Doping Dependence of the Pseudogap in La(2-x)Sr(x)CuO(4)

7 Pages·0.3 MB·English

by  J. G. Naeini

Doping Dependence of the Pseudogap in La Sr CuO
2−x x 4
J. G. Naeini, X. K. Chen, and J. C. Irwin
Department of Physics, Simon Fraser University, Burnaby, British Columbia, V5A 1S6, Canada
M. Okuya,∗ T. Kimura,† and K. Kishio
9
Department of Superconductivity, University of Tokyo, Bunkyo-ku, Tokyo 113, Japan
9
(Phys. Rev. B59, 1 April1999)
9
1
We report the results of Raman scattering experiments on single crystals of La2−xSrxCuO4 that

n spantherangefromunderdoped(x=0.10)tooverdoped(x=0.22). Thespectraareconsistentwith
a theexistenceofastronganisotropicquasiparticleinteractionthatresultsinanormalstatedepletion
J of spectral weight from regions of the Fermi surface located near the zone axes. The strength of

1 the interaction decreases rapidly with increasing hole concentration and the spectral evidence for
1 the pseudogap vanishes when the optimum doping level is reached. The results suggest that the

pseudogap and superconductinggap arise from different mechanisms.
]
n
PACS numbers: 74.25.Gz, 74.72.Dn, 74.25-q, 78.30.Er
o
c
- I. INTRODUCTION tions in Bi2Sr2CaCu2Oz (Bi2212). Finally the hole con-
r centration is determined simply by the Sr concentration
p
u It has been recognized for some time1,2 that the nor- if oxygen stoichiometry is maintained. As a result, one
s mal state electronic properties of the high temperature can obtain23 high quality and well characterized single
t. superconductors (HTS) are very different from those of crystals of La214 that enable one to study the system-
a a conventional Fermi liquid. It is also generally agreed aticevolutionoftheelectronicpropertiesthroughoutthe
m
that an understanding of the normal state properties is complete doping range. We have measured the relative
d- importantinidentifying the physicalmechanismrespon- strengthsandfrequencydistributionsoftheB1g andB2g
n sible for superconductivity in HTS. As a result an in- Raman continua as a function of temperature and dop-
o creasing amount of attention has focused on the tem- ing. Theresults ofthese experiments areconsistentwith
c perature region T>Tc and recent studies3–17 of under- a PG mechanism that is intrinsic to a single CuO2 layer
[ doped compounds have revealed remarkable deviations and is present only in the underdoped regime of La214.

2 from Fermi liquid behavior. These results4–17 have been OurresultsalsoimplythattheenergyscaleEg associated
v interpreted in terms of the opening of a normal state with the PG is quite different from the superconducting
2 pseudogap (PG), a term that is generally used to mean gap energy ∆.
6 a large suppression of low energy spectral weight.
2
Despite an extensive experimental effort the physical
4 origin of the PG remains undetermined. Some of the II. EXPERIMENT
0
8 proposed mechanisms involve precursor pairing3,18,19 at
9 T>Tc . For example Emery and Kivelson18 have sug- Thesamplesstudiedinthisworkweregrownbyatrav-
/ gested that incoherent pairs form above T , with phase elling floating-zone method23. The crystals used here
t c
a coherence and hence superconductivity occurring at T . were cut from larger single crystals and typically had
c
m Or if one has separation of charge and spin19 d-wave dimensions of about 2×2×0.5 mm3. Characterization
- pairing of spinons can take place at T>Tc, with charge ofthesamplesincludedsusceptibilityandtransportmea-
d condensation at T . On the other hand in the nearly surements of their critical temperatures23. The sample
c
n
antiferromagnetic Fermi liquid (NAFL) model proposed surfaces were prepared for light scattering experiments
o by Pines and coworkers20–22, the PG and superconduct- by polishing with diamond paste and etching24 with a
c
: ing mechanisms are competing. In this model the PG bromine-ethanol solution. Finally the samples were ori-
v
or spectral weight depletion (SWD) is caused by strong entedinthebasalplaneusingLauex-raydiffractionpat-
i
X anisotropic antiferromagnetic (AFM) interactions that terns. The physicalparameterscharacterizingthe La214
r are peaked near the AFM ordering vector Q= (π,π). samples are summarized in Table I.
a To obtain additional, complementary information on The Ramanspectra were obtained in a quasibackscat-
the nature and extent of the PG regime we have car- tering geometry using the 488.0 or 514.5 nm lines of an
ried out electronic Raman scattering experiments on Ar+ laser. In most cases the light was focussed onto the
La Sr CuO [La214(x)]. This compound is an excel- sample using a cylindrical lens, but on occasion a com-
2−x x 4
lent material for these studies for several reasons. It has binationofa sphericalandcylindricallenses wasusedto
a relativelysimple structure with a single CuO plane in reduce the dimensions of the laser beam at the sample.
