Quasiparticle tunneling spectroscopy of high {Tc} cuprates

Quasiparticle Tunneling Spectroscopy of High-T, Cuprates* John Zasadzinski, L. Ozyuzer, 2. Yusof Science and Technology Centerfor Superconductivity Physics Department Illinois Institute of Technology, Chicago, IL 60616 Jun Chen, K.E. Gray, R. Mogilevsky, D.G. Hinks Materials Science Division Argonne National Laboratory, Argonne, IL 60439 1 7 1996 J.L. Cobb, and J.T. Markert University of Texas at Austin, Austin, TX 78712 SPIE International Symposium on Lasers and Integrated Optoelectronics, January 27-February 2, 1996, San Jose, CA. $De&rOSCOD -ic Studies of Hiph T, CUDrates, edited by I. Bozovic and D. van der Marel (SPIE, Bellingham, 1996) DISCLAIMER This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. *Work supported by the U.S. Department of Energy, Basic Energy Sciences-Materials Science under Contract #W-31-109-ENG-38 (KEG, DH) and the National Science Foundation Office of Science and Technology Centers for Superconductivity under Contract #DMR91-20000 (JZ, JC, ZY). Scicncc and Tcchnology Center for Supcrconductivity Illinois Institute of Technology. Physics Dcparimcni Chicago, Illinois 60616 Junc Chen, K. E. Gray. R. Mogilevsky, D. G . Hinks - Materials Science Division Argonne National Laboratory Argonne, Illinois 60439 J. L. Cobb and J. T. Markert University of Texas at Austin Austin, TX 78712 ABSTRACT Superconductor-insulator-normal metal (SIN) and superconductor-insulator-superconductor (SIS) tunnel junctions provide important information on pairing state symmetry and mechanism. Measurements of such junctions on high Tc superconductors (HTS) are reported using mechanical point contacts, which generally display the optimum characteristics that can be obtained from HTS native-surface tunnel barriers. New tunneling data on the infinite-layer cuprate, SrlxNdxCu02 are reported which show a remarkable similarity to another electron-doped cuprate, Ndl .gsCq)JyjCu04. In particular, there is a strong, asymmetric linear background conductance that is indicative of inelastic tunneling from a continuum of states. A discussion is given of the anomalous "dip" feature found in the tunneling and photoemission data on BSCCO 2212. It is shown that a similar feature is found in many cuprate junctions and that this dip scales with the gap energy over a wide range. New data on the single-layer, tetragonal cuprate, T12Ba2CuOg (Tl2201) are presented and discussed in light of recent published results on the similar compound HgBa2CuOq (Hg1201). The Hg1201 data display a low, flat sub-gap tunneling conductance which is consistent with a BCS density of states whcreas the Ti2201 data display a cusp-like feature at zero bias which is more consistent with dx2-y2 symmetry. 1 INTRODUCTION Quasiparticle (single-electron) tunneling spectroscopy has long been viewed as a powerful probe of the superconducting state.' In conventionaI superconductors such as Pb and Nb, the gap in the density of states is seen in the tunneIing conductance and the strong-coupling phonon structures are found as well. Inversion of the data allows the determination of the gap energy, the electron-phonon spectral function, a2F(o),and the coulomb pseudopotential, p*, thereby providing a In the early studies of high Tc superconductors complete description of the superconductivity in conventional metals. (HTS) the tunneling spectroscopy results were plagued by poor reproducibility, however, more recent studies on better characterized samples has led to a consensus on energy gap values for a number of HTS. Tunneling datz on BSCCO 2212, for exam le, ar: highly reproducible and consistent results have been re orted by a number of groups using various junction methods3 A good review of HTS tunneling is given by Hasegawa et al. We report here progress made in the development of SIN and SIS junctions on oxide superconductors using a mechanical, point-contact tunneling (PCT) approach. All of the PCT data presented are from the ArgonnenIT tunneling group. For our instrument, the term "point contact" is somewhat of a misnomer in that the contact