ON INTRINSIC AND EXTRINSIC EFFECTS IN THE SURFACE IMPEDANCE OF CUPRATE SUPERCONDUCTORS

Surface impedance measurements in the normal and superconducting state are an excellent method to study conduction electron dynamics and extended defects. Electron dynamics show up most clearly in the relaxation range, i.e., for distances traveled in one rf period s = upsilon(F)/omega (upsilon(F) Fermi velocity) being smaller or of the order of the penetration depth lambda and mean free path l. For materials with upsilon(F) less-than-or-equal-to 10(7) cm/sec the relaxation range is easily accessible for f greater-than-or equal-to 0.1 THz. Then, in the normal state, relaxation defines the surface impedance with an intrinsic penetration depth lambda(I) approaching the London penetration depth lambda(L) and R(I) almost-equal-to mu0lambda(L)/2tau as surface resistance, allowing measurement of lambda(L) and relaxation time tau(T, omega). In the superconducting state the photon interaction scales with xi(F)/lambda(L) = 1/gamma (xi(F) dimension of Cooper pairs for l --> infinity) and causes at low frequencies an absorption rate growing with gamma, which is decreasing with xi(F)/l. The rate increase proportional to gamma turns to a decrease above 0.1 THz, which is accompanied by a decrease of lambda with frequency which is stronger for large gamma and small xi(F)/l. These characteristic dependences allow measurement of material parameters, anisotropy, and dynamics of electrons, especially the relaxation rate tau. But presently, the rf surface impedance Z is still shrinking with material improvements, which shows, clearly, that the Z = Z(I) + Z(res) is still dominated by extrinsic properties summarized in Z(res). We present evidence that Z(res) is due to the large leakage current j(bl) and the small j(cJ) of weak links where the latter destroys the intrinsic shielding from a lambda(I)-thin seam lambda(J) deep into the bulk. This causes rf residual losses R(res) almost-equal-to (omegamu0)2lambdaj3sigma(bl)/2. R(res) stays finite at T congruent-to 0 due to sigma(bl)(T-->0) almost-equal-to sigma(bl) (is-proportional-to j(bl) being amplified by (lambda(J)/lambda(I)3 > 10(3) as a weighting factor. The appropriate measure of weak links are the grain-boundary resistance R(bn)(is-proportional-to rho(0)) enhancing lambda(J) is-proportional-to R(bn) and R(res) is-proportional-to R(bn)2. Thus, Z(res) is minimal for minimal extrapolated resistivity rho(T --> 0). To identify the weak links as a new entity, the H-field dependence is most helpful, because at very low fields H(c1J)is-proportional-to 1/lambda(J) Josephson fluxons penetrate into the weak links. These Josephson fluxons show negligible flux flow or flux creep, and enhance Z(res) by lambda(J)(H, T) is-proportional-to 1/square-root j(cJ)(H, T). The measured j(cJ)(H, T) and j(bl) values explain Z(res) quantitatively as well as in temperature is-proportional-to (a + T) (n almost-equal-to 1, T < T(c)/2) and in field is-proportional-to (b + H(n)) (almost-equal-to 1, H > H(c1J)) dependence. The strength of the field dependence dZ(res)/dH is-proportional-to Z(res)(H(c1J)*)/H(c2J)(T) is not only a measure of Z(res) and H(c2J)(T) but is crucial for nonlinear effects and (fluxon) noise also, which limit the performance of rf devices.