2
the primitive cell. Thus the resultsare notinfluenced by In all cases the incident power level was controlled to be
structuralcomplicationssuchas those introducedby the lessthan10W/cm2. Withthesepowerlevelsanestimate
chains in YBa Cu O (Y123) or the structural modula- of about 11K has been obtained for the amount of local
2 3 y
1
La2−xSrxCuO4 Nominal Srcontent (x) Tc (◦K) (a) La1.87Sr0.13CuO4
Underdoped 0.10 27
300K
Underdoped 0.13 35 250K
Optimally doped 0.17 37 _ 200K
150K
Overdoped 0.19 32 _
_ 100K
Overdoped 0.22 30
_ 50K
)
s
TABLEI. Thephysicalparameterscharacterizingthesam- nit _ X'Y' (B1g)
u
panledsxstiusdtihedeninomthinisalphaopleer.coTnccewntarsatdioetne2r3m.ined magnetically arb. (b) 300K
(
χχ'' 250K
_ 200K
heating by the laser. This estimation was obtained from
_ 150K
measurementsofthe powerdependence ofthe spectra at
_ 100K
temperatures near T , and the Stokes/anti-Stokes ratio
c _ 50K
atsomewhathighertemperatures. The temperaturesre- _
XY (B )
2g
ported in this paper are the ambient value plus 11K, or
0 200 400 600 800 1000 1200
the estimated temperature in the excited region of the
RAMAN FREQUENCY (cm-1)
sample.
In a Raman spectrum the intensity of the scattered FIG. 1. The B1g (a) and B2g (b) response functions of
lightisproportionaltothesquareoftheRamantensorγ, theunderdopedLa214(0.13)crystalfortemperaturesbetween
whosecomponentscanbeselectedbyappropriatechoices 50K and 300K. The spectra are offset vertically for clarity.
of the incident and scattered polarizations25. In this pa-
perspectrawereobtainedinthez(xy)z¯andz(x′y′)z¯scat-
ture,towithin the accuracyofourmeasurements(15%).
teringgeometrieswherethelettersinsidethebracketsde-
From figures 2a and 2b, it is evident that the B spec-
1g
notethepolarizationsoftheincidentandscatteredlight.
tra obtained for each sample, above and below T , are
c
Herethexandy axesareparalleltotheCu-Obondsand
essentially identical and are thus unaffected by the su-
x′ and y′ axes are rotated by 45◦ with respect to (x,y).
perconducting (SC) transition. In direct contrast, the
The xy component of the Raman tensor γ transforms
xy low energy B spectra of both underdoped compounds
2g
as k k or the B irreducible representation of the D
x y 2g 4h (Figs. 2c and 2d) undergo a superconductivity induced
point group, and γx′y′ as (k2x−k2y) or B1g. Thus γ(B1g) renormalization(SCIR) as the temperatures of the sam-
must vanish along the diagonal directions (±1,±1) in k-
ples are lowered through T . The observed immunity of
c
space and γ(B ) must vanish along the k and k axes.
2g x y the B spectra to passage into the SC state, and con-
1g
These two scattering geometries allow us to probe com-
trasting SCIR in the B channel, is analogous to the
plementaryregionsoftheFS25 sincetheB channelwill 2g
1g behavior observed in underdoped Y12313.
sampleregionsoftheFSlocatednearthek andk axes
x y
whiletheB channelwillproberegionslocatednearthe
2g
diagonal directions. (a) X'Y' (B1g) XY (B2g) (c)
40 K
Throughoutthispaperwedealwiththefrequencydis- 40 K
tributionoftheRamanresponsefunctionχ′′(ω,T)which
is obtained by dividing the measured intensity by the
15 K _ 15 K
thermal factor [1–exp(-h¯ω/k T)]−1. _
B s)
nit
u La1.87Sr0.13CuO4
III. RESULTS arb. (b) X'Y' (B1g) XY (B2g) (d)
χχ'' ( 35 K
AdetaileddescriptionofRamanspectraofLa214(0.17)
sample,whichwewillconsidertoberepresentativeofthe
optimally doped state, has beenpublished recently25. In 15 K
35 K _
this paperwe will focus onthe changesthat occur inthe
_
spectra of the underdoped and overdoped compounds. 15 K La Sr CuO
1.9 0.1 4
We will first discuss spectra obtained from the under-
0 100 200 300 400 5000 100 200 300 400 500
doped samples (x = 0.10 and 0.13). The normal state RAMAN FREQUENCY (cm-1)
response functions are shown in Fig. 1 for the 0.13 sam-
ple and the low temperature response functions for both FIG. 2. The B1g response above and below Tc for the
La214(0.13) crystal (a) and theLa214(0.10) crystal (b). The
samples are presented in Fig. 2. As is shown in Fig.