area of the tip and sample can be large compared to atomic dimensions. This is by design as the intention is not to obtain atomic scale images as with an STM but rather :c obtain stable contacts with relatively low resistances of typically 1 WZ to 20 k f i for improved signal-tonoise. Barrier height analysis of such junctions on Nb4 using a Au tip indicated a contact diameter of -2400 A. Furthermore, the observation of the Nb phonon structures consistent with planar junctions4 demonstrated that this method could be used for sensitive spectroscopy. This mechanical method has proven to be a reliable and versatile tool for making many * s ,- 2. TUNNELING DENSITY OF STATES No discussion of tunneling data can procecd without a brief discussion of the density of states (dos) for cuprate superconductors. The tunnel current in an SIN or SIS junction can be written as, Here, p 1(E) and p2(E) are the quasiparticle dos in the two electrodes and C is a constant which depends on (among other things) junction area. The Fermi functions, f(E) account for thermal population of quasiparticle states. Here, p(E)=l for a normal metal which points out an often neglected fact that all band structure effects have mysteriously disappeared from the integral. The absence of the band structure dos in the tunneling data of conventional superconductors is experimentally established and a theoretical argument for this has been given by Harrison (see chap. 2 of ref I). The tunneling matrix element, t2, has an energy dependence which is assumed to be weak over the voltage range of interest (energy gap region) and has been taken out of the integral. In the limit, T=O K, the tunneling conductance, os =dYdV, for an SIN junction becomes. where we have now set E=eV. Eq. 2 simply states that a measurement of the tunneling conductance at low temperatures should reveal the quasiparticle dos, which for a BCS superconductor is given by p(E)= E/(E2-A2)li2. An exact determination of p(E) requires a measurement of the weak, voltage-dependent background (or normal state) conductances to divide out the prefactors of eq. 2, however such a measurement can be difficult due to the high critical fields and temperatures of HTS. Also, in some cases the background conductances are far from having a weak voltage dependence as we will show. In the absence of a normal state measurement, it is common to normalize the data by a constant taken at some arbitrary high bias voltage. To examine the dos of HTS cuprates, we take a simple model identical to that described by Fedro and Koelling6 The two-dimensional Cu-02 planes are represented by a single, tight-binding band with nearest neighbor (NN) and second NN hopping described by t and t'. We choose t'=O for simplicity and a hole concentration of 0.18 which shifts the van Hove singularity above the Fermi energy at Ed. Fig. 1 shows what the superconducting energy gap looks like in the dos. 2 2 -7 1.6 1.6 n 3 v 1.2 1.2 0.8 0.8 0.4 0.4 v) Q 0 -( ,6 -0.2 0.2 I 0.6 0- oI2t Figure 1 Two-dimensional, tight-binding band dos with (a, isotropic gap and (b) J-wave gap. 6 oI2t A Ln 0 3 W b E 1'0 3.1 'Tlrc ;amnulous "clip" featurc PCT data on 1%-clopcd Bi2Sr$aCu2OX (BSCCO 2213) with a Tc=b6 K wcrc rcportcdI4 i n 1989. Two SIN tiinncling conductancc curves (Au tip) from that article arc' rcproduccd in Figs. 4(a) and 4(b) for thc temperatures. 4.2 K and 77 K respectively. The background conductances decrease with applicd voltage, an nnonialous feature in itself but one which is typical for BSCCO 2212. In Fig. 4(a) the positive bias (Au tip is + with respect to the sample) is characterized by a sharp conductance peak at -22 mV, a pronounced dip near 45 mV followed by another peak near 70 mV. For the negative bias direction, the dip feature is weaker (but nevertheless observable) and we originally described it as a shoulder. The observation of a dip feature near 2A in the SIN junctions on BSCCO and at 3A in SIS junctions has been found by many 0.34 0.48 0.44 -0.28 2 tn E 00.22 -2 0.4 v v 0 o 0.36 c 0 C 0 Q 0 w 0.32 0.16 3 7J 0 5 0.28 0 0.1 0 -a C 0.24 0.04 I . . . . I . . . . i . . . . i .... 