards had been mixed with Stycast
1266 epoxy and placed into trie 7 T magnet. Trie nudear spin-lattice relaxation rates bave
been measured with Bruker MSL 300 puise spectrometer using standard inversion-recovery and saturation-recovery techniques.
3. Resuits aud Discussion
3.1. STATIC MAGNETIC SUSCEPTIBILITY. Trie normal state susceptibility of Tl:2201 single crystals was found to contain a Curie.like contribution, ,c~~~~
= GIT; which bas appeared to be isotropic within tire experimental errer and growing with increasing oxygen content. In
Figure 1, this contribution for trie H [ (ab) orientation is plotted ~ersus 1/T for trie single crystals with T~ = 26 and 94 K. The Curie constants, C, obtained from these curves are
,É
/ /
~ /
l /~
P
£ P
~ ~
© M
/
__-' _,""
'~
D ~
0.00 0.02 0.04 0.06
1/T(K.1)
Fig. l. The Curie.like contribution to magnetic susceptibility of Tl:2201 single crystals with Tc = 94 K (C) and 26 K (o) agamst inverse temperature.
o
' Q
'
Q '
' '
'
_
° '
~ '
- '
~~ "
'
O ' '
° ° '°°
'
0 2 4 6
n (%)
Fig. 2. Critical temperature us. the number of spm 1/2, g
= 2 locahzed paramagnetic centers per Cu site in Tl:2201 single crystals.
340 x 10~6 K emu 79 x 10~~ K emu, respectively., which corresponds to concentrations nrn -S
4.6~ and 1.1~ of spin 1/2, g = 2 localized paramagnetic centers pet CU site.
In Figure 2, trie transition temperatures of ail surgie crystals investigated are plotted ~e~sus
nn,. Considerable variation of nrn was observed within trie groups of crystals with close T~'s, which may result from a non-uniform distribution of trie paramagnetic centers. Trie correlation
betweeii trie concentration of trie paramagnetic centers and T~ exists aise when the oxygen stoichiometry is changea in a given crystal. Two surgie crystals with Tc -S 30 K hâve been annealed for 12 heurs at 350 °C in vacuum (-J 10~~ atm) and two others, with Tc
= 94 K,
for î heurs at trie same temperature under oxygen pressure of
-J 2 atm. As a result, trie transition temperature of trie first pair of crystals mcreased up to 75 80 K and that of trie second pair decreased clown to 20 K. In Figure 3, trie xcur~e vs. 1/T curves are depicted for one
2268 JOURNAL DE PHYSIQUE I N°12
,o
/ J°
,n'
R "
p ' D
/~ 'D
~
©
~ l'
,' D
'
" ~
"
"
0.00 0.02 0.04
1/T(K-1)
Fig. 3. The Curie.like contribution to magnetic susceptibility of Tl:2201
us. inverse temperature ai
o-1 T: (o) Tc = 26 K (as grown), C
= 4.4 x 10~~ K
emu: (Zh) after annealing in vacuum, Tc m 65 K, C = 1.9 x lo~~ K
emu; (D) after the recovery annealing in oxygen, Tc m 30 K, C
= 2.8 x 10~~ K emu.
of these samples, where T~ was increased and then decreased back. As is seen from Figure 3, these changes in trie transition temperature are accompanied by trie corresponding variations of trie Curie-like contribution to susceptibility.
3.2. THE NUCLEAR SAIN-LATTICE RELAXATION. In trie whole temperature range (4.2
300 K) 2°5Tl (spm 1/2) magnetization growth after trie saturation puises comb was smgle- exponential, though for below T~ some slight deviations from exponentiality bas appeared.
As we bave reported recently [10], ~°5Tl nudear spin-lattice relaxation rate, 2°5W, is field- dependent at low temperatures for any sample orientation. 2°5W field dependence for trie
samples with T~ = 100 and 32 K at T
= 4.2 K and H [ (ab) is depicted in Figure 4. It is seen that trie relaxation rate grows with diminishing field and trie magnitude of trie elfect is langer
for trie sample with Tc
= 32 K. Upon heating, trie field dependence weakens and disappears
above 50 K in trie "100 Il" sample and above 100-120 K in trie "32 K" one.
Unlike 2°5Tl, trie kinetics of ~~Cu relaxation
was found strongly non-exponential at low temperatures. In Figure .5, trie initial stage of ~~Cu recovery is plotted ~ersits t~/~ (t is delay
between trie inversion puise and spin-echo peak) for trie two samples at T
= 4.2 K and H
= 1.8
and 7 T. One con easily see that at short times trie NMR signal grows as t~/~, while in
exponential case it should rather behave bene linearly in time (e~~ c~ t at t - 0). Linear
interpolation of trie data, for 7 T field plotted in Figure 5 gives trie slopes -S 0.24 and 0.5î for trie samples with T~ = 100 and 32 K, respectively. Like in trie case of ~°~Tl, trie field dependence is aise present in trie relaxation of ~3CU: as it con be seen in Figure 5, trie curves
measured at lower fields bave steeper slopes indicating to trie relaxation rate mcreasmg with
dimimshing field.