1a,themagnitudeofχ′′(B )isindependentoftempera- B2g response for the same crystals are shown in (c) and (d).
1g
2
As a measure of the strength of the response func-
(a) 15K (b) 15K (c) 15K (d) 25K (e) 15K
tions we will use the integral of χ′′(ω) over the low fre-
La214(x) B1g quency spectral region. In this context the strength
) B1g B1g of χ′′(B2g) is approximately the same (Fig. 3) for all
ts the crystals studied here, while the strength of χ′′(B )
ni 1g
u changesdramaticallyasweproceedfromtheunderdoped
rb. totheoverdopedsamples. TheratioR=R dωχ′′(B1g)/
χχ'' (a x=0.10 x=0.13 x=0.17 x=0.19 x=0.22 Rmadnωrχe′s′(pBo2ngs)e,,ainncdretahseersefboyreatfhaectsotrreonfgatbhoouftt1h6eaBs1wgeRgao-
B2g B1g B2g B2g B2g B2g from the underdoped x = 0.10 sample to the optimally
doped sample with x = 0.17. The ratio R continues to
B1g increase, but at a slower rate as one proceeds into the
0 150 300 0 150 300 0 150 3000 150 300 0 150 300 overdoped regime. This result is summarized quantita-
RAMAN FREQUENCY (cm-1) tively in Fig. 4, where we have plotted the ratio R as a
functionofholeconcentration. Forcomparisonpurposes,
FIG.3. Direct Comparison of the low energy B1g and B2g
Raman response functions measured at temperatures below data obtained previously13 on Y123 are also included28
Tc in La214(x) for (a) x = 0.10; (b) x = 0.13; (c) x = 0.17; in Fig 4. It is evident that the ratio R for the two com-
(d) x = 0.19; and (e) x = 0.22. The scale of χ′′ is the same pounds exhibits a similar trend with doping.
for all fiveframes.
IV. DISCUSSION
We have also obtained spectra from two overdoped
crystals (x = 0.19 and 0.22) and we have found that
A. Spectral Weight Depletion
the magnitudeofthe responsefunctions χ′′(B )inboth
1g
samples (for ω ≥ 300cm−1) are again approximately
independent26,27 oftemperature. Incontrastto the situ- Relatively few Raman experiments have been carried
out to investigate the nature of the PG. Nemetschek et
ation in the underdoped samples, there is a pronounced
SCIR that occurs at T in both the B and B chan- al.14 have attributed a small loss of spectral weight at
c 1g 2g
nels for both the optimally doped25 and the overdoped ω ≤700cm−1intheB2g channeltothepresenceofaPG.
samples26,27. Finally the low temperature spectra of the Blumberg et al.15, in spectra obtained from underdoped
Bi2212haveobserveda sharpfeature atabout 600cm−1
overdoped crystals, in both B and B channels, are
1g 2g
which they have attributed to a bound state associated
compared to the underdoped and optimally doped spec-
with precursor pair formation above T . In our studies
tra in Fig. 3. c
of underdoped Y123 and La214 we have focused on the
20 10 0 φφ (degrees) strikinglossofspectralweightintheB1gchannel,andthe
contrasting constancy of the B spectra, features which
2g
La SrCuO appearto be common13,29,30 to the hole doped cuprates.
2-x x 4
5 Toillustratethedramaticchangeintheintensityinthe
YBa Cu O
) g 2 3 y B1g channel and the relative constancy of the B2g chan-
2
B 4 nel, we can calculate the unscreened Raman response
χχ'' ( functions inthe superconducting state using the conven-
) / g 3 tional model of light scattering31
1
B
χχ'' ( 2 χ′γ′(ω)∝ hωpγ2ω(2k−)∆42∆(k2)(k)i, (1)
1
where h···i imply an average over the FS for k-vectors
such that ω > 2|∆(k)|. The results shown in Fig. 5
0 wereobtainedwithγ(k)calculatedusingasecondneigh-
0.10 0.15 0.20
bor tight binding model with the parameters described
Hole Concentration (x)
previously25 and ∆(k) = ∆ [cos(k a)−cos(k a)]. The
0 x y
FIG. 4. The ratio R of the integrated Raman response response functions obtained by integrating (1) over the
functions plotted as a function of hole concentration for full FS (dashed lines in the inset of Fig. 5) are shown
La214 (•) and Y123 (◦). The ratio was determined for in Fig. 5a. To obtain a representation of the Raman
0 ≤ ω ≤ 600cm−1 in all cases. The dashed line represents response in the underdoped case we integrate (1) over
the calculated R using equation (1) for different values of φ portions of the FS lying between φ and 90◦−φ in each
(Fig. 5) superimposed using the top scale. The dotted line
quadrant. That is, regions of the FS near the axis (de-
serves only as a guide tothe eye.
finedbyangleφ)wereexcludedfromtheintegration,and
3
(a) φ = 0o (b) φ = 10o ky perature(forT> Tc),whichisdifficulttoreconcilewith
the FS contraction envisaged by Norman et al.11.