2 0 0 - 1 5 0 - 1 0 0- 5 0 i . . . . i . . . . t . . . . , . . . . J 0 50 Voltage (mV) 100 150 200 0.2 I . . . . , . . . . I . . . . I . . . . , . . . . , . . . . , . . . . , . . . . J 200-150-100 - 5 0 0 50 100 150 200 Voltage (mV) Figure 4. Superconducting dI/dV (solid lines) on Pb-doped BSCCO 2212 for (a) T = 4.2K and(b) T = 77K. Dashed lines are estimates of the normal state background as described in the text. Voltage is that of tip with respect to sample. groups using various junction methods.2 Note that features are shifted by an additonal factor of A in SIS junctions compared to SIN. The dip is clearly a reproducible feature and appears to be connected to a similar dip feature found i n the spectral weight function measured in photoemission experiments.15 To gain some insight into this feature we have constructed estimates of the normal state curves by fitting the data for IVb40 mV to a high order polynomial. These curves are shown as dashed lines in Fig. 4. Considering the different conductance values for the two junctions and noting the different sensitivities in the ordinates, the inferred background shapes are quite similar. The reduced gap features of the 77 K junction put a greater focus on the peak near 70 mV and perhaps it is this feature to which attention should be paid. Similar dip features are found in other cuprate junctions including the electron-doped NCC02, although in that case it is a subtle feature exposed only because of the ability to measure the normal state conductance. SIN junctions on T12Ba2CaCu20x (TBCC0)16 generally displayed pronounced dips for both bias values, again at a voltage nearly twice that of the conductance peak. The dip features are symmetric and much more pronounced in SIS junctions2 and for this reason we have chosen to plot the SIS characteristics of various cuprate junctions on a single plot and this is shown in Fig. 5. In the cases of NCCO and TBCCO we generated the SIS curves from SIN data using eq. 1 and the superconducting dos for each electrode. The TC values of the cuprates range from 5.5 K for BSCCO 2201 to 100 K for TBCCO and to plot the data on a single graph we have normalized the voltage axis by Vp/2 where Vp is the voltage of the conductance peak. Using this normalization, the x-axis is in units of A. It is clear from Fig. 5 that the dip and subsequent peak features scale with the superconducting gap, which varies by a factor of 30 over the cuprates examined. 4 1.4 3.5 (d -r 1.2 3 1 0 3 T) c 2.5 0.8 0 0 2 0.6 1.5 .0.4 1 0.2 0.5 0 -8 0 -6 -4 -2 0 2 4 6 8 2WV P Figure 5 . Representative SIS normalized tunnel conductances for various cuprate superconductors (T., from 5.5 K to 100 K) plotted on a voltage axis renormalized in units which are A. Left scale: NCCO (solid line), BSCCO 2212 (dashed line). Right scale: TBCCO (dashed-dot line), BSCCO 2201 (solid line), BSCCO film (dashed line). - Note that in the case of NCCO the dip is about 3.5 meV from the conductance peak, far below any phonon peak energies (typically 10 meV-7OmeV) and therefore does not interfere2 with the determination of a2F(o). Linking the dip feature to a superconducting energy scale is important to understanding its origin. D. Coffey has argUedl7 that a natural explanation is an intrinsic, energy dependent quasiparticle decay mechanism, F(o), put into the dos of eq.3, which turns on at a characteristic energy, 2A or 3 4 for d-wave and s-wave respectively. He further argues that that the location of the dip at 3A in SIS in Fig. 5 (consequently 2A in the dos) is evidence for d-wave superconductivity. This is an attractive explanation however, the dip is found at the same location in NCCO which we have shown is most likely an s-wave superconductor. Perhaps the focus at the present time should not be on the precise location of the experimental dip as it might be affected by tunneling background shapes for example. An alternative explanation put forth recently by L. Coffey and K. Kouznetsov ( ~ ~ 1 is 1 in 8 terms of inelastic quasiparticle scattering processes off a strong spin fluctuation spectrum