Probing the Yb$^{3+}$ spin relaxation in Y$_{0.98}$Yb$_{0.02}$Ba$_{2}$Cu$_{3}$O$_{x}$ by Electron Paramagnetic Resonance

0.38 MB·English

by  A. Maisuradze

PREPRINT(January 22, 2009)
Probing the Yb3+ spin relaxation in Y Yb Ba Cu O by Electron Paramagnetic
0.98 0.02 2 3 x
Resonance
A. Maisuradze,1 A. Shengelaya,2 B. I. Kochelaev,3 E. Pomjakushina,4,5 K. Conder,4 H. Keller,1 and K.A. Mu¨ller1
1Physik-Institut der Universit¨at Zu¨rich, Winterthurerstrasse 190, CH-8057 Zu¨rich, Switzerland
2Institute of Physics, Tbilisi State University, Chavchavadze av. 3, GE-0128 Tbilisi, Georgia
3Department of Physics, Kazan State University, Kazan, 420008, Russia
4Laboratory for Developments and Methods, Paul Scherrer Institut, CH-5232 Villigen PSI, Switzerland
5Laboratory for Neutron Scattering, ETH Zu¨rich and Paul Scherrer Institut, CH-5232 Villigen PSI, Switzerland
9
0 The relaxation of Yb3+ in YBa2Cu3Ox (6 < x < 7) was studied using Electron Paramagnetic
0 Resonance(EPR).ItwasfoundthatbothelectronicandphononicprocessescontributetotheYb3+
2
relaxation. Thephononicpartoftherelaxationhasanexponentialtemperaturedependence,which
n can be explained by a Raman process via the coupling to high-energy (∼500 K) optical phonons
a or an Orbach-like process via the excited vibronic levels of the Cu2+ ions (localized Slonczewski-
J modes). In a sample with a maximum oxygen doping x=6.98, the electronic part of the relaxation

2 follows a Korringa law in the normal state and strongly decreases below Tc. Comparison of the
2 samples with and without Zn doping proved that the superconducting gap opening is responsible
] ofofrtthheeYshba3+rprdeleacxraeatisoenoifnYtbh3e+srueplaerxcaotniodnucintinYgBsat2aCteu3foOll6o.9w8s. tIhtewsaasmsheotwemnptheraatttuhree edleepcternodneinccpeaarst
n 63Cu and 17O nuclear relaxations despite the huge difference between the corresponding electronic
o
and nuclear relaxation rates.
c
-
r PACSnumbers: 74.72.-h,76.75.+i,74.25.Dw,74.25.Ha
p
u
s I. INTRODUCTION magnetic Resonance (EPR) signal of the RE ions.12
.
t There have been a few EPR studies of the RE ions
a
with non-zero orbital moment (Er and Yb) doped in
m It is well known that in the high-T superconduc-
c cuprates.13,14,15,16,17,18,19,20 Again, the situation with
d- lteonrtYrBarae2-Ceaur3tOhx(RthEe)suiobnsst,ituhtaivoningofloyctatrliummagbnyetiiscov4af- EPR is similarly contradictory as in INS studies. Some
n groupsobservedbothelectronic(Korringa)andphononic
moments, does not change the critical temperature T
o c (Orbach) contributions to EPR relaxation.13,16,17 While
considerably.1 This makes it possible to use these lo-
c others conclude that the Korringa contribution is negli-
[ cal moments as useful paramagnetic probes of the elec-
1 tronic states within the CuO2 planes without seriously gsuibffileciaenntdttoheexipnltaeirnactthioentewmitpherlaattutirceedveipbernatdieonncseoonflythies
perturbing them. Relaxation of the RE magnetic mo-
v EPR linewidth in cuprates.18,19,20
ments provides important information about fluctuating
5
2 electric and magnetic fields in cuprate superconductors. In this work we report a detailed EPR study of the
5 In the normal state, RE ions with non-zero orbital mo- Yb3+ relaxation in YBa2Cu3Ox samples with oxygen
3 ment can interact with phonons, spin fluctuations and content x covering the whole doping range (6 < x < 7).
. charge carriers. These interactions limit the life-time of By measuring the temperature dependence of the Yb3+
1
0 the crystalfield (CF) excitations and leadto broadening EPR linewidth in a broad temperature range, it was
9 oftheobservedCFtransitions. Inelasticneutronscatter- foundthatbothphononicandelectronicmechanismscon-
0 ing(INS)iswidelyusedtostudytherelaxationoftheRE tribute to the relaxation. The electronic contribution
: magnetic moments in cuprates superconductors by mea- decreases with decreasing oxygen content x, while the
v
i suring the linewidths of the observed CF transitions.2 phononic contribution is practically doping independent
X However, the mechanism of the relaxation of RE ions in and has an exponential temperature dependence. In a
r cuprates is the issue of hot debates.3,4 Most of the au- samplewithamaximumoxygendopingx=6.98,theelec-
a
thorsconsideraninteractionof4f spinswithchargecar- tronic part of the relaxation follows a Korringa law in
rier spins (Korringa mechanism) as a dominant channel the normalstate,andasharpdropofthe relaxationrate
of relaxation.5,6,7,8 Other authors, in opposite, conclude is observed below Tc. In the superconducting state the
thattheinteractionsofREspinswithchargecarriersare electronic part of the Yb3+ relaxation rate follows the
negligible and that the interactions with lattice vibra- same temperature dependence as 63Cu and 17O nuclear
tions are only responsible for the relaxation behavior of relaxations.
the 4f spins in cuprates.9,10,11 At the moment there is This paper is organized as follows: Sample prepara-
no consensus on this subject and therefore it is impor- tion and experimental details are described in Sec. II.
tant to disentangle the electronic and phononic sources The EPRspectraandthe procedureoftheir analysisare
of relaxation. discussed in Sec. III. Sec. IV presents the temperature
Relaxationofthe4f magneticmomentscanbestudied and doping dependence of the Yb3+ relaxation rate in
also by measuring the linewidth of the Electron Para- the normal and superconducting states. In Sec. V we
2
summarize our results and the conclusions of this study.

