PHYSICAL REVIEW B VOLUME 57, NUMBER 13 1 APRIL 1998-I Quasiparticle transport below T c in the cuprate superconductors F. Gollnik and R. P. Huebener Physikalisches Institut, Lehrstuhl Experimentalphysik II, University of Tuebingen, Morgenstelle 14, D-72076 Tuebingen, Germany ~Received 10 September 1997! The interpretation of experimental data on the Nernst and Seebeck effects in the mixed state of a type-II superconductor requires a careful treatment of the temperature-dependent electrothermal conductivity S n (T)/ r n (T) below T c ~S n 5Seebeck coefficient, r n 5resistivity!. For the cuprate superconductors such as YBa2Cu3O72d, a simple extrapolation from the normal-state behavior above T c is inadequate and can lead to erroneous results, due to the opening of the superconducting energy gap and its effect on the temperature dependence of the quasiparticle scattering rate. In particular, the value of the thermal Hall angle of vortex motion depends sensitively on the correct treatment of the electrothermal conductivity below T c . Taking experimental data for YBa2Cu3O72d films as an example, we show that the thermal Hall angle of vortex motion obtained from the data attains the expected small value only if the opening of the energy gap below T c is taken into account. @S0163-1829~98!07013-1# II. THERMAL DIFFUSION OF QUASIPARTICLES AND MAGNETIC FLUX QUANTA I. INTRODUCTION In the mixed state of a type-II superconductor, quasiparticles and magnetic flux quanta respond to a temperature gradient by thermal diffusion, in this way generating the Seebeck and Nernst effects, respectively.1–3 Since there exists a Hall contribution both to the thermal diffusion of quasiparticles and of magnetic flux quanta, two Hall angles appear in this vortex dynamics. From an analysis of our measurements of the Seebeck and Nernst effects in the mixed state of epitaxial c-axis-oriented YBa2Cu3O72d films, we have concluded recently that the Hall angle for the thermal diffusion of vortices appears to be 30–50 times larger than expected.2,3 In this paper we argue that this discrepancy is due to the opening of the superconducting energy gap and its effect on the quasiparticle transport below T c , and that the thermal Hall angle is consistent with the expected value. A key ingredient in the previous analysis2,3 has been the assumption that the electrothermal conductivity S n (T)/ r n (T) below T c can simply be obtained by extrapolation from its normal-state value above T c . Here S n and r n are the Seebeck coefficient and the electric resistivity, respectively. However, various experiments have shown recently that in the cuprate superconductors below T c the quasiparticle transport is sensitively affected because of the opening of the superconducting energy gap. This results mainly from the fact that electron-electron scattering is dominating in these materials. In the following we show that this modification of the quasiparticle transport below T c can account for the measured Seebeck and Nernst effects in the mixed state of YBa2Cu3O72d and that the discrepancy in the thermal Hall angle is removed. In Sec. II we summarize the theoretical description of the vortex motion in the mixed state, leading to the Seebeck and Nernst effects. The role of the electrothermal conductivity below T c is emphasized. In Sec. III we discuss the implication of the opening of the superconducting energy gap below T c for the electrothermal conductivity and how a large thermal Hall angle can appear instead as an artifact, if this implication is ignored. 0163-1829/98/57~13!/7495~4!