In trie mixed state of type-II superconductors, trie additional field-dependent source of trie nuclear relaxation con be associated with vortices (for example, see a review by MacLaughlin iii]). Particularly, thermal motion of fluxoids con grue rise to trie contribution to W with a
negative field depeiidence (due to stilfening of trie fluxoid lattice with field) resembling that observed for 2°5W (Fig. 4). However, m our experiments we bave applied trie field parallel
to conductiiig loyers. H [ (ab). For this geometry fluxoids m strongly anisotropic layered
j's
300
fi
D T =100K~
~ /=32K 20
~ ~
i o
0
0 2 4 6 8
H (T es la)
Fig. 4. ~°~T1 spin-lattice relaxation rate versus magnetic field in the samples with T~ = 100 K la)
and 32 K (O) at 4.2 K, H ii (ab). Sand fines: best fils according to expression (2).
OR fia
- ~ ~D
~ ~ D
'fi
~
ù
-1
0 2 4 6 8 10
fi/3 (m8eCl/3)
Fig. 5. The ~~Cu NMR spm-echo amplitude after the delay t from the inversion puise as a function of t~/~ in the samples with Tc
= lo0 K (C) and 32 K (O), measured ai 4.i K in fields î T (open
symbols) and 1.8 T (dotted symbols): R ii (ab).
superconductors are intrinsically pmned and trie fluxoid motion-mduced relaxation rate is small [12j. Moreover, in the sample with T~ = 32 K trie field dependence persists well above the transition temperature up to 100-120 K that indeed cannot be associated with vortices.
Thus we assume trie field-dependent contribution to trie nuclear spin-lattice relaxation, as
well as the non-exponential relaxation kinetics of ~~Cu observed in
our expenments bas some dilferent origin. Namely, it con appear due to the presence of trie locahzed paramagnetic
centers (LPC). Actually, the dipole-dipole interaction of a nuclear spm with a spm 1/2 LPC
residing at distance ~ from the nucleus results m relaxation with a rate [13j,
H~LPC = G~~. G
= Î~i~i th~
liil
)1 + i~~~~ 11)
2270 JOURNAL DE PHYSIQUE I N°12
where ~B is trie Boby magneton, kB is trie Boltzmann constant, ~vn = inH is trie nuclear reso-
nance frequency, ~n is trie nuclear gyromagnetic ratio. and T is trie LAC spin-Îattice relaxation
time (for high LPC concentrations, however, it can measure aise trie spin-spin correlations).
Trie factor [1- th~(pBH/kBT)] describes trie saturation of localized moments at high fields
and low temperatures, and ~fil + (un~)2j is trie amplitude of trie LPC Lorentz-spectrum fluc- tuations at frequency un. Trie relaxation rate of 2°5Tl (Fig. 4) was found to follow trie field dependence of equation il fairly well (see solid fines in Fig. 4). Trie experimental data are
be~t-fitted with trie formula.
~°~ÎV
=
a~G + ~°~W~, (2)
1N.here a is a fit parameter and 2°~~[ is a field-independent background (relaxation through
trie quasiparticle spin system). Trie best lits give a = 1-à x 10~~ and 4.4 x 10~~ cm~~ for trie samples1N.ith T~ =100 and 32 K, respectively, and 7 m 10 ns for both samples.
One would justly argue that trie factor ~~~ in equation (2) should cause a non-exponential behavior of an NMR sigiial after trie saturation since, depending on a distance from trie LPC dilferent nuclei would relax with diiferent rates and a set of relaxation rates would appear, while trie "~Tl signal recovery bas been found single-exponential as ~ve bave noticed above.
We explain this is due to a fast spin diffusion in ri spin system. Really, trie inhomogeneous broadening of trie spin il? 2°~Tl NàIR fine is about 50 kHz at î T (2°~w -S 174 MHz) in Tl:2201 (14], while its homogeneous linewidth, /l1/~ -J 10-15 kHz. This gap it is non too large
to weaken essentially trie eiiergy-conserving fiip-flop interaction between neighboring nuclear spms. Further we can estimate trie relaxation time for 2°~Tl spin at distance a m 4 x 10~~ cm (lattice unit in (ab)-plane in Tl:2201) from trie spin 1/2 LPC. ivith trie LPC spin-lattice
relaxation time 7 m 10~~ s, one gets from equation il) Ti(a) -S 3 x10~~
s for H = o-à T
and Ti(a) -S (2 6) x 10~~ s (depending on temperature) for H = 7 T. As one con see, Tila) < 7D, where ~D
+~ T2/30 is trie spin diffusion titre of trie nudear spin system (15] and trie nuclear spin-spin relaxation time T2 " 1/~/li /2
+~
10~~ s. In this case which is known as strong spin diffusion [16], fast energy transfer through trie nuclear spin system equalizes trie spm polarization, making thus the NMR signal recovery exponential. Furthermore, in contrast with the case of free spin diffusion (15] giving a H~~/2 field dependence for trie LPC-induced relaxation rate, for the strong spin diffusion calculations [16] predict W c~ H~2 that is observed
m ouf experiments.