s) a)
II. EXPERIMENTAL DETAILS nit H | c
u
b. Yb3+
weTreheprpepoalyrcerdysbtyalltihnee sstaamndpalersd osofliYd1s−tyaYtebyrBeaac2tCioun3Obyx gnal (ar T = 40 K
Si
usingY2O3,Yb2O3,BaCO3 andCuOofaminimumpu- R
rityof99.99%. Appropriateamountsofstartingreagents EP
were mixed and calcinated at temperatures 800-920oC x = 6.4
H || c
during at least 150 h in air, with several intermediate
grindings. Finally, the as-prepared sample was oxidized 150 200 250
in oxygen atmosphere (1 bar of O2) at 500oC. After the Field (mT)
oxidation the sample had an oxygen content close to 7
(6.98). The required oxygen content in the samples was
adjusted by gettering in a closed ampoule with metallic
copper (850oC, 10 h; cooling 10oC/h). The oxygen con- s) H | c b)
tent in the reduced samples was checked by comparing nit
u
theoretical and real mass changes of the oxidized getter b.
ar
and the reduced sample. For all the samples phase pu- al ( Yb3+
ritywascheckedwithaconventionalX-raydiffractometer n
g T = 25 K
(SIEMENS D500). si
R
The dilute levelofthe Ybdoping(y=0.02)waschosen P
E
in order to minimize broadening effects from Yb-Yb in- x = 6.98
teractions and at the same time to obtain a sufficiently H || c
strong EPR signal. The EPR measurements were per-
150 200 250
formed with an X-band BRUKER EMX spectrometer
Field (mT)
equipped with an Oxford Instruments helium flow cryo-
stat. In order to avoid a signal distortion due to skin FIG. 1: EPR spectra of Yb3+ in grain oriented
effects, the samples were ground and the powder was
Y0.98Yb0.02Ba2Cu3Ox with different oxygen contents: (a) x
suspendedinepoxy. Thec-axesgrain-orientationwasob-
= 6.4; (b) x = 6.98. Two orientations correspond to the ex-
tainedbyplacingthesamplesina9Tmagneticfielduntil ternal magnetic field along and perpendicular to the crystal
theepoxyhardened. Asaresultoftheorientationproce- c-axis.
dure the c-axes of the grains were preferentially aligned
along the magnetic field direction.21
EPR spectra were measured in five samples signals would be observed only for this doublet, which
Y0.98Yb0.02Ba2Cu3Ox withx=6.1,6.4,6.5,6.6,6.98and can be described in an effective S = 1/2 notation.
critical temperatures T of 0, 12(1), 51(1), 60(1), and
c Fig. 1(a) shows Yb3+ EPR spectra of
93(1) K, respectively. In addition, one Zn-doped sample
Y Yb Ba Cu O at T=40 K with the exter-
Y Yb Ba (Cu Zn ) O was measured with 0.98 0.02 2 3 6.4
0.98 0.02 2 0.97 0.03 3 6.95 nalmagneticfieldalongandperpendiculartothecrystal
T =57(1) K.
c c-axis. ThespectracorrespondtoYb3+ withaneffective
spin S=1/2 with g values g =3.13(3) and g =3.49(3).
k ⊥
The average value g =3.37 is close to g=3.43 expected
av
III. ANALYSIS OF THE YTTERBIUM EPR for the isolated Γ ground doublet.22 Fig.1(b) shows
7
SPECTRA Yb3+ EPR spectra for the optimally doped sample
x=6.98 at T=25 K. Note that the EPR line splits
In the YBa Cu O structure, the rare-earth site is into two components for the H c orientation. This
2 3 x ⊥
eight-fold coordinated by oxygens lying in the CuO bi- reflects the transition from tetragonal to orthorhombic
2
layers. Group theoretical considerations show that the crystalsymmetryinYBa2Cu3Ox withincreasingoxygen
eight-folddegeneracyoftheground-statemultiplet2F content. Enhanced noise seen in these spectra is related
7/2
oftheYb3+ ions(4f13)issplitbythecrystalelectricfield to the superconducting state. It is known that in the
oforthorhombicsymmetryintofourKramersdoublets.22 superconducting state the strong noise is generated due
Generally, this splitting is large enough so that only the to vortex motion in the modulating magnetic field used
lowestlyingdoubletisappreciablypopulatedatlowtem- in standard EPR spectrometers.
peratures. In fact, inelastic neutron scattering measure- Fig. 2 shows typical EPR spectra for the x=6.4 sam-
mentsshowedthatinYbBa Cu O thefirstexciteddou- ple at different temperatures with the magnetic field ap-
2 3 7
blet lies 1000 K above the ground doublet.23 So, EPR plied perpendicular to the c-axis. The EPR lines in the
3
g 2 region are due to Cu2+ defect centers, which are geneous and homogeneous broadening mechanisms. The
alw≈ayspresentinYBa Cu O .24 Theconcentrationof homogeneous broadening leads to Lorentzian line shape
2 3 6+x
the Cu2+ defect centers in our samples correspond to 1- while the inhomogeneous low-temperature line shape is
2% of the total copper content, as determined from the usually approximated as either Lorentzian or Gaussian.