/$15.00 57 In a temperature gradient the quasiparticles and the vortices present in the mixed state of a type-II superconductor both experience the thermal force f th52S * •“ x T. ~1! Here and in the following, we assume the temperature gradient along the x direction and the magnetic field in the z direction. In Eq. ~1!, S * denotes the transport entropy of the species being considered. In the case of charge carriers, S * is related to the Seebeck coefficient S via the relation S * 5S•e, ~2! where e is the elementary charge. ~Note that e is negative for electrons and positive for holes.! In the case of vortices, the thermal force ~1! and the transport entropy are usually given per unit length of vortex line. In the following we denote the transport entropy per unit length of vortex line by S w* . As discussed in detail elsewhere,2 the longitudinal ~E x , Seebeck effect! and transverse ~E y , Nernst effect! electric fields in the presence of the temperature gradient “ x T are given by the following expressions: S w* r m Sn Ex 5rm a, ~ 11 ab ! 1 “ xT rn w0 ~3! S w* r m Ey Sn 5 1rm ~ b2a !. “ xT w0 rn ~4! Here w 0 is the magnetic flux quantum and r m the electric resistivity in the mixed state. a 5tan uv and b 5tan uqp , where u v and u qp are the Hall angles associated with the motion of vortices and quasiparticles, respectively. The expressions in Eqs. ~3! and ~4! were derived earlier by Maki4 from the time-dependent Ginzburg-Landau theory. Similar results were also obtained by Samoilov et al.5 We see that the longitudinal electric field E x contains a first-order contri7495 © 1998 The American Physical Society 7496 BRIEF REPORTS bution from the Hall angle a of the thermal diffusion of the vortices, whereas the transverse electric field E y only contains higher-order contributions from the Hall angle of the thermal diffusion of the quasiparticles. On the right-hand side ~RHS! of Eq. ~4!, the difference b 2 a appears, since the Hall angles of the quasiparticles and of the vortices act in opposite directions and partly cancel each other. Within the Bardeen-Stephen model6 we have a 5 b , and the second term in Eq. ~4! vanishes. The first term on the RHS in Eq. ~3! indicates that in the mixed state the Seebeck coefficient also shows the broadening of the transition to the superconducting state typical for the resistivity r m in the mixed state of the p-doped cuprates. For epitaxial c-axis-oriented films of YBa2Cu3O72d and Bi2Sr2CaCu2O81x , it has been found that the first term on the RHS of Eq. ~3! with ab !1 approximately accounts for the observed Seebeck coefficient in the mixed state.3,7 In Ref. 2 a detailed analysis of experimental results for epitaxial c-axis-oriented YBa2Cu3O72d films has been performed. By extrapolating the temperature-dependent factor S n (T)/ r n (T) from the normal state above T c , the difference Ex S n~ T ! 2 r m~ T ! DS5 “ xT r n~ T ! S w* r m w0 a, r n (T) in the denominator of the second term on the RHS will turn out to be too large from such a simple extrapolation. Hence, this second term will be too small and the difference DS and the thermal Hall angle too large. In the following we examine this situation in more detail, using only highly simplified arguments. We write the electrothermal conductivity S n (T)/ r n (T) in the form14 E S n~ T ! 1 52 r n~ T ! Te ` 0 ~6! and the thermal Hall angle a has been determined from DS. In this way a thermal Hall angle in the range a 50.3– 0.5 has been found. However, this range is much larger than the value a '0.01 obtained from electric transport measurements. We emphasize that this analysis had been performed with the assumption that below T c the temperaturedependent electrothermal conductivity S n (T)/ r n (T) can be obtained simply by extrapolation from its normal-state value above T c . In the following we critically reevaluate the electrothermal conductivity below T c . In particular, we show that such a simple extrapolation becomes unjustified because of the opening of the superconducting energy gap. III. ELECTROTHERMAL