According to Fig. 4 the SWD or pseudogap has ap-
φφ kx proximately the same strength in La214 and Y123. This
s) isnotconsistentwiththesuggestion32thatthePGarises
t
ni fromstrongAFMinteractionsbetweenthecloselyspaced
u CuO layers of bilayer compounds such as Y123. The
2
.
b present results thus suggest that the PG arises from in-
r (c) φ = 20o (d) φ =30o
a tralayer interactions and in this regardare in agreement
(
" with IR4,5, ARPES7, specific heat33, and NMR34 mea-
χ
surements.
_ _
B. Magnitude (Eg) of the Pseudogap
0 1 0 1 0 1 0 1 2
In the spectra we have obtained from La214 there is
(ω/2∆ )
max no clear indication of an onset temperature T∗ for the
PG. Estimates21,34 of T∗ ≤ 200K have been obtained
FIG. 5. Calculated B1g (thin lines) and B2g (thick lines) from the analysis of NMR experiments carried out on
Raman response functions in La214 (a) by integrating over
slightly underdoped La214 samples (x = 0.14 and 0.15).
the full Fermi surface (dashed lines in the inset) and (b-d)
Onthe otherhandtransport35, IRreflectivity5,andspe-
by integrating only over region lying between φ and 90◦−φ
cific heat33 measurementsonunderdopedLa214haveall
in each quadrant. The inset shows polar plots of the angu-
foundT∗ >300K.Ourspectra(Fig. 1)indicatesomemi-
lar dependence of the B1g (thin lines) and B2g (thick lines)
norchangesatabout200K,thatinclude the appearance
components of the Raman tensor γ(k).
of a broad and weak feature at about 600cm−1 in both
channels. In fact the variations that occur in the B
2g
the resulting response functions are shown in Figs. 5b- spectra below about 200Kreflect a decrease in intensity
5d for φ=10◦, 20◦, and 30◦. As is evident an increasing atlowenergiessimilartothatobservedbyNemetscheket
truncation of the FS leads to a rapid depletion of the al.14 in Bi2212. However,no significant changes in spec-
B spectrum, but has much smaller effect on the B tralweight,whichonemightexpecttomarktheonsetof
1g 2g
spectrum. the PG, occur at any temperature below 300K.
We have also calculated the ratio R of the integrated ThelackofanexperimentalconsensusforT∗ inLa214
response functions as a function of φ and the results are is perhaps supportive of the suggestion36 that one can-
superimposed(dashed line) on the data points in Fig. 4. notassociateawelldefinedT∗ with the PG,butonlyan
For the purpose of this comparison both the measured energyscale Eg. In this regard,however,there is alsono
and calculated response functions were integrated from clear evidence of a gap edge in the B1g spectra we have
ω =0to600cm−1. Itshouldbenotedthatthe apparent obtained (Fig. 1) from underdoped La214. This is per-
agreement between calculation and experiment is quite haps not surprising if Eg is of order of J in underdoped
fortuitous given the simplistic nature of the model. The samples and decreases with increasing hole concentra-
results do, however, provide plausibility to the existence tionashasbeensuggestedbyphotoemission7andspecific
of localized depletion of spectral weight from regions of heat33 measurements. This magnitude for the PGmight
the FS located near (π,0). Finally we should note that be expectedifshortrangeAFMcorrelationwererespon-
the effective gapping of quasiparticles (QP) from these sible for the spectral weight depletion20. One possible
sameregion,atT>T ,isconsistentwiththeabsenceof scenario involves the existence of AFM ordered regions
c
a superconductivity induced renormalization in the B andseparateholerichregionsintheunderdopedsamples.
1g
spectraoftheunderdopedcompounds(Figs. 2aand2b). Evidence for this type of phase separation was obtained
The observeddependence onscatteringgeometrythus recently in the muon spin resonance experiments on un-
implies that the spectral weight loss is confined to re- derdoped La214 and Ca doped Y12337.
gions of the FS located near (±π,0) and (0,±π) and In Raman spectra one-magnonexcitations are not ob-
that QP located near the diagonals are essentially unaf- servedandhencethePGenergywouldbereflectedinthe
fectedbyareductionindopinglevel. GiventhatARPES two-magnonpeakthatoccursatabout3J≈3000cm−1in
experiments8–10 also find a SWD that reaches a max- undopedLa21438,39andinsulatingY12338,40,41. Theam-
imum in regions near (π,0), we will assume that this plitudeandenergyofthetwo-magnonpeakdecreases39,42
featureis characteristicofthe PG,andthatthe PGthus as the compounds are doped and such peaks have not
manifestsitselfintheRamanspectraprimarilyasaSWD been observed for doping levels greater than or equal to
in the B channel. We should note however, that the optimum. Although our spectrometer range is limited,
1g
B spectra is also approximately independent of tem- we have carried out preliminary measurements of the
2g
4
(a) La Sr CuO 6c) suggests pcr > popt. We would thus conclude that
1.78 0.22 4 the pseudogap “fills” before it “closes”.