Cu2+ EPRsignalintensity. The Yb3+ EPRspectrum at ForaLorentzian-Lorentzianconvolutionthereisasimple
40Kisrathercomplex,withadominantcentrallineand relation: ∆Br (T)=∆Bt (T) ∆B0 ,where∆B0 isa
pp pp − pp pp
shoulders on each side. The shoulders are due to par- residual,temperature-independent linewidth. The result
tially resolvedhyperfinecomponents. Naturalytterbium of a Lorentzian-Gaussian convolution is called a Voigt
has 69% even-mass isotopes with nuclear spin I=0, 14% function, which cannot be expressed in closed form.
171YbwithI=1/2,and16%173YbwithI=5/2. Bothodd In our case the low-temperature line shape is neither
isotopes give rise to nuclear hyperfine interactions. In Lorentzian nor Gaussian. The anisotropic nature of the
Fig. 2 one can see that with increasing temperature the Yb3+ signal, the partially resolved hyperfine structure,
Yb3+ EPR line broadens and the multiple-line structure and the non-ideal grain alignment, make it very difficult
of the Yb spectra gradually merges into one line which to accurately model the complex shape of the Yb EPR
continues to broaden with temperature. This broaden- spectra and its evolution with temperature. Therefore,
ing is due to the Yb relaxation and is the subject of the inordertoextractthelinewidthrelatedtorelaxation,we
present study. usedanapproachsimilartooneusedininelasticneutron
scattering studies of rare-earthrelaxation in cuprates.7
The inhomogeneous broadening of an EPR line is due
tothespreadoftheresonancefrequenciesofanassembly
of electronic spins. The width of the single spin packet
contributes to homogeneous broadening. This situation
x = 6.4, H | c can be described by a convolution integral,25
I(B)= p(B′)f(B B′)dB′ =p(B) f(B) (1)
− ∗
x10, 140K Z
where * is the convolution symbol, p(B) is the inhomo-
3+ 2+ geneous line shape and f(B) is the resonance line pro-
Yb Cu
file of the spin packet given by a Lorentzian line func-
tion. EPR spectra are usually detected in the form
of a first derivative, I′(B) = ∂I(B)/∂B. In this case

I′(B)=p(B) f′(B), where f′(B) is the first derivative
x5, 100K
∗
of a Lorentzian with amplitude A, center at B and the
0
peak-to-peak width ∆Br related with relaxation,
pp
2 −2
2 (B B ) 2 (B B )
40K f′(B)=A · − 0 3+ · − 0 . (2)
∆Bprp ·" (cid:18) ∆Bprp (cid:19) #
In order to describe the inhomogeneous line shape
p(B), we divided Yb EPR spectra in 120 points with
100 200 300 astepofδB=1mTintheintervalof120-240mT.Inthis
Field (mT) case an integralis replaced by a sum of 120 Lorentzians:
120
FIG. 2: EPR spectra in grain oriented I′(B)= p (B ) f′(B B )δB (3)
i i i
· −
Y0.98Yb0.02Ba2Cu3O6.4 at different temperatures for a i=1
X
magnetic field direction perpendicular to the crystal c-axis.
The solid lines are fits to the data, including relaxation as The residualfunction pi(Bi), characterizingthe inhomo-
described in the text. geneous line shape was obtained at low temperatures
where relaxational broadening is negligible. We found
The observed line shape can be interpreted in the fol- that in allsamples measuredin the presentwork,except
lowing way: a temperature-independent residual func- one Zn-doped sample, the relaxational broadening was
tion due to inhomogeneous broadening which is fur- absentbelow40K.Therefore,thistemperaturewasused
ther broadened by a temperature dependent relaxation. to determine the inhomogeneous line shape for samples
The method by which the relaxation-induced peak-to- without Zn doping. In Zn-doped sample the linewidth
peak derivative linewidth ∆Br is extracted from the continuesto decreasedownto 15K andthereforethe in-
pp
total peak-to-peak linewidth of the EPR signal ∆Bt homogeneous line shape was determined at this temper-
pp
depends upon the line shapes associated with inhomo- ature. As an example, the line drawn through the 40 K
4
data in Fig. 2 represents the residual function for x=6.4
sample. Havingestablishedtheresidualfunction,wekept

the coefficients p fixed at all temperatures and fitted
i 80
the data at high temperatures by convolving the resid- 20
ual function with the broadening function of Lorentzian
70 6.1 H | c
shape. The center B , width ∆Br , and amplitude A of
0 pp 6.4
the broadening function were the only variable parame- 60
tterarTsn.hsveTehrpseeearrkees-lutaolxt-ipantegiaokfintrlsianatereweTisd2−hto1hwan∆s fBionplrlpFowigiss.:22r6e.lated to the r B (mT)pp4500 666...5695 110511/ T (10 9
s
T−1 = √3gµB∆Bprp =7.62 106g∆Br (4) 30 ) -1
2 2~ × pp
20 5
For paramagnetic relaxation of RE ions in cuprates T−1
2
is equal to the spin-lattice relaxation rate T−1, as was 10
1
shown for Gd3+ in Y Gd Ba Cu O .27
0.99 0.01 2 4 8 0 0
60 80 100 120 140 160
T (K)
IV. TEMPERATURE AND DOPING
DEPENDENCE OF THE YTTERBIUM
FIG.3: (Coloronline). Temperaturedependenceofthewidth
RELAXATION
ofthebroadeningfunction∆Br andthecorrespondingspin-
pp
A. Yb3+ relaxation in the normal state lwaittthicedirffeelarxenattioonxyrgaetenTc1−on1toenftYbx3+forinHY⊥0.c98.YbT0h.0e2Bsoal2idCul3inOexs
represent the best fit to Eq. (6).