CONDUCTIVITY S n „T…/ r n „T… BELOW T c One of the special features of the p-doped cuprate superconductors is the strong indication for the dominance of electron-electron scattering instead of electron-phonon scattering in the normal state.8,9 Therefore, the opening of the superconducting energy gap below T c strongly affects the quasiparticle transport properties. For example, measurements of the microwave surface resistance of YBa2Cu3O72d have shown a strong decrease of the inelastic quasiparticle scattering rate with decreasing temperature and an abrupt change of its temperature dependence near T c . 10,11 Furthermore, below T c a particularly strong decrease of the quasiparticle scattering rate has also been found in measurements of the thermal conductivity12 and of the thermal Hall conductivity13 of YBa2Cu3O72d. Hence, below T c , the quasiparticle transport cannot be obtained simply by extrapolation from the normal state above T c . In Eq. ~5! the resistivity S D d« ~ «2 m ! 2 ]f s~ « !. ]« ~7! Here « is the energy, m the chemical potential, and f the Fermi function. s~«! is the contribution to the conductivity s from electrons in the interval d« around «, s5 E S ]] D s ` 0 f « d« 2 ~ « !. ~8! Expanding s~«! around the Fermi energy « F ' m , s~ « !5s~ «F!1 ~5! was calculated ~Fig. 3 of Ref. 2!. From Eq. ~3! we have DS5 57 ]s ]« U ~ «2 m ! , ~9! «F we see that it is the second term on the RHS, describing the nonsymmetry between electrons and holes, which yields a nonvanishing contribution to the electrothermal conductivity. Next, we consider the changes in the electrothermal conductivity due to the opening of the superconducting energy gap D. In a highly simplified way, we take the quasiparticle condensation into Cooper pairs into account, by leaving out the energy interval between m 2D and m 1D in the integrals ~7! and ~8!. Ignoring for the moment any influence of D on the quasiparticle scattering rate, we obtain F E s DÞ0 5 s D50 12 1D/k B T 2D/k B T dz 1 ~ 11e z !~ 11e 2z ! G ~10! and S n~ T ! r n~ T ! U 5 DÞ0 S n~ T ! r n~ T ! 3 E U F 12 D50 1D/k B T 2D/k B T dz 3 p2 G z2 . ~ 11e z !~ 11e 2z ! ~11! In adopting such a procedure, we have not included the increase in the quasiparticle density of states above m 1D and below m 2D and the corresponding redistribution of the quasiparticle energy levels. However, near the band edge the quasiparticle group velocity will be reduced in a similar way as the density of states is enhanced. Hence the contribution of these quasiparticles to the current will not be strongly affected. It is instructive to compare the two integrands I (0) (z) 5 @ (11e z )(11e 2z ) # 21 and I (2) (z)5(3/p 2 )z 2 I (0) (z) from Eqs. ~10! and ~11!, respectively. These two functions I (0) (z) and I (2) (z) are displayed in Fig. 1. From this figure we see that the opening of the superconducting energy gap immediately affects the quantity s DÞ0 , whereas the electrothermal conductivity S n (T)/ r n (T) u DÞ0 at first remains nearly unchanged. For illustrating this point, in Fig. 2 we BRIEF REPORTS 57 FIG. 1. The functions I ( 0 ) (z)5 @ (11e z )(11e 2z ) # 21 ~solid line! and I ( 2 ) (z)5(3/p 2 )z 2 I ( 0 ) (z) ~dashed line! plotted vs z 5(«2 m )/k B T. show the square brackets of Eqs. ~10! and ~11! plotted versus D/k B T. Taking for D(T) the function D(T)5D(0) 31.74(12T/T c ) 1/2 near T c with D(0)51.76k B T c for a classical BCS superconductor, we obtain the plot of the normalized quantity S n~ T ! r n~ T ! U Y U DÞ0 S n~ T ! r n~ T ! , D50 i.e., of the square bracket of Eq. ~11!, versus T/T c shown by the solid line in Fig. 3. Up to now we have ignored the detailed influence of D on the quasiparticle states close to the energy gap. @One expects modifications of Eq. ~9! if coherence factors are taken into account.