Strongly correlated s-wave superconductivity in the N-type infinite-layer cuprate

Strongly Correlated s-Wave Superconductivity in the N-Type In?nite-Layer Cuprate

C.-T.Chen,1P.Seneor,1N.-C.Yeh,1R.P.V asquez,2L.

D.Bell,2C.U.Jung,3J.Y.Kim,3

Min-Seok Park,3Heon-Jung Kim,3and Sung-Ik Lee3

1Department of Physics,California Institute of Technology,Pasadena,California91125

2Jet Propulsion Laboratory,California Institute of Technology,Pasadena,California91109

3National Creative Research Initiative Center for Superconductivity and Department of Physics,

Pohang University of Science and Technology,Pohang790-784,Korea

(Received17November2001;published16May2002)

Quasiparticle tunneling spectra of the electron-doped(n-type)in?nite-layer cuprate Sr0.9La0.1CuO2

reveal characteristics that counter a number of common phenomena in the hole-doped(p-type)cuprates.

The optimally doped Sr0.9La0.1CuO2with T c?43K exhibits a momentum-independent superconduct-

ing gap D?13.061.0meV that substantially exceeds the BCS value,and the spectral characteristics

indicate insigni?cant quasiparticle damping by spin?uctuations and the absence of pseudogap.The re-

sponse to quantum impurities in the Cu sites also differs fundamentally from that of the p-type cuprates

with d x22y2-wave pairing symmetry.

PACS numbers: 74.72.–h, 74.50.+r, 74.62.Dh

The predominantly d x22y2pairing symmetry[1,2],the

existence of spin?uctuations in the CuO2planes[3,4],

and the pseudogap phenomena[3–5]in the underdoped

and optimally p-type cuprates have been widely con-

ceived as essential to high-temperature superconductivity. However,recent scanning tunneling spectroscopic studies

have shown that the pairing symmetry may be dependent

on the hole-doping concentration,with?d x22y21s?

mixed symmetries in certain overdoped cuprates such

as?Y12x Ca x?Ba2Cu3O72d[6].Furthermore,whether the pairing symmetry is d x22y2or s wave in the one-

layer n-type cuprates such as Nd1.85Ce0.15CuO42d and

Pr1.85Ce0.15CuO42d remains controversial[7,8],and it has

been suggested that the pairing symmetry in the one-layer

n-type cuprates may change from d x22y2to s,depending

on the electron doping level[9].The nonuniversal pairing symmetries in cuprate superconductors imply that the sym-metry is likely the result of competing orders rather than a suf?cient condition for pairing.Nonetheless,an important consequence of either d x22y2or?d x22y21s?-wave pairing is that the resulting nodal quasiparticles can interact strongly with the quantum impurities in the CuO2planes [10,11],such that a small concentration of impurities can give rise to strong suppression of superconductivity and modi?cation of the collective Cu21spin excitations [6,12–17].In addition,Kondo effects could be induced by nonmagnetic impurities through breaking the nearest-neighbor antiferromagnetic Cu21-Cu21interaction[18]. Such strong response to nonmagnetic impurities is in sharp contrast to conventional s-wave superconductivity[19,20]. Despite signi?cant progress in the studies of cuprate superconductivity,the research on the simplest form of cuprates,the in?nite-layer system Sr12x L x CuO2(L?La, Gd,Sm),has been limited[21–23]due to the dif?culties in making single-phase samples with complete superconduct-ing volume.Recently,Jung et al.[24]have demonstrated single-phase samples of Sr0.9La0.1CuO2with nearly100%superconducting volume and a sharp superconducting tran-sition temperature at T c?43K,thus enabling reliable spectroscopic studies of the pairing symmetry and the ef-fects of quantum impurities.These single-phased in?nite-layer cuprates are n-type with P4?mmm symmetry,which differ signi?cantly from other cuprates in that no excess charge reservoir block exists between consecutive CuO2 planes except a single layer of Sr(La),as illustrated in Fig.1(a),suggesting stronger CuO2interplanar coupling. Furthermore,the c-axis superconducting coherence length ?j c?0.53nm?is found to be longer than the c-axis lattice constant?c0?0.347nm?[25],in stark contrast to other cuprate superconductors with j c?c0.Hence,the super-conducting properties of the in?nite-layer system are ex-pected to be more three-dimensional,as opposed to the quasi-two-dimensional nature of all other cuprates.In this Letter,we report experimental?ndings based on the scan-ning tunneling spectroscopy studies of pure in?nite-layer samples and those with a small concentration(1%)of ei-ther magnetic or nonmagnetic quantum impurities.A num-ber of surprising results are found and compared with the established phenomena in other cuprates.