Figure 3 shows the temperature dependence of
the peak-to-peak width ∆Br related with relaxation
and corresponding relaxatiopnp rate 1/T of Yb3+ in processinvolvingopticalphononsorlocalvibrations.29,30
1
In this case the relaxation rate is
Y Yb Ba Cu O for different oxygen contents. In
0.98 0.02 2 3 x
ordertodistinguishbetweenelectronicandphononiccon- 1/T =Cexp(Ω/T)/[exp(Ω/T) 1]2, (5)
1
tributions to relaxation, let us first consider the sample −
with the lowest oxygen content (x=6.1). One can see in where Ω is the optical phonon frequency.29,30 The solid
Fig. 3 that in this sample, which is antiferromagnetic lines in Fig. 4 represent a best fit to the data using
and insulating with practically no charge carriers, the Eq.(5). It is obvious that the Raman process involving
relaxation rate is comparable to those of samples with optical phonons can explain the phonon contribution to
much higher oxygen contents. Consequently, this sug- theYb3+ spin-latticerelaxation. Opticalphononswithin
gests that the phonon contribution to the rare-earth re- the energy range 500(50) K exist in YBa Cu O . These
2 3 x
laxation rate is significant at all oxygen doping levels, are: (i) the in-plane bond-bending (500-560 K) and (ii)
since in the x=6.1 sample the electronic contribution is the outofplaneB (470K)phonons.31 Theoreticalcal-
1g
expected to be negligible. culations are necessary in order to determine which of
We observed that the temperature dependence of the thetwoopticalphononmodes(in-planebond-bendingor
relaxation rate follows closely the exponential function outofplaneB )mostlycontributestoYb3+ spin-lattice
1g
Cexp( ∆/T) with ∆=520(30) K and 570(30) K, for relaxation.
−
x=6.1 and 6.4 respectively. This is demonstrated in It is interesting to note that an exponential tempera-
Fig.4,wheretherelaxationratesasafunctionofinverse turedependence ofthe spinrelaxationratewasobserved
temperature are plotted on a semi-logarithmic scale. in a previous EPR study of crystals containing Jahn-
Such an exponential dependence is expected for the Or- Teller (JT) transition metal ions.32,33 In this case relax-
bach relaxation process via an excited intermediate en- ationtakesplaceduetoanOrbach-likeprocessviatheex-
ergy level.28 In this case ∆ corresponds to the separa- citedvibroniclevelsoftheJTion(localizedSlonczewski-
tion between the ground state doublet and the excited modes).34 In our case such a scenario is also possible if
level. According to inelastic neutron scattering experi- Yb3+ spin relaxation occurs due to coupling to the vi-
ments the firstexcitedenergylevelofYbinYBa Cu O brations of surrounding CuO complexes, since Cu2+ is
2 3 7 6
is about 1000K above the ground state doublet.23 Since a strong JT ion. There are no reports on JT splitting
thereisnoexcitedcrystalfieldenergylevelwith∆ 500 of Cu2+ in SrTiO or similar perovskites, but in MgO
3
K, the traditional Orbach relaxation mechanism c∼an be and CaO it is approximately 1500 K,33 which is much
excluded. larger than 500 K found in the present work. However,
An exponential temperature dependence of the relax- it was observedthat for Ni3+ the JT splitting in SrTiO
3
ation rate is also expected for the Raman two-phonon is2-4timesreducedcomparedtoMgO,CaOorAl O .33
2 3
5
TABLE I: The fitting parameters of the Yb3+ relaxation in
Y0.98Yb0.02Ba2Cu3Ox using Eq. (6).

100
x Tc(K) C(G) Ω(K) b(G/K)
YBCO6.4 6.1 - 10800(500) 500(30) 0
YBCO6.1 10 6.4 12(1) 13190(500) 540(30) 0
YBCO6.98 6.98 93(1) 9090(400) 460(30) 1.27(10)
1
T) /T 6.95(Zn) 57(1) 10160(400) 490(30) 1.27(10)
m 1
(1pp0 (10
r B s 9 The obtained parameters C, Ω, and b are summarized
1 ) -1 in Table I. It is interesting to compare the value of the
Korringaconstantb=1.27(10)G/Kforx=6.98withthe
values obtained from EPR measurements of Gd-doped
1 YBa2Cu3Ox. In contrast to Yb3+, the Gd3+ ion has
zero orbital moment (L=0) and interacts very weakly
with lattice vibrations. Therefore, the Korringa relax-
5 6 7 8 9 10 11 12 13 14 15
ationdue to interactionwith chargecarriersisthe domi-
1000/T(K)
nant process and the Korringaconstant can be obtained
directly from the temperature dependence of the EPR
FIG. 4: (Color online). The width of the broadening func-
linewidth. Indeed, EPR measurements on a Gd-doped
tion ∆Bprp and the corresponding spin-lattice relaxation rate EuBa Cu O single crystal37 showed a linear broad-
T1−1 of Yb3+ in Y0.98Yb0.02Ba2Cu3Ox with x=6.1, 6.4, and ening2of t3he6E.8P5R line in the temperature range 90-300
6.98versusinversetemperatureplottedonasemi-logarithmic
K with the Korringa constant b=0.5 G/K. This value of
scale. For x=6.98 sample only phononic contribution to re-
b is comparable, but smaller than our value of b. This is
laxation is plotted. The solid lines represent the best fit to
Eq. (5). expected due to the smaller oxygen content (x = 6.85)
and consequently smaller density of states at the Fermi
level N(E ) compared to our sample (x=6.98).