arizing from an oxygen deficient layer on the surface of the HTS. It is clear that the inelastic tunneling processes are present in many HTS junctions and by treating such processes using a realistic spin fluctuation spectrum, it is possible that the dip and peak features can be explained for the tunneling data and photoemission as well, If indeed, the anomalous dip is due to an inelastic tunneling process arising from a surface layer, then this might explain the observation by Shimada et all9 of phonon structures in BSCCO 2212 using Schottky type junctions. This experiment is one of the few to not show the pronounced dip in BSCCO 2212. Perhaps in this case, the intimate contact of the semiconductor to BSCCO inhibits any oxygen-deficient, surface layer al!owing predoniinantly elastic tunneling to be observed. 3.2 PCT junctions on T12Ba~CuOg It was recently reported by Chen et a120.21 that PCT junctions on the single-layer Hg based cuprate, HgBa2CuOd (Hg1201) exhibit BCS-like tunneling conductances for both SIN (Au tip) and SIS' (Nb tip) junctions. Given the choice of the two curves shown in Fig. 1, it was clear that the Hg1201 data were more compatible with an s-wave order parameter. Furthermore, the SiS' I(V) data exhibited sharp current onsets at the gap voltage making the Hg1201 material a potential candidate for quasiparticle based devices such as mixer-based photon dmctors.2 Typical gap parameters were A- 13- 16 nieV. but one junction had A=24 meV. Considering that Hg1201 is tetragonal and has a single C u - 0 layer per unit cell, we decided to examine a very similar compound, T12Ba2CuOg (T12201), \vhich has the same structural properties as Hg1201 and a similar Tc (91 K-95 K). Single-crystals of TI2201 approximately 0.5mm on edge were grown by Mogilevsky and Argonnc. The PCT incthod ol'l'crs distiiicl advantaycs I1)r tlic study 01. such sinail cryst;ils. Over 2 0 0 junctions have I x x i i iiiatfc on 20 tlil'lercnt crystals using both Au and Nb tips. The data :ire reproducible and wc show ;i rcprcscntativc set of t'our.junctioiis l'rom two samples in Fig. 6. Note here the voltagc is that of the sample rclativc to the tip. Hiiiks :it TI 2201 2 1.5 W > U 1 7 3 1 0.5 0 -200 -150 -100 -50 0 v 50 100 200 150 (mV> Fig. 6. Representative set of SIN (Au tip) junction conductances for two crystals of T12201. The curves are labelled (a) through (d) going from top to bottom. Junction (b) is represented by dots, all others by solid lines. The voltage, V, is that of the sample relative to the tip. First note the background shape which is weakly decreasing with applied bias voltage similar to that typically found in BSCCO 2212. The junction conductances are characterized (in many cases) by very sharp conductance peaks located near 20 mV. Estimating A=20 meV and using 91 K for Tc one obtains 2AkTc=5.1. We note that the ratio of peak conductance to the estimated normal state at 2 2 0 mV is often greater than 2 and in some cases is as high as 3.5, the latter value being one of the largest ratios we are aware of for any SIN cuprate junction, 3 2.5 2 bc \ t I- 1.5 bv) 1 0.5 0 - 1 00 -50 0 50 100 wmv> Figure 7. Normalized conductances ofT12201 junctions (a) and (d) of Fi.g. 7. Junction (c) ol' Fig,..0 is the niost hroadcncd ( 3 1 Ihc: foui c t i ~ t f111cioi1t1~1ii;iricc p k volt:~gchas hccn sliil'tcct to ;I slightly largcr valuc. This shirt of thc: conciuctnncc pcdi \oltagc to higher v:ilucs for hnatlcnctl junctions is ;1 co111111ol1 cl't'ccr i n oxidc supc.r.c:ori~iuctors.?~c ilrgtlc t~iattliosc junctions with thc xtiarpcst cont~uc~ancc pcaks arc ttic most rcprcscnlot ivc o f t11cbulk d o s o f TI720 I :tiid thus junctions (a) and (ti) tlroin two diffcrcnt crystals ;lrc chosen for luthcr consideration. Thc c u r ~ c swcrc noriidimd by ;i constant :ind arc shown in Fig. 7. Comparing thc normalizcci Conductances to thc dus plots of Fig. I , i t appcars thnt TI220 I more closcly rcsctnblcs a superconductor with ci,2-y2 symmetry. This is evident from thc pronounced cusp l'caturc in the data at zero bias. This cusp can be seen in all of the junctions of Fig. 6 and is a general feature of rnost of thc SIN junctions we have studied. Quantitative fits of the data using the d-wave model discussed above are currently underway. In comparing these data to those of Hgl'Ol, an obvious question arises. W h y would t w o such similar compounds display junction characteristics indicating different gap symmctries'? 