Supercond. Sci. Technol. 12 (1999) 682–689. Printed in the UK PII: S0953-2048(99)04672-2 Flux creep in a thin disc of YBa2Cu3O7−δ superconductor E Moraitakis, M Pissas, G Kallias and D Niarchos Institute of Materials Science, National Centre for Scientific Research, ‘Demokritos’, 153 10 Aghia Paraskevi, Athens, Greece Received 1 June 1999, in final form 23 July 1999 Abstract. We investigated the flux-creep behaviour of a thin disc of YBa2 Cu3 O7−δ superconductor in the framework of an analytical model, in the temperature range T = 10–60 K and for a perpendicular applied field Ha = 5–50 kG. The relaxation of the magnetic moment shows a logarithmic dependence on time for the whole temperature range. The field and temperature dependences of the critical current jc and the pinning potential U0 can be derived accurately by fitting the relaxation data. The results are discussed within collective pinning theory. The critical current jc shows a power law field dependence of the form jc (H ) ∝ H −b with b ' 0.50, which is kept during the relaxation process. The E versus j characteristics are quantitatively derived from the relaxation data at the circumference of the disc and show a very steep E(j ) relation described by a power law E(j ) = Ec (j/jc )n for the whole temperature range, while a change of this behaviour is observed for T = 60 K. The activation barrier U (j, T = 0, H ) is derived with the application of Maley’s method based on the exact relations for a thin disc and this gives an exponent µ ' 0.68 for H = 20 kG. 1. Introduction Thermal activation of vortices in high-Tc superconductors (HTSs) yields to strong relaxation of the irreversible magnetization limiting the current-carrying capacity of these materials. The macroscopic electrodynamic properties of HTS specimens become sensitive to the highly nonlinear current–voltage characteristics (E versus j relation) in the subcritical region (j < jc ), determined by microscopic mechanisms of flux dynamics and pinning [1]. The case of a superconducting thin film in a perpendicular magnetic field exhibits novel features as compared with the well-known case of a slab in a parallel field (parallel case) [2]. Static analytical calculations [3–10] were carried out and numerical solutions [11–15] for the current j (r ) and magnetic field B (r ) were also obtained for a variety of sample geometries in a perpendicular applied field. Flux dynamics in thin films in a perpendicular field has attracted much interest and a detailed theoretical description has been given for the linear E versus j relation (Ohmic case) which holds in both the flux-flow regime and the TAFF regime observed for low currents at temperatures above the irreversibility line [16]. Also, a detailed analysis has been given for the general nonlinear E versus j relation, well below the irreversibility line [17, 18]. In the latter case, the formulation developed in [17] permits a detailed description of a thin strip or disc. It was shown that the flux diffusion is one-dimensional, nonlinear and nonlocal and it is described by an integral equation which can be solved analytically. Exact relations for the relaxing electric field, for the relaxing current density profiles and for the relaxing magnetic moment are derived. Furthermore, the magnetic moment exhibits the 0953-2048/99/100682+08$30.00 © 1999 IOP Publishing Ltd well-known logarithmic dependence on time for long-enough times relative to characteristic ones. This formulation was generalized in [18] in order to treat long superconductors of arbitrary cross section with arbitrary E versus j relation. It must be noted that most of the relaxation experiments have made use of rectangular specimens. However, in this case the patterns of the sheet current and of the flux density have some peculiar features (especially at low fields) relative to the one-dimensional distributions in circular discs and strips that are essential in order to understand the flux dynamics in superconductors [14]. In this paper magnetic relaxation measurements taken for different magnetic fields in a thin disk of YBa2 Cu3 O7−δ superconductor are presented and analysed in the framework of [17]. The use of a specific (disc) geometry and an analytical model which describes magnetic relaxation in this case permits the accurate