It is interesting to compare the PG energy scale E to
g
100K that of the superconducting gap ∆. In Raman spectra
_
the energy of the pair-breaking peak in the B channel
1g
X'Y'(B1g) 50K can be used43 as a measure of ∆. In optimally doped
its) (b) La1.87Sr0.13CuO4 Lisag2i1v4enthbeymωa(xBim1gu)m≈v2a∆lumeaoxf t≈he20su0pcemrc−o1n,dausctisinaglsgoapev25-
n
ident from Fig. 3. In the overdoped region ∆ de-
u max
. 100K creases (Fig. 3), consistent with the behavior in other
rb _ cuprates44–46. In the underdoped regime the absence of
a X'Y'(B ) 50K
( 1g SCIR inthe B channelmeans that we do not haveany
χχ'' direct measure1gof the gap maxima. We can note how-
(c) La Sr CuO
1.9 0.1 4 ever that the frequency of the pair-breaking peak in the

B channel remains approximately constant [ω(B ) ≈
2g 2g
130±20cm−1]asthedopinglevelisdecreasedfromopti-
100K
_ mum. We thus infer that ∆ remains approximately con-
X'Y'(B1g) 50K stantthroughoutthedopingrange0.10≤x≤0.17. This
behavior is also consistent with that found in ARPES
0 500 1000 1500 2000 2500 3000
and tunneling measurements9,47 on Bi2212. Therefore
Raman Frequency (cm-1) we have an amplitude ∆ ≈100cm−1 for the SC gap
max
and E ∼J∼700cm−1 for the PG. These very different
FIG. 6. The high energy B1g Raman response functions g
energy scales suggest that different physical mechanisms
measuredat50Kand100KinLa214(x) for(a)x=0.22, (b)
are involved.
x = 0.13, and (c) x = 0.10.
highenergyB spectraforthreesamples(x=0.10,0.13, C. Summary
1g
and 0.22). The high energy (ω < 3000cm−1) response
functions measuredat50Kand100Kfor the three sam- The most noticeable aspect of the spectra shown in
ples are shown in Fig. 6. From the figure one can see Fig. 3 is the SWD that occurs in the B channel when
1g
a broad, weak peak at about 2000cm−1 (J∼ 700cm−1) the doping level is reduced below optimum, while the
in the x = 0.10 spectrum. This peak essentially vanishes strengthof the B spectrum is essentially unaffected by
2g
as we proceed to higher doping levels (Figs. 6a and 6b). changes in doping. These results suggest that the QP
Wemustemphasizethatthesespectrahavenotbeencor- scattering rate is highly anisotropic48 and leads to the
rected for variations in the optical constants or for spec- existence of “hot spots” on regions of the FS located
trometer response and are shown only to illustrate the near (±π,0) and (0,±π). This observation is consis-
effect of doping. Our spectrometer response decreasesin tent with the predictions of Pines and coworkers20 who
the redandcorrectionmightpush the peak inFig. 6c to have pointed out that QP located near these regions
higher energies and also decrease its relative amplitude. can be coupled by the AFM ordering vector Q = (π,π)
Such corrections would not however alter the observed and hence interact strongly in underdoped compounds.
trend that is induced by changes in doping. Conversely, the QP on regions of the FS located near
It appears plausible that the pseudogap or SWD in |k | = |k | are weakly coupled20, and have been des-
x y
the B1g channel is due to short range AFM correlations ignated as “cold”20,49. As the doping level is increased
and the formation of “hot spots”20 on regions of the FS toward optimum the SWD decreases rapidly, consistent
located near (±π,0) and (0,±π). As the doping level withaweakeningoftheAFMinteractionsbetweenQP20.
is increased the AFM interactions become weaker, as InthislanguageourresultsthensuggestthatthehotQP
is evidenced by a weakening (Fig. 6) and shift of the aregappedaboveT andhencearenotaffectedbytheSC
c
two-magnon peak to lower energies39,42. Thus as the transition. Only the cold QP, as evidenced by the SCIR
doping level is increased, Eg decreases and some spec- in the B2g channelat Tc, participate in the transitionto
tral weight is shifted from higher to lower energies. At the SC state.