4. SUMMARY Quasiparticle tunneling data on HTS cuprates can potentially give important information on pairing state symmetry and mechanism. However, it is clear that in many cases unusual background shapes occur in the tunneling conductances. The fact that these background shapes can vary from increasing with bias (often in a linear fashion) to decreasing with bias for the samc cuprate suggests that they are arising from a conduction process which is not elastic tunneling. A strong likelihood is inelastic tunneling which has been shown to produce linearly increasing backgrounds when the spin fluctuation spectrum is flat. Perhaps a more rigorous treatment of inelastic tunneling, including a realistic spin fluctuation spectrum, a tight binding band structure and directional tunneling effects will result in an explanation of all of the background shapes. The background must be understood before a quantitative analysis of the elastic tunneling part can be undertaken. One way of minimizing the problem is to focus on those junctions which have a relatively weak voltage dependent background at least for one bias direction. This approach worked with NCCO where a2F(w) spectra have been obtained. The origin of the anomalous dip feature is still not understood. Our observation that the dip feature scales with the energy gap puts severe constraints on its interpretation. The possibility of an energy dependent decay rate, T(w), is an attractive one and readily leads to the observed scaling behavior. However, there are still aspects of the dip feature which are not explained by this model, for example, the observed asymmetry of the effect with bias voltage as found in BSCCO 2212. There is also the possibility that the dip feature is associated with the same inelastic tunneling processes that affect the general background shape. Quasiparticle tunneling cannot probe the sign of the order parameter and therefore cannot directly determine the pairing state symmetry. However, strong inferences can be made. For example, the observation of highly reproducible gap values in NCCO strongly supports an isotropic s-wave state and this is verified by a number of other experiments on NCCO including penetration depth and Raman scattering. The origin of the small sub gap conductance remains a puzzle. Other electron doped systems, including the infinite layer cuprate, Srl -xNdxCuq2 display similar gap reproducibility which leads to the suggestion that they may be s-wave as well. Another way to probe the pairing state is to note that the s-wave and d-wave densities of states are quite different and should be reflected in the quasiparticle tunneling spectra. The observation of low, flat sub-gap conductances in junctions of Hg1201, for example, are difficult to reconcile with a d-wave pairing state. Strong directional tunneling effects must be invoked to explain the Hg1201 data within a d-wave scenario. In the case of the similar compound, T12201, the tunneling dos looks very much like the d-wave dos, including a pronounced cusp feature. In this case one must argue that there are no preferred tunneling directions and the total dos is being probed. For other cuprates, the presence of sub-gap conductance without any obvious cusp feature gives little information about pairing symmetry. Finally, it should be mentioned that under no circumstances has the van Hove singularity feature ever been observed in PCT or to our knowledge, any other tunneling method. Such a distinctive feature as seen in Fig. 1 might be expected to show up in the tunneling data. Being strictly a band structure effect, its absence is probably linked to the common absence of band structure effects in the tunneling dos as found in conventional metals. 