F
A similar reduction of the JT splitting can be expected The comparable values of Korringa constants b for
alsoforCu2+ inperovskites. Inaddition,incupratesthe Gd3+ and Yb3+ in YBCO shows that they have sim-
Cu2+ionsaresituatednexttoeachother. Insuchacoop- ilar exchange coupling Jsf with charge carriers. The
erativesituationtheenergyoftheSlonczewskimodewill observed Korringa constants of RE ions in YBCO are
be lower compared to the isolated Cu2+ ions. It would at least one order of magnitude smaller than the cor-
be interesting to search for localized vibronic modes in responding quantities found for RE ions in conventional
cuprates using inelastic neutron scattering. metals.12ThesmallvalueofbforREionsinYBa2Cu3Ox
Foranoxygencontentx>6.4,Eq.(5)cannotdescribe isduetotheweakcouplingJsf betweentheREmoments
the relaxation data well, and it was necessary to take and the holes in the CuO2 planes. This explains natu-
into account the Korringa relaxation mechanism where rally the small effect of RE magnetic moments on Tc in
localized Yb3+ moments couple to mobile charge carri- YBa2Cu3Ox.38
ers in the CuO planes throughan exchangeinteraction.
2
In normal metals the Korringa relaxation has a linear
temperature dependence bT.12 The parameter b is pro- B. Yb3+ relaxation in the superconducting state
2
portional to the product [J N(E )] , where J is the
sf F sf
exchangeintegralbetweenYb3+ momentsandholes,and Generally, it is difficult to measure EPR in the super-
N(E )isthedensityofstatesattheFermienergy. Inun- conductingstatebecauseofthestrongmicrowaveabsorp-
F
derdopedcupratesthedensityofstatesattheFermilevel tion and the noise due to vortex motion in the modulat-
istemperaturedependentduetothepseudogapopening. ing magnetic field used in standard EPR spectrometers.
This leads to the nonlinear temperature dependence of Nevertheless, in grain-aligned samples with small grain
the relaxation rate as demonstrated by 89Y NMR35 and size it was possible to observe an Yb3+ EPR signal be-
Gd EPR36 experiments in YBa2Cu3Ox. No exact for- low Tc (see Fig. 1(b)). This allowed us to study the
mula exists to describe this nonlinear temperature de- temperature dependence of the Yb3+ relaxation in the
pendence. Therefore we did not fit the data for under- superconducting state. We observed a strong reduction
doped samples (6.4 < x < 6.98). However, the use of oftheYb3+ relaxationratebelowTc. Thisisclearlyseen
theKorringalawisjustifiedinouroptimallydopedsam- in Fig. 3 for x=6.98,where the relaxationrate falls be-
ple x=6.98,where the pseudogapis absent.35 The solid low the theoreticalline givenby Eq. (6). It is naturalto
lines in Fig. 3 correspondto fits of the data using a sum attributethedropoftherelaxationtothe openingofthe
of phononic and electronic contributions: superconducting gap.
In order to check this possibility, we measured the re-
1/T =Cexp(Ω/T)/[exp(Ω/T) 1]2+bT (6) laxation of Yb3+ in Y Yb Ba (Cu Zn ) O
1 0.98 0.02 2 0.97 0.03 3 6.95
−
6

80
20
70 YBCO7
a) YBCO7:Zn 1.0
T) 60 151
r B (mpp345000 TC = 93 H K | c 101/ T (10 s) 9-1 alized) 0.8 Y1673bOC3u,+ ,, NENMPMRRR,,, HHH || | ccc,,, YYYBBBCCCOOO677. 98
20 TC = 57 K 5 m 0.6
or
10 n
(
0 0 T 0.4
1.0 T1
0.8 b) 1 / 0.2
T
el bT01.6 0.0
1/
0.4 H | c 0.4 0.6 0.8 1.0
T/T
C
0.2
YBCO7
YBCO7:Zn
0.00 20 40 60 80 100 120 140 160 FIG. 6: (Color online). (1/T1T) of Yb3+ in the su-
T (K) perconducting state normalized by its value at 100 K in
Y0.98Yb0.02Ba2Cu3O6.98 versus reduced temperature T/Tc,
FIG. 5: (Color online). (a) Temperature dependence compared with those of 17O NMR (Ref. 41) and 63Cu NMR
of the width of the broadening function ∆Bprp and (Ref. 42) in YBa2Cu3O7−δ.
the corresponding spin-lattice relaxation rate T−1
1
of Yb3+ in Y0.98Yb0.02Ba2Cu3O6.98 (YBCO7) and
Y0.98Yb0.02Ba2(Cu0.97Zn0.03)3O6.95 (YBCO7:Zn) for H⊥c. electronic part of the Yb3+ relaxation after subtracting
The solid lines represent the best fit to Eq. (6) of the
the phonon contribution in pure and Zn-doped samples.