optimum doping, where a SCIR is observed in the B1g Thus far our attention has been focused on the most
channel (Fig. 3c), the PG is assumed to be filled. Fur- dominant feature of the spectra obtained from under-
thermore on the basis of their specific heat measure- doped compounds, namely the SWD that is confined to
ments in several cuprates Loram et al.33 have proposed the B channel. As mentioned previously Nemetschek
1g
that Eg ≈J(1− p/pcr) where p is the hole concentra- et al.14 found that a relatively small depletion of spec-
tion and pcr the value at which Eg “closes”. Given that tral weight also occurs in the B2g channel of Y123 and
J≈ 1000cm−1 in La2CuO4 and Eg ≈ 700cm−1 (Fig. Bi2212 below ω ≈ 700cm−1 and T≈ 250K. This en-
5
ergy and onset temperature are in good agreement with conductinggaparisefromdifferentphysicalmechanisms.
IR measurements on the same compounds4. This ap- Since superconductivity grows while the spectral weight
pearsreasonablesince,giventheSWDintheB channel loss or pseudogap diminishes, it appears that the two
1g
one would expect the results of IR and transport mea- mechanisms compete with each other for the available
surements, for example, to be determined mainly by the quasiparticles.
properties of the cold QP14,20,49, or those on regions of
the FS located near the diagonal directions. Further-
more, one can see that a similar small depression of VI. ACKNOWLEDGEMENTS
spectral weight appears below 700cm−1 and 200K in
the B2g spectra of the x = 0.13 crystal (Fig. 1b), in The authors gratefully acknowledge useful conversa-
agreementwith IRmeasurementscarriedout5 onunder- tions with T. P. Devereaux, T. Startseva, T. Timusk,
doped La214. Thus the loss of spectral weight in the and J. Schmalian. One of us (JCI) has also benefited
B2g channel14, the depression of the scattering rate ob- greatlyfromdiscussionswithR.G.Buckley,J.L.Tallon,
servedin IR experiments, and the spectral weight deple- A. J. Trodhal, and G. V. M. Williams during a visit to
tion observed in the B1g channel (Figs. 3 and 4) are all the Industrial Research Laboratory in Lower Hutt, New
characterized by the same energy scale. This suggests Zealand. The financial support of the Natural Sciences
that there is a common physical mechanism underlying andEngineeringResearchCouncilofCanadaisgratefully
the apparently different phenomena that are observedin acknowledged.
the two channels. However, the identification of a pos-
sible interrelation between the two channels will require
a very carefulanalysis of temperature dependence of the
spectrafor50K≤T≤300K.Suchananalysisshouldalso
provide important information about the existence and
magnitude of crossover temperatures T∗ that have been
observedinotherexperiments5,21,34,35andareimportant 1P. B. Allen, Z. Fisk, and A. Miglieri, Physical Properties
features of some models18,20. It is also possible that the of HTS, V. I, edited by D. M. Ginsberg (World Scientific,
SWD in the two channels arises from different mecha- Singapore, 1989).
nisms and that there are different pseudogaps as have 2I.Bozovic, D.Kirillov, A.Kapitulnik,K. Char, M. R.Ha-
been suggested7 by Ino et al. han,M.R.Beasley,T.H.Geballe,Y.H.Kim,A.J.Heeger,
Phys.Rev.Lett. 59, 2219 (1987).
3M.Randeria,inProc.ofInternat.SchoolofPhys.,Verenna,
V. CONCLUSIONS 1997,editedbyG.IadonisiandJ.R.Schrieffer(IOSPress,
Amsterdam, 1998); cond-mat/9710223.
4A. V. Puchkov, D. N. Basov, and T. Timusk, J. Phys.
In the Raman spectra obtained from La214 there is a
Condensed. Matter 8, 10049 (1996); A. V. Puchkov, P.
loss of low energy spectral weight in the B channel as
1g Fournier, T. Timusk, and N. N. Kolesnikov, Phys. Rev.
the doping level is decreased below optimum, while the
Lett. 77, 1853 (1996).
intensity in the B2g channel remains unchanged. These 5T. Startseva, T. Timusk, A. V. Puchkov, D. N. Basov,
results imply the existence of a strong anisotropic scat-
H. A. Mook, T. Kimura, and K. Kishio, Phys. Rev. B 59
tering mechanism and the formation of hot spots20 on
March 1 (1999); T. Timusk and B. Statt to be published
regionsofthe Fermisurfacelocatedneark andk axes.
x y in Reports Progress in Physics.