5. ACKNOWLEDGEMENTS Wc acknowledge the :issistance at Argonne of Qiang Huang and Paoia Romano for tunneling measurements, John W s g n t r for bulk sample preparation , R. T. Kampwirth and C. Romeo for thin films OF BSCCO; This work is partially supported by the US. Department of Energy, Division of Basic Energy Sciences-Matcrials Sciences (KEG,DH), under contract No.w-3 1- 109-ENG-33 and rhc National Science Foundation, Office of Science and Technology Centers (JZ, JC, ZY), under contract NO. DMR 9 I -2oooO. 6. REFERENCES 1. E.L. Wolf, Principles of ElecrrofiTumieling Spectroscopy. Oxford University Press, New York, 1985 2. J.F. Zasadzinski el al, "Tunneling density of states in cuprate and bismuthate superconductors", J. Phys. Chem Solids, vol. 12, pp. 1635-1639, November 1992. 3. Tetsuya Hasegawa, Hiroshi Ikuta and Koichi Kitazawa, "Tunneling Spectroscopy of Oxide Superconductors", in Physcial Propenies of High Temperature Sriperconducrors III, ed. D.M. Ginsberg, World Scientific Publishing, 1992. 4. Q. Huang, J.F. Zasadzinski and K.E. Gray, "Phonon spectroscopy of superconducting Nb using point-contact tunneling", Phys. Rev. B. vol. 42, pp. 7953-7959, November, 1990. 5. Q. Huang et al, "Tunnelling evidence of predominantly electron-phonon coupling in superconducting Bal-xKxBi@ and Nd2-xCexCuOq-y",Nature, vol. 347, pp. 369-372, September, 1990. 6. AJ. Fedro and D.D. Koelling, "Interplay between band structure and gap symmetry in a two-dimensional anisotropic superconductor", Phys. Rev. B vol. 47, pp. 14342-14347, June 1993-1. 7. DJ. Scalapino, 'The case for dX2-,,2pairing in the cuprate supercoriductors"Physics Reports, vol. 250 No. 6, pp. 329365, January 1995. 8. P. Monthoux, A.V. Balatsky and D. Pines, "Toward a theory of high temperature superconductivity in the antiferromagnetically correlated cuprate oxides", Phys. Rev. Letters, vol. 67, pp. 3448-3451, December, 1991. 9. B. Stadlober et al, "Is Nd2-xCexCuOq a high-temperature superconductor?"; Phys. Rev. Letfen, vol. 74, pp. 49114914, June 1995. 10. N. Tralshawala, J.F. Zasadzinski, L. Coffey and Qiang Huang, "Interpretation of resistivity of Nd1-85Cq-15Cu04-y using the electron-phonon spectral function determined from tunneling data", Phys. Rev. B, vol. 44,pp. 12102-12105 11. June Chen, J. Zasadzinski, K.E. Gray, J.L. Cobb and J.T. Markert, "Tunneling study of the infinite layer compound Srl-xNdxCuO;?and the comparison to other n-doped cuprates" (preprint) 12. R.C. Dynes, V. Narayanamurti, and J.P. Garno, "Direct measurement of the quasiparticle lifetime broadening in a strong-coupled superconductor", Phys. Rev. Letters, vol. 41, pp. 1509-1512, November, 1978. 13. J.R. Kirtley. S. Washburn, and DJ. Scalapino, "Origin of the linear tunneling conductance background", Phys. Rev. B VOI. 45, pp. 336-345, January 1992-1. 14. Q. Huang et al, "Electron tunneling study of the normal and superconducting states of Bi1.7Pb0.3Sr2CaCu20x1', Phys. Rev. B. vol. 40, pp. 9366-9369, November, 1989. 15. Z.X. Shen et al, "Anomalously large gap anisotropy in the a-b plane of BiZSr2CaCu208", Phys. Rev. Leners, v d . 70, pp. 1553-1556, March 1993. 16. Qiang Huang, J.F. Zasadzinski, K. E. Gray, E.D. Bukowski and D.M. Ginsberg, "Point-contact tunneling study of the normal and superconducting states of Tl2Ba2CaCu2w, Physicu C. vol. 161, pp- 141-144,1989. 17. D. Coffey, "Effect of spontaneous decay of superconductior quasiparticlesin the tunneling density of states", J. Phys. Chem. Solids,vol. 54, pp. 1369-1376,1993. 18. K. Kouznetsov and L. Coffey, "Unified method of analysis of tunneling and photoemission spectra in high temperature superconductors", (preprint) 19. D. Shimada, Y. Shiina, A. Mottate, Y. Ohyagi and N. Tsuda, "Phonon structure in the tunneling conductance of Bi2S~CaCu20g",Phys. Rev. B.. vol. 51, pp. 16495-16498, June 1995-II. 20. Jun Chen et at, "Point-contact tunneling study of HgBa2CuOq+& BCS-like gap structure", Phys. Rev. B, vol. 49, pp. 3683-3656, February, 1994-1. 21 John Zasadzinski et al, Device-quality tunnel junctions on the high-Tc superconductor HgBazCuO4+$' Applied Supercondricrivity, vol. 2, pp. 7W-7 14, 1994 ~ "