Within this scenario, trie nuclear relaxation con depend on LPC concentration due to trie factor r~~ (Eq. (l)) [lîj. Trie fit parameter a in equation (2) bas a dimension of concentration
(but surely is net trie real concentration of trie LPC'S), and for trie sample with T~ = 32 K it is three times higher thon for trie one 1N.ith T~
= 100 K indicating to trie growth in trie number of LPC'S with oxygen doping.
As for trie ~~Cu relaxation, both trie field dependence and non-exponential kinetics are ob- served, hence it also bas a contribution associated with the LPC'S but the spin diffusion is weak here. The hindering of spin diffusion may result from the fact that ~~CU is a 3/2-spin nucleus and thus its NIVIR fine is shifted due to quadrupolar interactions and strongly inhomogeneously
broadened (about 500 kHz at 7 T. ~3~v -S 81 MHz) owing to the lattice imperfections. On the other bond, internuclear spin coupling in trie ~~Cu system is about 3 kHz [18j which is fairly
small compared to trie inhomogeneous broadening. As a consequence, trie levels of neighbonng
nuclear spins misoverlap and this channel of energy transfer is forbidden, so that each spin relaxes direct,ly to trie LPC. In this case the relaxation kinetics of a t1N.o-dimensional nuclear spm system to the localized centers would have a form of exp[-(ÎVt)~/~] [19]. 1N.hich at small t grues tl/~ time dependence for the recovery signal. As one con see in Figure 5, this behavior is observed at the initial stage of ~~CU rela~~ation kinetics. Calculations give the expression for
trie spm 3/2 NMR signal amplitude at short times after the inversion puise in trie form:
Ait) -S 4.6~nrn Ù 1, (3)
where nrn is trie number of LPC'S per ~~Cu spin and G is defined in equation Il). Using the value of 7 -S 10 ns derived from 2"W field dependences and trie slopes of trie curves obtained from the hnear interpolation of the î T field data in Figure 5, one has nrn -S 1.9~ and 4.5$io for the samples ~vith T~ = 100 and 32 K, respectively. Actually, expression (3) was obtained in trie assumption of sniall concentrations of impurity spins and is trot valid for strict quantitative description for trie case of as high concentrations as umts of percent. This explains some dis-
agreement with trie values of nrn obtained from trie susceptibility measurements. Nevertheless it is dear that qualitatively trie result is trie sanie: trie number of LPC'S gro1N.s with oxygen
doping.
3.3. ~/ÎAGNETISM AND SUPERCONDUCTIVITY. TÎ1e àc susceptibility anà NÀÎR results men- tioned above show that oxygen doping in Tl:2201 produces localized paramagnetic centers. We would like to emphasize two important facts. First, the presence of the LPC'S is really related to the excessive oxygen since their concentration con be changea by varying trie oxygen content within one single crystal (Fig. 3). Secondly, these centers reside not in some impurity phase
different from that of Tl:2201 but are distributed over the lattice: the locally sensitive NMR technique distinguishes these two cases easily.
Consequently, trie suppression of T~ with oxygen doping in Tl:2201 may be explamed by enhancement of trie pair breaking processes caused by spin-flip scattering on paramagnetic
centers (20j. Trie estimate for finis suppression (20j may be obtained from trie expression
T~ = T~o ~h/4kB7~, (4)
where T~o is trie critical temperature of a pure sample and 1/7s is a scattering rate due to
spin-flips on magnetic impurities, .1/7~ c~ N(0)nrn, N(0) is trie density of states at trie Fermi level. Taking into account that the maximum T~ observed in Tl:2201 is 110 K (4j and assuming N(0) is.not alfected essentially by oxygen doping [îj, one estimâtes trie ratio of trie magnetic
impunty concentrations m trie Tl:2201 samples with T~ = 26 and 94 K as
NC/n©
= ~/~/7/~
= â.25, (5)
which is m agreement ~vith trie growth of the LPC concentration observed in trie susceptibility
measurements.
A well knowii consequence of trie Abrikosov-Gor'kov theory (20j is that trie superconducting
gap, A, is much stronger suppressed by trie magnetic impurities thon T~. Trie increase m trie number of trie LPC'S provides therefore a natural explanation for tire decrease m trie A/T~
ratio with lowenng T~ m Tl:2201 observed in (6j.
Dur recent NMR investigations I?ii bas allowed to separate ~~Cu spm-lattice relaxation due
to quasiparticle spm excitations, ~~W~, and that associated with trie LPC'S. At T < T~/2,
~~W~ behaves as T2 ~ that may mdicate to trie presence of gap zeroes at trie Fermi surface [22j.
Though it is usually mterpreted as an evidence m favor of trie d-wave pairmg, we beheve that these gap anomalies con occur also m trie amsotropic s-wave case due to trie preseiice of
paramagnetic centers.
According to trie results Îeported in this paper, thé paramagnetic centers produced by ex-
cessive oxygen in Tl:2201 system niay be totally responsible for suppressing T~. Trie possibility
thon electron spin correlations of antiferromagnetic (AF) type also affect superconductivitv, as