Furthermore, in underdoped compounds the B1g chan- 6C. P. Slichter, R. L. Corey, N. J. Curro, S. M. DeSoto,
nel is unaffected by the superconducting transition. The K. O’Hara, T. Imai, A. M. Kini, H. H. Wang, U. Geiser,
pseudogapisthusmanifestedinRamanspectrabothbya J. M. Williams, K. Yoshimura, M. Katoh, and K. Kosuge,
depletionoflowenergyspectralweightorreducedinten- Philosophical Magazine B 74, 545-561 (1996).
sity and by the absence of a superconductivity induced 7A.Fujimori, A.Ino, T. Mizokawa, C. Kim, Z.-X. Shen,T.
renormalization in the B channel. In this context the Sasagawa,T.Kimura,K.Kishio,M.Takaba,K.Tamasaku,
1g
pseudogap in La214 and Y12313 is present only in the H.Eisaki,andS.Uchida,J.Phys.Chem.ofSolids59,1892
underdoped regime. Since the spectral weight depletion (1998);A.Ino,T.Mizokawa,K.Kobayashi,A.Fujimori,T.
is similar in the two compounds, we conclude that the Sasagawa, T. Kimura, K.Kishio, K.Tamasaku, H.Eisaki,
pseudogap is intrinsic to a single CuO plane and does and S. Uchida, Phys. Rev. Lett. 79, 2101 (1997); Phys.
2
not arise from interlayer interactions. Furthermore the Rev.Lett. 81, 2124 (1998).
results suggest that the loss of low energy intensity in 8D.S.Marshall, A.G.Loeser, Z.-X.ShenandD.S.Dessau,
Phys.Rev.Lett.76,4841(1996);A.G.Loeser,Z.-X.Shen
the B channelis due to strongantiferromagneticinter-
1g
actions which in turn implies E ≈J≈700cm−1 for the and D. S.Dessau, Science273, 325 (1996).
g 9J. M. Harris, Z.-X. Shen, P. J. White, D. S. Marshall, M.
pseudogap, an energy scale that is much larger than the
C. Schabel, J. N. Eckstein, and I. Bozovic, Phys. Rev. B
energy of the superconducting gap (∆ ≈ 100cm−1).
max 54, R15665 (1996).
Theseenergyscalesimplythatthepseudogapandsuper- 10H. Ding, T. Yokoya, J. C. Campuzano, T. Takahashi, M.
6
Randeria, M. R. Norman, T. Mochiku, K. Kadowaki, and 32A. J. Millis and H. Monien, Phys. Rev. Lett. 70, 2810
J. Giapintzakis, Nature 382, 51 (1996); H. Ding, M. R. (1993).
Norman,T.Yokoya,T.Takeuchi,M.Randeria,J.C.Cam- 33J. W. Loram, K. A. Mirza, J. R. Cooper, J. Phys. Chem.
puzano, T. Takahashi, T. Mochiku, and K. Kadowaki, of Solids 59, 2091 (1998).
Phys.Rev. Lett.78, 2628 (1997). 34H.Yasuoka, Physica C 282-287, 119 (1997).
11M. R. Norman, H. Ding, M. Randeria, J. C. Campuzano, 35B. Batlogg, H. Y. Hwang, H. Takagi, R. J. Cava, H. L.
T.Yokoya,T.Takeuchi,T.Takahashi,T.Mochiku,andK. Kao, and J. Kwo, Physica C 235-240, 130 (1994).
Kadowaki, P. Guptasarma, and D. G. Hinks, Nature 392, 36G.V.M.Williams,J.L.Tallon,E.M.Haines,R.Michalak,
157 (1998). and R. Dupree,Phys.Rev.Lett. 78, 721 (1997).
12J. W. Loram, K. A. Mirza, J. R. Cooper, N. Athanas- 37Ch. Niedermayer, C. Bernhard, T. Blasius, A. Golnik, A.
sopoulou,andW.Y.Liang,P341inProc.of10thAnniver- Moodenbaugh,andJ.I.Budnick,Phys.Rev.Lett.80,3843
sary HTS Workshop, (Wolrd Scientific, Singapore, 1996). (1998).
13X. K. Chen, J. G. Naeini, K. C. Hewitt, J. C. Irwin, R. 38K. B. Lyons, P. A Fleury, L. F. Schneemeyer, and J. V.
Liang, and W. N.Hardy,Phys. Rev.B 56, R513 (1997). Waszczak, Phys. Rev. Lett. 60, 372 (1988); K. B. Lyons,
14R. Nemetschek, M. Opel, C. Hoffmann, P. F. Muller, R. P. A Fleury, A. S. Cooper, and T. J. Negran, Phys. Rev.
Hackl, H. Berger, L. Forro, A. Erb, and E. Walker, Phys. B 37, 2353 (1988).
Rev.Lett. 78, 4837 (1997). 39S.Sugai,S.Shamoto,andM.Sato,Phys.Rev.B38, 6436
15G.Blumberg,M.Kang,M.V.Klein,K.Kadowaki,andC. (1988).
Kendziora, Science 278, 1427 (1997). 40D. Reznik, S. L. Cooper, M. V. Klein, W. C. Lee, D.