YBCO7 and YBCO7:Zn data above their superconducting
The linear temperature dependence in the normal state
transition temperatures Tc=93 K and Tc=57 K,respectively.
and a sharp decrease below T is clearly seen.
(b) Temperature dependence of the electronic part of the c
Yb3+ relaxation 1/bTelT in YBCO7 and YBCO7:Zn. The In superconducting samples at low temperatures ex-
1
dashed line represents the normal state relaxation expected tra broadening could be present due to the distribution
by theKorringa law. of the diamagnetic shifts related to the irregular shape
of the powder grains. However, we observed no addi-
tionalbroadeningeveninoptimallydopedsuperconduct-
where Zndoping reducesT to 57K.Fig.5(a)showsthe ing sample down to 40 K. In fact, the superconducting
c
temperaturedependenceofthepeak-to-peakwidth∆Br and nonsuperconducting samples had the same inhomo-
pp
relatedwithrelaxationandcorrespondingrelaxationrate geneous linewidth at this temperature. This is in agree-
1/T in samples with and without Zn doping. It is ex- ment with the estimations of the diamagnetic shift 3
1 ∼
pected that 3% Zn doping should not change strongly mT at 10 K in H c orientation in YBa2Cu3O7, which
the phonon spectra and the electronic density of states. further decreases w⊥ith increasing temperature.36 There-
In fact, above 90 K the relaxation rates are very close fore, extra broadening from the distribution of this dia-
for both sampl∼es. However, below 90 K the behavior magnetic shift in our optimally doped sample at 40 K is
of relaxation is different. While th∼e relaxation rate of estimatedtobe about1mT,whichis muchsmallerthan
the sample without Zn doping sharply decreases below the observed inhomogeneous linewidth ∆Bpp=11 mT.
this temperature due to the onset of superconductivity, It is expected that the electronic part of the relax-
in the Zn-doped sample relaxationcontinues to decrease ation of the 4f magnetic moments on the yttrium site
graduallyuntil TZn=57 K,where a similar sharpturn is in YBCO measured by EPR has the same tempera-
c
observed. This result unambiguously shows that the su- ture dependence as 89Y nuclear relaxation,since in both
perconducting gapopening is responsible for the dropof cases the relaxation is proportional to the imaginary
Yb3+ relaxation in YBa Cu O . Moreover, it proves part of the dynamic spin susceptibility.39 This was con-
2 3 6.98
the presence of the electronic channel of relaxation de- firmed by EPR measurements of Gd3+ spin relaxation
scribed by the Korringa term bT in Eq. (6). Zn doping in YBa Cu O in normal state.36,40 Also, in the present
2 3 x
helps to reveal the Korringa term, which is masked at work, the electronic part of the Yb3+ spin relaxation in
hightemperaturesbyphononrelaxationduetothemuch optimally doped YBCO aboveT showsa linear temper-
c
strongertemperaturedependence,andbelowT byopen- ature dependence (Korringa behavior) like 89Y nuclear
c
ing of the superconducting gap. Figure 5(b) shows the relaxation.35 Note, however, that there is a huge differ-
7
encebetweenthemagnitudes(morethanafactor109)of T T , since it becomes much smaller than the inho-
c
≪
theelectronandnuclearrelaxationratesduetothelarge mogeneous residual linewidth. This prevents a reliable
differencebetweenthecorrespondingcouplingconstants. measurement of the relaxation rate below T = 0.5T . It
c
It would be interesting to compare the temperature is expected that at low temperatures in the supercon-
dependences of the electron and nuclear spin relaxation ducting state the Yb3+ relaxation rate will decrease be-
rates on the yttrium site in the superconducting state, low 107 s−1. In this case T can be measured directly
1
where relaxation drops due to the opening of the super- using the pulse EPR techniques. It would be interesting
conducting gap. Unfortunately, no detailed 89Y nuclear to perform such experiments at low temperatures.
relaxation data exists for YBa Cu O in the supercon-
2 3 x
ductingstate. Averysmallcouplingbetweentheyttrium
nuclei and the charge carriers leads to very long relax- V. SUMMARY AND CONCLUSIONS
ation times and makes 89Y NMR measurements in the
superconducting state extremely difficult. Therefore, we
To summarize, we performed a detailed study of the
plotinFig. 6theelectronicpartof1/T T versusreduced
1 temperature dependence of the Yb3+ EPR linewidth,
temperature T/T for Yb3+ in x=6.98 sample together
c i.e., the relaxation in YBa Cu O from the undoped
with the corresponding quantities from NMR measure- 2 3 x
insulating to the optimally doped superconducting re-
ments for 17O and 63Cu nuclei in YBa Cu O .41,42 The
2 3 7 gion (6 x 7). It was found that both electronic
relaxationratesareplottedforH⊥corientationand63Cu and phon≤onic≤processes contribute to Yb3+ relaxation.