The samples studied in this work included high-density granular materials of Sr0.9La0.1CuO2(SLCO), Sr0.9La0.1?Cu0.99Zn0.01?O2(1%Zn-SLCO),and Sr0.9La0.1?Cu0.99Ni0.01?O2(1%Ni-SLCO)[24].X-ray diffraction(XRD)con?rmed the single-phase nature of all samples,and both XRD and scanning electron microscopy [24]revealed random grain orientation and a typical grain size of a few micrometers in diameter.Magnetization studies revealed nearly100%superconducting volume for all samples,with T c?43K for SLCO and1%Zn-SLCO,and T c?32K for1%Ni-SLCO.Structurally,the in?nite-layer system with up to,3%Zn or Ni substitu-tions was stoichiometrically homogeneous[24].However, the superconductivity appeared to be sensitive to the type of impurities.While nonmagnetic Zn had little effect on

FIG.1.(a)Comparison of the structure of the in?nite-layer system Sr12x L x CuO2(L?La,Gd,Sm),with those of the one-layer p-type(T-phase)and one-layer n-type(T0-phase)cuprates.

(b)A representative surface topography of an area of SLCO with subnanometer?atness.The typical area with atomic-scale ?atness where most tunneling spectra were taken was greater than(20nm320nm),and the work function of the spectra was 0.1?1eV.(c)A zoom-out view of the region shown in part(b) (indicated by the dashed box)over an area(49nm340nm). Also shown in the lower left corner is a grain boundary.

T c for up to3%concentration,strong suppression of T c already occurred with1%Ni,and nearly complete sup-pression of T c was reached with only2%Ni[24].Thus, the global response of SLCO to impurities appeared dif-ferent from that in the p-type cuprates[6,12–17]and was similar to that in conventional superconductors[19,20]. Quasiparticle tunneling spectra were taken using a low-temperature scanning tunneling microscope on hundreds of randomly oriented grains for the three different in?nite-layer samples,so that a range of different quasiparticle momenta relative to the crystalline axes of the local grains could be sampled.The sample surface was prepared ac-cording to the chemical etching method described else-where[26],and a nearly stoichiometric surface with no discernible chemical residue was con?rmed with the x-ray photoemission spectroscopy[26].A typical surface topog-raphy of the pure SLCO sample for our spectroscopic stud-ies with subnanometer?atness is exempli?ed in the left panel of Fig.1(b),and a zoom-out view of this area is il-lustrated in Fig.1(c).Con?rming the local?atness for the tunneling spectra was to ensure that the average momen-tum of the incident quasiparticles relative to the crystalline axes of a grain was well de?ned.A set of representative differential conductance?dI?dV?vs biased voltage?V?spectra for such a?at area is given in Fig.2(a).In general, all spectral characteristics revealed long-range(.50

nm)FIG.2.(a)Representative dI?dV vs V quasiparticle spectra of SLCO taken at4.2K.The curves correspond to spectra taken at?1.5nm equally spaced locations within one grain and have been displaced vertically for clarity except the lowest curve.Left inset:a typical spectrum taken at4.2K(solid line)compared with the corresponding high-voltage background(dashed line). Right inset:comparison of a typical spectrum taken at4.2K with one taken slightly above T c.(b)A spectrum normalized relative to the high-voltage background given in the left inset of(a),together with a BCS theoretical curve for the normal-ized DOS at?T?T c??0.1and a corresponding c-axis tunnel-ing spectrum for a pure d x22y2-wave superconductor(thin solid line).Left inset:a normalized c-axis tunneling spectrum of an optimally doped YBa2Cu3O72d(T c?92.560.5K).Right inset:a typical spectrum for the1%Zn-SLCO sample taken at4.2K.

spatial homogeneity within each grain and small variations in the superconducting gap value(D?13.061.0meV) from one grain to another.Here?2D?e?was de?ned as the conductance peak-to-peak separation in the spectra.This observation was in sharp contrast to our previous?ndings of strongly momentum-dependent spectra in the p-type cuprates with d x22y2pairing symmetry[6].The absence of the zero bias conductance peak(ZBCP)[6],known as a hallmark for unconventional pairing symmetry,for over1000spectra provided additional support for a fully gapped Fermi surface.