16J. W. Quilty,H. J. Trodhal, and D. M. Pooke, Phys. Rev. M. Ginsberg, A. A. Maksimov, A. V. Puchkov, I. I. Tar-
B 57, R11097 (1998). taovskii, S. Cheong, Phys. Rev.B 48, 7624 (1993).
17J. L. Tallon, G. V. M. Williams, N. E. Flower, and C. 41G. Blumberg, P. Abbamonte, M. V. Klein, W. C. Lee, D.
Bernard, Physica C 282-287, 236 (1997). M. Ginsberg, L. L. Miller, A. Zibold, Phys. Rev. B 53,
18V. J. Emery and S. A. Kivelson, Nature 374, 434 (1995); R11930 (1996).
V.J.EmeryandS.A.Kivelson,andO.Zachar,Phys.Rev. 42M. Rubhausen, C. T. Rieck, N. Dieckmann, K. O. Subke,
B 56, 6120 (1997). A.Bock, and U.Market, Phys.Rev.B 56, 14797 (1997).
19P.W.Anderson,Science235,1196(1987); X.G.Wenand 43T. P. Devereaux, D. Einzel, B. Stadlober, R. Hackl, D. H.
P. A.Lee, Phys. Rev.Lett. 76, 503 (1996). Leach,andJ.J.Neumeier,Phys.Rev.Lett.72,396(1994).
20D.Pines,Z.Phys.B103,129(1997);PhysicaC282-287, 44X. K. Chen, E. Altendorf, J. C. Irwin, R. Liang, and W.
273 (1997). N. Hardy, Phys. Rev. B 48, 10530 (1994); X. K. Chen, J.
21V.BarzykinandD.Pines,Phys.Rev.B 52,13585 (1995); C.Irwin,R.Liang,andW.N.Hardy,J.Supercond.7,435
Y. Zha, V. Barzykin and D. Pines, Phys. Rev. B 54, 7561 (1994); Physica C 227, 113 (1994).
(1996). 45P. J. White, Z.-X. Shen, J. M. Harris, A. G. Loeser, P.
22A. V. Chubukov, D. Pines, and B. Stojkovic, J. Phys.: Fournier,A.Kapitulnik,Phys.Rev.B54,R15669(1996).
Cond.Matter8,10017(1996);J.Schmalian,D.Pines,and 46C. Kendziora, R. J. Kelley, and M. Onellion, Phys. Rev.
B. Stojkovic, PhysRev.Lett. 80, 3839, (1998). Lett. 77, 727 (1996).
23T. Kimura, K. Kishio, T. Kobayashi, Y. Nakayama, M. 47Ch. Renner, B. Revaz, J. -Y. Genoud, K. Kadowaki, and
Motohira, K. Kitazawa, and K. Yamafuji, Physica C 192, O.Fischer, Phys.Rev. Lett. 80, 149 (1998).
247 (1992). 48T. P. Devereaux and A. Kampf, J. Phys. Chem. of Solids
24D. J. Werder, C. H. Chen, M. Gurvitch, B. Miller, L. F. 59, 1972 (1998); (cond-mat/9711039).
Schneemeyer, and J. V. Waszczak, Physica C 160, 411 49L.B.IoffeandA.J.Millis,Phys.Rev.B58,11631(1998).
(1989).
25X.K.Chen,J.C.Irwin,H.J.Trodhal,T.Kimura,andK. ∗Present address: Department of Material Science, Shizuoka
Kishio, Phys. Rev. Lett. 73, 3290 (1994); Physica C 295, University,Johoku, Hamamatsu 432, Japan.
80 (1998). †Present address: Joint Research Center for Atom Technol-
26J. G. Naeini, X. K. Chen, K. C. Hewitt, J. C. Irwin, T. ogy, Higashi, Tsukuba305, Japan.
P.Devereaux,M.Okuya,T.Kimura,andK.Kishio,Phys.
Rev.B 57, R11077 (1998).
27J. G. Naeini, J. C. Irwin, M. Okuya, T. Kimura, T.
Sasagawa, Y. Togawa, and K.Kishio, to bepublished.
28See J. L. Tallon, C. Bernard, H. Shaked,R.L. Hitterman,
and J. D. Jorgensen, Phys. Rev. B 51, 12911 (1995), for
an estimation of the hole concentration in Y123.
29T.Katsufuji,Y.Tokura,T.Ido,andS.Uchida,Phys.Rev.
B 48, 16131 (1993).
30M. Opel, R. Nemetschek, C. Hoffmann, P. F. Muller, R.
Philipp, R. Hackl, H. Berger, L. Forro, A. Erb, and E.
Walker, J. Phys. Chem. of Solids 59, 1942 (1998).
31M. V. Klein and S. B. Dierker, Phys. Rev. B 29, 4976
(1984).
7