NMR relaxation data is plotted for weak magnetic field
We were able to separate these processes and studied
H=0.45 T where the fluxoid core contribution to relax-
their relative contributions to relaxation as a function
ationissmall.42Theelectronandnuclearrelaxationrates
of oxygen doping. As expected, the electronic contribu-
werenormalizedtotheirvaluesaboveT atT=100K.As
c tion decreases with decreasing oxygen doping, while the
isevidentfromFig.6,(1/T T)showsaverysimilartem-
1 phonon contribution is practically doping independent.
perature dependence for the Yb3+ electronic spins and
It was found that the phononic partof relaxationhas an
the 63Cu and 17O nuclei.
exponential temperature dependence, which cannot be
Previously,the relaxationrate1/T ofYb3+ inthe su-
1 explained by a traditional mechanism involving acous-
perconductingstateofoptimallydopedYBa Cu O was
2 3 7 tic phonons. Instead, a Raman process via the coupling
extractedfrom170Yb Mo¨ssbauerspectra.43 The temper-
to high-energy ( 500 K) optical phonons or an Orbach-
ature dependence of 1/T1 was followed only up to 90 K like process via∼the excited vibronic levels of the Cu2+
limitedbythe decreaseinthe intensityofthe Mo¨ssbauer
ions (localized Slonczewski-modes) is responsible for the
effect. Therefore, a sharp decrease of the relaxationrate
phononic part of the Yb3+ relaxation in YBa Cu O .
at transition from normal to superconducting state was 2 3 6+x
In a sample with maximum oxygen doping x = 6.98,
not observed. Nevertheless, it is interesting to compare
theelectronicpartofrelaxationfollowstheKorringalaw
the absolute values of the relaxation rates of Yb3+ ob-
in the normal state, and a sharp drop of the relaxation
tained by Mo¨ssbauer and EPR techniques. Such a com-
rate was observed below T . Comparison of the EPR
parison is shown in Table II. The excellent quantitative c
linewidths in samples with and without Zn doping al-
agreement of the relaxation rates obtained by two dif-
lowed us to prove that the superconducting gap opening
ferent experimental techniques is remarkable and pro-
isresponsibleforthesharpdecreaseofYb3+relaxationin
vides strong support of the methods of extracting re-
YBa Cu O . It was shown that the electronic part of
laxation rates from Mo¨ssbauer spectroscopy43 and from 2 3 6.98
theYb3+relaxationrateinthesuperconductingstatefol-
EPR spectra presented in this work.
lows a very similar temperature dependence as the 63Cu
and the 17O nuclear relaxation rates, despite the huge
TABLE II: The relaxation rates of Yb3+ in YBa2Cu3O7 difference between the corresponding electronic and nu-
clear relaxation rates.
at different temperatures obtained by M¨ossbauer technique
(1/TMS,Ref.43)andbyEPRinthepresentwork(1/TEPR) There is an excellent quantitative agreement between
1 1
relaxation rates of Yb3+ in YBa Cu O obtained previ-
2 3 7
ouslybyMo¨ssbauerspectroscopyandinthepresentwork
T(K) 1/TMS(109 s−1) 1/TEPR(109 s−1) by EPR.This providesstrong supportof the methods of
1 1
60 0.5(1) 0.6(1) extractingrelaxationratesfromMo¨ssbauerspectroscopy
70 1.3(1) 1.2(1) and from EPR spectra. One should note however, that
80 2.5(1) 1.8(1) the EPR signal from Yb3+ in YBa Cu O can be fol-
2 3 x
90 4.0(1) 4.4(1) lowed up to at least 160 K, while Mo¨ssbauer measure-
ments in the same compound using 170Yb are limited by
90KduetothedecreaseintheintensityoftheMo¨ssbauer
Relaxation measurements at low temperatures (T effect with temperature.
T ) can provide information about the superconduc≪t- The present results demonstrate that Yb3+ can serve
c
ing gap symmetry.44 However, in the present EPR ex- as a very effective microscopic spin probe to study elec-
periments we could not detect relaxation broadening at tronic, magnetic and lattice properties of YBa Cu O .
2 3 x
8
VI. ACKNOWLEDGMENTS the NCCR programMaNEP.
ThisworkwassupportedbytheSwissNationalScience
Foundation,theSCOPESgrantNo. IB7420-110784,and
1 P.H. Hor, R.L. Meng, Y.Q. Wang, L. Gao, Z.J. Huang, J. Solid State Commun. 81, 999 (1992).
Bechtold,K.Forster,andC.W. Chu,Phys.Rev.Lett. 58, 24 J. Sichelschmidt, B. Elschner, A. Loidl, and B.I.
1891 (1987). Kochelaev, Phys. Rev.B 51, 9199 (1995).
2 J. Mesot and A.Furrer, J. Supercond.10, 623 (1997). 25 A. I. Smirnov and R. L. Belford, J. Magn. Reson. A 113,
3 A.T. Boothroyd, Phys.Rev. B64, 066501 (2001). 65 (1995).
4 S.W. Lovesey and U. Staub, Phys. Rev. B 64, 066502 26 Gh.Cristea, T.L.Bohan,andH.J.Stapleton,Phys.Rev.
(2001). B 4, 2081 (1971).
5 U. Walter, S. Fahy, A. Zettl, S.G. Louie, M.L. Cohen, P. 27 V.A. Atsarkin, V.V. Demidov, G.A. Vasneva, T. Feher,
Tejedor, and A.M. Stacy,Phys.Rev.B 36, 8899 (1987). A. J´anossy, and B. Dabrowski, Phys. Rev. B 61, R14944
6 G. Amoretti, R. Caciuffo, P. Santini, O. Francescan- (2000).
geli, E.A. Goremychkin, R. Osborn, G. Calestani, M. 28 S.A. Dodds, J. Sanny, and R. Orbach, Phys. Rev. B 18,
Sparpaglione, and L.Bonoldi, Physica C221, 227 (1994). 1016 (1978).