Despite suggestive evidence for s-wave pairing symme-try,the unusually large ratio of?2D?k B T c??7.0as com-pared with the BCS ratio of3.5was indicative of strong coupling effects.Moreover,the commonly observed“sat-ellite features”in the quasiparticle spectra of p-type cuprate superconductors[6],as exempli?ed in the left in-set of Fig.2(b),were invisible in SLCO.The satellite fea-tures in p-type cuprates were associated with quasiparticle

damping by many-body interactions such as the collective spin excitations[6,27,28].Thus,the absence of satellite features in SLCO is consistent with weakened spin?uc-

tuations as the result of diluted antiferromagnetic coupling due to the presence of Cu11introduced by electron dop-ing.In addition,D was found to completely vanish above T c,with no apparent energy scale associated with any depression of the density of states(DOS)at T.T c,as

shown in the right inset of Fig.2(a),and the tunneling spectra were nearly temperature independent from just above T c to?110K.The absence of any spectroscopic pseudogap in the n-type in?nite-layer system was dis-tinctly different from the?ndings in optimally doped and

underdoped p-type cuprates[5]and was independently veri?ed by the NMR studies on similar samples[29].

By normalizing a typical spectrum in Fig.2(a)relative to the background conductance shown in the left inset of Fig.2(a),we compared the quasiparticle DOS of SLCO with the BCS theoretical curve,as illustrated in Fig.2(b).

The spectral weight of SLCO for quasiparticle energies at j E j$D was smaller than the BCS prediction,whereas additional DOS appeared for j E j,D and the DOS ap-proached0at the Fermi level(i.e.,V?0).Such behavior cannot be accounted for by the simple inclusion of disorder

in the BCS weak-coupling limit,because the latter would have only broadened the width of the conductance peaks and also increased the DOS near V?0substantially.The spectra also differed fundamentally from those of pure d x22y2-wave cuprates[6]because of the lack of discernible

gap variations and of the absence of ZBCP in all spectra taken on random grain orientations.Even in a special case of c-axis tunneling,j d2I?dV2j V!06would have been a positive constant in a d x22y2-wave superconductor,as simulated by the thin solid line in Fig.2(b),which is in contrast to the?nding of j d2I?dV2j V!06?0in SLCO.

Interestingly,recent Knight shift data from NMR studies of similar SLCO samples have revealed much smaller normal-state DOS at the Fermi level as compared with those of other cuprates[29],which corroborates the inap-plicability of weak-coupling theory to SLCO.We there-

fore suggest strongly correlated s-wave pairing in the in?nite-layer system based on the empirical?ndings of momentum-independent quasiparticle spectra,absence of ZBCP,and j d2I?dV2j V!0?0for all grain orientations. In the case of1%Zn-SLCO,the spectral characteristics also revealed long-range spatial homogeneity in the

spectra and a similar gap value(D?13.062.5meV) for randomly sampled areas in different grains,as ex-empli?ed in the right inset of Fig.2(b).Given that the average separation among Zn impurities is??1.831.83 1.6?nm3,our exhaustive spectral studies should have

covered a signi?cant number of Zn impurities.However, no signi?cant local variations were found in the spectra of the1%Zn-SLCO,which differed fundamentally from our observation of atomic-scale spectral variations in a YBa2?Cu0.9934Zn0.0026Mg0.004?3O6.9single crystal near nonmagnetic Zn or Mg impurities using the same apparatus [6].Nevertheless,the conductance peaks in1%Zn-SLCO were signi?cantly broadened relative to pure SLCO,with an increase in the DOS for j E j,D,as shown in the right inset of Fig.2(b).These features suggest that Zn impu-rities resulted in reduced quasiparticle lifetime while re-taining T c,similar to the response of conventional s-wave superconductors[19,20].