# Basically, the preceding discussion is a justification for such a simple treatment of the electrothermal conductivity close to T c , since in this temperature regime the diffusion current due to a temperature gradient is mainly carried by quasiparticles at energies appreciably beyond the gap. ~Note the position of the peaks of the dashed curve of Fig. 1.! From Fig. 3 we note that at T/T c 50.8 according to Eq. ~11! the opening of the energy gap reduces the electrothermal conductivity only by less than 20%. For providing 7497 FIG. 3. Effect of the opening of the superconducting energy gap on the electrothermal conductivity below T c . The ratio @ S n (T)/ r n (T) # u DÞ0 / @ S n (T)/ r n (T) # u D50 obtained from Eq. ~11! plotted vs T/T c ~solid line!. An additional factor of ( j / Aj 2 2D 2 ) in the corresponding integral for the electrothermal conductivity ~as discussed in the text! leads to the dotted line, whereas multiplication with the factor (T c /T) 4 results in the dashed curve. The open circles show the extrapolated values needed to yield the expected thermal Hall angle a '0.01 based on Eqs. ~5! and ~6!. additional insight we compare this result with the following modified calculations. We assume a superconductor with a geometrically limited mean free path l 0 ~e.g., due to impurities!. Here the reduction of the group velocity v qp 5 v F ( Aj 2 2D 2 / j ) is compensated by an increase of the scattering time t qp5 t F ( j / Aj 2 2D 2 ), where v F is the Fermi velocity, t F 5l 0 / v F , and j 5«2 m . 15 With the quasiparticle density of states N( j )5N 0 ( j / Aj 2 2D 2 ), this introduces an additional factor of ( j / Aj 2 2D 2 ) in the corresponding integral for the electrothermal conductivity. However, this divergent factor only leads to an enhancement of S n~ T ! r n~ T ! D/k B T. line! and plotted vs DÞ0 S n~ T ! r n~ T ! D50 of about 40% at T/T c 50.8, as is shown by the dotted line in Fig. 3. We conclude that a substantial increase of the electrothermal conductivity below T c has to involve a considerable reduction of the scattering rate for quasiparticles at higher energies. Since electron-electron scattering appears to be the dominating process in high-T c materials, the scattering rate is expected to be proportional to the quasiparticle concentration n qp . Not too far below T c , n qp of a superconductor with an isotropic gap is well approximated by the corresponding expression of the two-fluid model,16 yielding n qp;(T/T c ) 4 . Hence the opening of the energy gap at temperatures below T c is expected to introduce the additional factor (T c /T) 4 into the normalized quantity S n~ T ! r n~ T ! 1D/k T FIG. 2. $ 12 * 2D/k B T dz @ 1/(11e z )(11e 2z ) # % ~solid B k 1D T $ 123/p 2 * 2D/kB T dz @ z 2 /(11e z )(11e 2 ) # % ~dashed line! B U Y U U Y U DÞ0 S n~ T ! r n~ T ! , D50 derived from Eq. ~11!, because of the reduced quasiparticle scattering. This curve is shown in Fig. 3 by the dashed line. We emphasize that it is mainly this effect of the opening of the superconducting energy gap on the electrothermal conductivity below T c which is not taken into account by 7498 BRIEF REPORTS 57 simply extrapolating this quantity from above T c . Numerical calculations on the electron-electron scattering in the cuprates below T c have been reported by Rieck et al.17 In our discussion we have not yet included the vortex structure present in the mixed state. The experiments mentioned in Refs. 10–12 also deal only with the case B50. Turning now to the case BÞ0 ~with B oriented parallel to the c axis!, we note that in YBa2Cu3O7 the vortex core is close to the superclean limit.18,19 In this limit the electronic structure of the vortex core is given by the discreet levels of the Andreev bound states, with the level spacing D« 5D 2 /« F being larger than the level smearing d «5\/ t due to