7 A. Mukherjee, A.T. Boothroyd, D.MK. Paul, M. P. Srid- 29 B.I. Kochelaev, JETP 10, 171 (1960).
har Kumar, and M. A. Adams, Phys. Rev. B 49, 13089 30 Chao-Yuan Huang, Phys.Rev.154, 215 (1967).
(1994); A.T. Boothroyd, A. Mukherjee, and A.P. Murani, 31 L. Pintschovius, W. Reichardt, M. Klaser, T. Wolf, and
Phys. Rev.Lett. 77, 1600 (1996). H.V. Lohneysen, Phys. Rev. Lett. 89, 037001 (2002); M.
8 J.Mesot,G.B¨ottger,H.Mutka,andA.Furrer,Europhys. Opel, R. Hackl, T.P. Devereaux, A. Virosztek, A. Zawad-
Lett. 44, 498 (1998). owski, A. Erb, E. Walker, H. Berger, and L. Forro, Phys.
9 U. Staub, M. Gutmann, F. Fauth, and W. Kagunya, J. Rev.B 60, 9836 (1999).
Phys.: Condens. Matter 11, L59 (1999). 32 U. H¨ochli and K.A. Mu¨ller, Phys. Rev. Lett. 12, 730
10 S.W.LoveseyandU.Staub,Phys.Rev.B61,9130(2000). (1964).
11 M.Roepke,E.Holland-Moritz,B.Bu¨chner,H.Berg, R.E. 33 U.H¨ochli,K.A.Mu¨ller, andP.Wysling, Phys.Lett. 15, 1
Lechner, S. Longeville, J. Fitter, R. Kahn, G. Coddens, (1965).
and M. Ferrand,Phys. Rev.B 60, 9793 (1999). 34 K.A. Mu¨ller, in Magnetic Resonance and Relaxation,
12 S.E. Barnes, Adv.Phys. 30, 801 (1981). edited by R. Blinc (North-Holland Publishing Company,
13 L.Kan,S.Elschner,andB.Elschner,SolidStateCommun. 1967), pp. 192-208.
79, 61 (1991). 35 H. Alloul, T. Ohno, and P. Mendels, Phys. Rev. Lett. 63,
14 M.V.Eremin,I.N.Kurkin,M.P.Rodionova,I.H.Salikhov, 1700 (1989).
and L.R. Tagirov, Supercond. Phys. Chem. Techn. (Rus- 36 A. J´anossy, L.C. Brunel, and J.R. Cooper, Phys. Rev. B
sia) 4, 716 (1991). 54, 10186 (1996).
15 I.N.Kurkin,I.Kh.Salikhov,L.L.Sedov,M.A.Teplov,and 37 D. Shaltiel, C. Noble, J. Pilbrow, D. Hutton, and E.
R.Sh.Zhdanov,JETP 76, 657 (1993). Walker, Phys.Rev.B 53, 12430 (1996).
16 H. Shimizu, K. Fujiwara, and K. Hatada, Physica C 288, 38 N.E. Alekseevskii, A.V. Mitin, V.I. Nizhankovskii, I.A.
190 (1997). Garifullin, N.N. Garifyanov, G.G. Khaliullin, E.P. Khly-
17 H. Shimizu, K. Fujiwara, and K. Hatada, Physica C 299, bov,B.I.Kochelaev,andL.R.Tagirov,J.LowTemp.Phys.
169 (1998). 77, 87 (1989).
18 V.A.Ivanshin,M.R.Gafurov,I.N.Kurkin,S.P.Kurzin,A. 39 B.I. Kochelaev andG.B. Teitelbaum, in Superconductivity
Shengelaya, H. Keller, and M. Gutmann, Physica C 307, in Complex Systems, (Springer, 2005), pp.205-266.
61 (1998). 40 V.A. Atsarkin, V.V. Demidov, and G.A. Vasneva, Phys.
19 L.K. Aminov, V.A. Ivanshin, I.N. Kurkin, M.R. Gafurov, Rev.B 52, 1290 (1995).
I.Kh. Salikhov, H. Keller, and M. Gutmann, Physica C 41 J. A. Martindale, P.C. Hammel, W. L. Hults, and J. L.
349, 30 (2001). Smith,Phys. Rev.B 57, 11769 (1998).
20 M.R.Gafurov,L.K.Aminov,I.N.Kurkin,andV.V.Izotov, 42 J.A.Martindale,S.E.Barrett,C.A.Klug,K.E.O’Hara,
Supercond.Sci. Technol. 18, 352 (2005). S. M. DeSoto, C. P. Slichter, T. A. Friedman, and D. M.
21 D.E. Farrell, B.S. Chandrasekhar, M.R. DeGuire, M.M. Ginsberg, Phys.Rev. Lett. 68, 702 (1992).
Fang,V.G.Kogan,J.R.Clem,andD.K.Finnemore,Phys. 43 J. A. Hodges, P. Bonville, P. Imbert, and G. J´ehanno,
Rev.B 36, 4025 (1987). Physica C 184, 259 (1991).
22 A.AbragamandB.Bleaney,ElectronParamagneticReso- 44 For a review see, K. Asayama, Y. Kitaoka, G.Q. Zheng,
nanceofTransition Ions(Clarendon Press, Oxford,1970). and K. Ishida, Prog. Nucl. Magn. Reson. Spectrosc. 87,
23 M. Guillaume, P.Allenspach, J. Mesot, U. Staub,A.Fur- 221 (1996).
rer,R.Osborn,A.D.Taylor,F.Stucki,andP.Untern¨ahrer,