In contrast,two types of spectra were observed in1% Ni-SLCO,as illustrated in Fig.3(a).The majority spec-tra(.90%)exhibited suppressed coherence peaks,large zero bias residual conductance,strong electron-hole spec-tral asymmetry,and gradual spatial evolution over a long range.In contrast,the minority spectra(,10%)exhib-ited sharp spectral peaks,small zero bias conductance, and varying electron-hole spectral asymmetry over a short range(,1nm),as exempli?ed in the inset of Fig.3(a)for two representative minority spectra.The signi?cant spec-tral asymmetry implied different phase shifts in the elec-tronlike and holelike quasiparticle states as the result of broken time-reversal symmetry[20,30],which may be re-sponsible for the global suppression of the superconducting phase coherence and thus a reduction in T c.

Assuming homogeneous Ni-impurity distributions,the average Ni-Ni separation would be d Ni?1.8nm in the ab plane and?1.6nm along the c axis in each grain. The impurity wave function with poor screening from the carriers would have extended over a coherence vol-ume?j2ab j c?[20,30].Given the coherence lengths j ab?4.8nm and j c?0.53nm[25],?30%volume probabil-ity in each grain could be considered as under signi?cantly weaker impurity in?uence.In the limit of completely ran-dom grain orientation in1%Ni-SLCO,the STM studies of the grain surfaces would have?20%probability for ?nding surface regions with weak impurity in?uence and spatial extension over a short range(?0.5nm)along the c axis.This simple estimate is in reasonable agreement

-0.5

0.0

0.5 n

o

r

m

a

l

i

z

e

d

d

I

/

d

V

V(mV)

d

I

/

d

V

(

n

S

)

V(mV)

FIG.3.(a)Main panel:comparison of a normalized majority spectrum of1%Ni-SLCO and that of pure SLCO at4.2K.The normalization was made relative to the background conductance shown by the dashed line in part(c).Inset:two minority spectra with different electron-hole asymmetry.(b)Spectral difference of the majority spectra relative to that of the pure SLCO.(c)A series of spectra taken on the same grain of1%Ni-SLCO at ?3nm apart.The conductance of all curves except the lowest one has been displaced up for clarity.

with our observation of?10%minority spectra with short-

range(,1nm)spatial homogeneity.However,due to the

lack of direct information for the Ni distribution on the sample surface,the true origin for two types of spectra in

1%Ni-SLCO remains uncertain.

Considering the spectral difference between the major-

ity spectrum of1%Ni-SLCO and that of pure SLCO,as

shown in Fig.3(b),we?nd that the spectral characteris-tics resemble the?ndings in Ref.[20]and are representa-

tive of the impurity-induced state.On the other hand,the

slowly varying majority spectra of1%Ni-SLCO,as shown

in Fig.3(c),were possibly the result of strong overlapping

in the Ni-impurity wave functions and of weak screen-ing effects due to low carrier density in SLCO,which

differed markedly from the rapidly diminishing impurity

effects away from an isolated Mn or Gd atom on the

surface of Nb[20],and also from the strong atomic-scale

spectral variations near Ni impurities in the p-type cuprate

Bi2Sr2Ca?Cu12x Ni x?2O81x[17].The contrast in the spa-tial extension of the Ni-impurity effects may be attributed

to the variation in the impurity coupling strength and range,

and also to the degree of impurity screening by carri-

ers.We suggest that the strongly interacting Ni impurities

in1%Ni-SLCO are analogous to a Kondo alloy,which cannot be explained by the Abrikosov-Gor’kov theory for magnetic impurities in BCS superconductors[30].

The parent materials of all p-type cuprates are Mott

insulators with strong on-site Coulomb repulsion[3,4].

Thus,the formation of d x22y2-wave pairing symmetry is energetically favorable in reducing the Coulomb repulsion while retaining the quasi-two-dimensionality.On the other hand,the strong three-dimensional coupling in the in?nite-layer system could favor s-wave pairing symmetry by com-pensating the resulting increase in the Coulomb repulsion with a large gain in the condensation energy.Thus,the pairing symmetry of cuprate superconductors may be de-pendent on the speci?c structures and various competing energy scales.Similarly,the pseudogap phenomena may be due to competing orders and need not be universal for all cuprates.