scattering ~t 5quasiparticle scattering time!. In YBa2Cu3O7 it appears that only a single Andreev bound state exists in the vortex core.20,21 Therefore, the quasiparticle scattering is not much affected by the vortex cores, and our arguments given above for the case B50 can be extended also to the case BÞ0 without introducing a large error. However, this approach will be restricted to the low-field limit B!B c2 . The (T/T c ) 4 factor reduces the quasiparticle scattering rate at 0.8T c by a factor of 2.5 compared to its value at T c . This appears consistent with the other experimental data for YBa2Cu3O72d. 10–13 The temperature-dependent electrother- mal conductivity displayed by the dashed curve in Fig. 3 also removes the discrepancy of the thermal Hall angle a discussed above. In Fig. 3 we show for six temperatures below T c ~open circles! the factor by which the quantity S n (T)/ r n (T) must be enhanced in order to obtain the expected value a '0.01 using Eqs. ~5! and ~6!. These values were calculated from the experimental data presented in Fig. 3~b! of Ref. 2, and they are close to the dashed line in Fig. 3. We conclude that our simplified treatment leads to a reasonable understanding of the Seebeck and Nernst effect data in the mixed state of YBa2Cu3O72d, without envoking an unusually large value of the thermal Hall angle. Finally, we emphasize that our discussion given above has been highly simplified, only pointing out the main aspects of the underlying physics. In particular, we have ignored implications of the possible d-wave symmetry of the order parameter in the cuprates. A detailed theoretical analysis of these questions needs to be done and remains an interesting task. R. P. Huebener, Supercond. Sci. Technol. 8, 189 ~1995!. H.-C. Ri, F. Kober, A. Beck, L. Alff, R. Gross, and R. P. Huebener, Phys. Rev. B 47, 12 312 ~1993!. 3 H.-C. Ri, R. Gross, F. Gollnik, A. Beck, R. P. Huebener, P. Wagner, and H. Adrian, Phys. Rev. B 50, 3312 ~1994!. 4 K. Maki, Prog. Theor. Phys. 41, 902 ~1969!; J. Low Temp. Phys. 1, 45 ~1969!. 5 A. V. Samoilov, A. A. Yurgens, and N. K. Zavaritsky, Phys. Rev. B 46, 6643 ~1992!. 6 J Bardeen and M. J. Stephen, Phys. Rev. 140, A1197 ~1965!. 7 H.-C. Ri, F. Kober, R. Gross, R. P. Huebener, and A. Gupta, Phys. Rev. B 43, 13 739 ~1991!. 8 D. M. Newns, H. R. Krishnamurthy, P. C. Pattnaik, C. C. Tsuei, C. C. Chi, and C. L. Kane, Physica B 186–188, 801 ~1993!. 9 J. Ruvalds, Supercond. Sci. Technol. 9, 905 ~1996!. 10 D. A. Bonn, Ruixing Liang, T. M. Riseman, D. J. Baar, D. C. Morgan, Kuan Zhang, P. Dosanjh, T. L. Duty, A. MacFarlane, G. D. Morris, J. H. Brewer, W. N. Hardy, C. Kallin, and A. J. Berlinsky, Phys. Rev. B 47, 11 314 ~1993!. 11 F. Gao, J. W. Kruse, C. E. Platt, M. Feng, and M. V. Klein, Appl. Phys. Lett. 63, 2274 ~1993!. 12 1 2 ACKNOWLEDGMENT The authors are grateful to N. Schopohl for helpful discussions. R. C. Yu, M. B. Salamon, Jian Ping Lu, and W. C. Lee, Phys. Rev. Lett. 69, 1431 ~1992!. 13 K. Krishana, J. M. Harris, and N. P. Ong, Phys. Rev. Lett. 75, 3529 ~1995!. 14 N. W. Ashcroft and N. D. Mermin, Solid State Physics ~Holt, Rinehart and Winston, New York, 1976!, p. 254. 15 L. P. Gor’kov and N. B. Kopnin, Sov. Phys. Usp. 18, 496 ~1976!. 16 M. Tinkham, Introduction to Superconductivity ~McGraw-Hill, New York, 1996!. 17 C. T. Rieck, W. A. Little, J. Ruvalds, and A. Virosztek, Phys. Rev. B 51, 3772 ~1995!. 18 Y. Matsuda, N. P. Ong, Y. F. Yan, J. M. Harris, and J. B. Peterson, Phys. Rev. B 49, 4380 ~1994!. 19 M. Golosovsky, M. Tsindlekht, and D. Davidov, Supercond. Sci. Technol. 9, 1 ~1996!. 20 K. Karrai, E. J. Choi, F. Dunmore, S. Liu, H. D. Drew, Qi Li, D. B. Fenner, Y. D. Zhu, and Fu-Chun Zhang, Phys. Rev. Lett. 69, 152 ~1992!. 21 I. Maggio-Aprile, Ch. Renner, A. Erb, E. Walker, and O ” . Fischer, Phys. Rev. Lett. 75, 2754 ~1995!.