A recent study of the angular-resolved photoemission

spectroscopy(ARPES)on three different families of

p-type cuprates suggested that an abrupt change of the electron velocity in the5080meV energy range was ubiquitous and might be associated with the longitudinal optical oxygen phonon modes in the CuO2planes[31]. Such changes in ARPES approximately coincided with a“dip”feature in the quasiparticle tunneling spectra of some cuprates[6].However,our tunneling spectra of SLCO revealed a dip energy at?20meV,and that of YBa2?Cu0.9934Zn0.0026Mg0.004?3O6.9at?30meV,much smaller than the energy?50meV for pure YBa2Cu3O6.9 [6].Thus,the only ubiquitous features among all cuprates appear to be the strong electronic correlation and the background antiferromagnetism of Cu21ions in the CuO2 planes.

The research at Caltech was supported by NSF Grant No.DMR-0103045and the Caltech President’s Fund.Part of the work was performed by the Center for Space Mi-croelectronics Technology,Jet Propulsion Laboratory,and was sponsored by NASA.The work at Pohang University was supported by the Ministry of Science and Technology of Korea through the Creative Research Initiative

Program.

[1]C.C.Tsuei and J.R.Kirtley,Rev.Mod.Phys.72,969

(2000),and references therein.

[2]D.J.VanHarlingen,Rev.Mod.Phys.67,515(1995).

[3]J.Orenstein and https://www.sodocs.net/doc/c611801773.html,lis,Science288,468(2000),and

references therein.

[4]S.Sachdev,Science288,475(2000),and references

therein.

[5]T.Timusk and B.Statt,Rep.Prog.Phys.62,61(1999),

and references therein.

[6]N.-C.Yeh et al.,Phys.Rev.Lett.87,087003(2001),and

references therein.

[7]L.Alff et al.,Phys.Rev.Lett.83,2644(1999).

[8]C.C.Tsuei and J.R.Kirtley,Phys.Rev.Lett.85,182

(2000),and references therein.

[9]J.A.Skinta et al.,Phys.Rev.Lett.88,207005(2002).

[10]A.V.Balatsky,M.I.Salkola,and A.Rosengren,Phys.

Rev.B51,15547(1995).

[11]M.E.Flatte,Phys.Rev.B61,14920(2000).

[12]H.Alloul et al.,Phys.Rev.Lett.67,3140(1991).

[13]Y.Sidis et al.,Phys.Rev.Lett.84,5900(2000).

[14]S.H.Pan et al.,Nature(London)403,746(2000).

[15]W.A.MacFarlane et al.,Phys.Rev.Lett.85,1108(2000).

[16]K.Ishida et al.,Phys.Rev.Lett.76,531(1996).

[17]E.W.Hudson et al.,Nature(London)411,920(2001).

[18]A.Polkovnikov,S.Sachdev,and M.Vojta,Phys.Rev.Lett.

86,296(2001).

[19]P.W.Anderson,J.Phys.Chem.Solids11,26(1959).

[20]A.Yazdani et al.,Science275,1767(1997).

[21]J.D.Jorgensen et al.,Phys.Rev.B47,14654(1993).

[22]T.Siegrist,S.M.Zahurak,D.W.Murphy,and R.S.Roth,

Nature(London)334,231(1988).

[23]M.G.Smith et al.,Nature(London)351,549(1991).

[24]C.U.Jung et al.,Physica(Amsterdam)366C,299(2002);

Phys.Rev.B65,172501(2002).

[25]M.-S.Kim et al.,cond-mat/0102420.

[26]R.P.Vasquez,C.U.Jung,J.Y.Kim,M.S.Park,H.J.Kim,

and S.I.Lee,J.Phys.Condens.Matter13,7977(2001).

[27]A.V.Chubukov,N.Gemelke,and A.Abanov,Phys.

Rev.B61,R6467(2000),and references therein. [28]C.L.Wu,C.Y.Mou,and D.Chang,Phys.Rev.B63,

172503(2001).

[29]G.V.M.Williams et al.,cond-mat/0111421.

[30]A.A.Abrikosov and L.P.Gor’kov,Sov.Phys.JETP12,

1243(1961).

[31]https://www.sodocs.net/doc/c611801773.html,nzara et al.,Nature(London)412,510(2001).