determination of the critical current density jc and the pinning potential U0 . We also derive the E versus j characteristics from the relaxation data in a quantitative way and apply the method of Maley and Willis [19] for the determination of the activation barrier U (j, T = 0, H ), based on the exact relations for a thin disc. 2. Analysis of relaxation measurements In the following we summarize the main topics from [17] and derive analytical relations for the case of a disc in a perpendicular field to be used for the analysis of our data. For an HTS, thermal activation of vortices induces an electric field which can be written in the form E = Ec exp[−U (j )/kT ] (1) Flux creep in a thin disc of YBa2 Cu3 O7−δ where U (j ) is a flux-creep potential barrier that vanishes at j = jc and Ec is a crossover electric field between the fluxflow and flux-creep regimes which defines the critical current jc by E(jc ) = Ec . The above relation determines the E versus j characteristics in the subcritical region j < jc which together with Maxwell equations completely determine the electrodynamic properties of HTS specimens [8, 17, 18]. For Anderson–Kim flux creep [20] the potential barrier is linearized as U (j ) = U0 (1−j/jc ) and leads to an exponential E versus j relation: E = Ec exp[−U0 (1 − j/jc )/kT ]. For the vortex glass and collective creep models [1, 21, 22] the potential barrier U (j ) is a highly nonlinear function of j and is generally described by the so-called interpolation formula U (j ) = (U0 /µ)[(jc /j )µ − 1], where the exponent µ depends on the dimensionality and the particular fluxcreep regime [23]. In the case of a 3D collective creep model the values µ = 1/7, 5/2 and 7/9 are proposed for pinning of single vortices, small flux bundles and large flux bundles respectively [1, 21]. In the case of vortex glass model 0 < µ 6 1 holds [22]. For the case of a superconducting thin disc under a perpendicular magnetic field the formulation which was developed in [17] gives a onedimensional, nonlinear and nonlocal flux diffusion which is described by the integral equation (in Gaussian units) 1 ∂(rE) 1 ∂Ha =− r ∂r Z c ∂t 2 a ∂J ∂E(u, t) E(k) K(k) − 2 − du (2) c 0 ∂E ∂t r −u r +u where E(k) and K(k) are complete elliptic integrals, Rd k = (4ru)1/2 /(r + u) and J (r) = 0 j (r, z) dz = j (r)d is the sheet current density flowing in the disc plane. However, since the ratio kT /U0 1 well below the irreversibility line, we can approximate ∂j/∂E =j1 /E = (kT /U0 )jc /E over a wide region of E values except for exponentially small fields E < Eg ≈ Ec exp[−U0 /(1 + µ)kT ] Ec . In this case, after some transient time τ , the relaxing electric field E(r , t) = f (r )g(t) separates into a universal profile f (r ) and a time dependence g(t) = 1/(t + τ ). The relaxing current density profile is then obtained by inserting this general electric field into the material law j = j (E) and it is approximated by j (r, t) = jc − j1 ln[Ec /E(r, t)]. In this way the relaxing magnetic R a moment of the disc can be extracted: m(t) = (π/c) 0 r 2 J (r, t) dr, where J (r, t) = j (r, t)d is the sheet current in the disc. This formulation is exact for the Anderson–Kim linearized flux-creep barrier U (j ) = U0 (1 − j/jc ) but holds equally well for a sufficiently nonlinear E(j ) relation as it was described above. Only for the Ohmic case (flux-flow regime) or the TAFF regime [24] at j j1 does E(j ) become linear (E = ρj ), and the profiles of E and j attain a different, also universal form and decay exponentially [16]. When the applied magnetic field increases with constant ramp rate R = dHa /dt, the first term in equation (2) dominates and induces a steady-state electric field: E(r) = −Rr/2c. (3) After the applied magnetic field is stabilized at t = 0, the first term in equation (2) vanishes and the electric field E(r, t) begins to decay with time, owing to the nonzero resistivity of a superconductor for j < jc , caused by thermal activation of vortices. The exact solution for the case of disc of radius a is written in the form E(r, t) = 2j1 ad fdisc (η) c2 (t + τ ) (4) where fdisc (n) is a universal function defined in [17]. The transient period τ during which the relaxation of E(r, t) depends on the initial condition equation (3) can be determined by equating m(t) for t = 0 to the unrelaxed (initial) magnetic moment Z π a J (r, 0)r 2 dr m(0) = c 0 S Ra 1 π a3 Jc 1 + S ln (5) − = c 3 2cEc 3 where Jc = jc d is the sheet current, S = kT /U0 and R = dHa /dt is the field ramp rate. The result is τ = 2.568J1 /cR, where J1 = (kT /U0 )jc d. We finally obtain m(t) = m(0) − m1 ln(1 + t/τ ) (6) where m(0) is the initial magnetic moment from equation (5) and m1 = π a 3 jc dS/3c. For t τ (regime of steady-state relaxation), the time τ in equation (6) cancels out and thus m(t) becomes independent of the initial conditions, as can be easily verified. The result is m(t) = mc − m1 ln(t/t0 ) (7) where t0 = 0.920jc daS/c2 Ec , mc = π a 3 jc d/3c and m1 = π a 3 jc dS/3c. The above logarithmic expression is exact for the whole time window for the Anderson–Kim flux creep. The problem of magnetic relaxation in a thin disc, in terms of the current–voltage characteristics and the initial conditions, was also treated appropriately by Zhukov et al [25, 26]. A similar approach was also applied in the case of an infinite plate in a parallel field (longitudinal geometry) [27]. A universal electric field profile E(r , t) was also found in this case and the initial stage as well as the long-time stage of the relaxation of the magnetization were discriminated and observed experimentally. Now both jc and t0 depend on the electric field criterion Ec . Notice that instead of jc and Ec one can take another pair jc0 and Ec0 related to the previous pair by jc0 = jc − j1 ln(Ec /Ec0 ). This transformation does not change equation (7), provided that the quantity mc + m1 ln t0 remains constant [27]. In other words the above transformation links two critical current densities jc and jc0 defined at different electric field criteria. For t t0 one can write down the interpolation formula for the whole process of magnetic moment relaxation [1, 21]. This formulation clears out the question about the origin of ‘microscopic’ times t0 , τ in the relaxation process. Both of them depend on the material’s parameters but τ depends also on the initial conditions R = dHa /dt, through well-defined analytical relations for the case of a disc. 683 E Moraitakis et al 3. Experimental details High-quality thin films of YBCO superconductor were deposited with a simple sputtering technique as reported elsewhere [28]. The films were grown epitaxially on LaAlO3 (100) substrates with the c-axis normal to the substrate and exhibited very good superconducting properties (Tc ∼ 89–92 K, 1Tc ∼ 1–2 K, Jc (77 K) > 1×106 A cm−2 ). A thin YBCO disc with a diameter of 5 mm and a thickness of 400 nm was patterned using standard photolithographic–wet etching techniques. No degradation of Tc ' 90.2 K, 1Tc ' 1.2 K, was measured after the patterning by an ac susceptibility technique. Magnetic measurements were performed with a commercial SQUID magnetometer (Quantum Design MPMSII), applying a magnetic field perpendicular to the film surface (parallel to the c-axis of the film). A scan length of 2 cm was chosen in order to avoid magnetic field inhomogeneities during the measurement scan. In all the relaxation measurements the crystal was cooled in zero field, and then the magnetic field was increased to the value Ha with a ramp rate R = dHa /dt which depends on the field value. Then the field was fixed (initial time t = 0) and the relaxation of m(t) was measured within the time window ti ' 102 s 6 t 6 tf ' 104 s. Hysteresis loops were collected for the YBCO disc up to a maximum field of 50 kOe, with different sweep rates R = dHa /dt, the field increment 1Ha being varied from 200 to 10 000 Oe. 4. Results and discussion Figure 1 is a semilogarithmic plot of the variation of the magnetic moment with time at T = 10 K, with applied field Ha k c-axis and for the field range Ha = 5–50 kOe. The main characteristic of the measurements is the linear variation of m(t) in the semilogarithmic plot for all the magnetic fields. Within experimental accuracy the same logarithmic time dependence of m(t) is also observed for the whole temperature range T = 10–60 K, examined in this work. The data were fitted using equation (7) which describes the steady-state relaxation. Fits are shown by straight lines in figure 1. The following expression was used: m(t) = −P1 + P2 ln(t P3 /P2 ), where P1 = −mc , P2 = −m1 are free parameters and P3 = cEc (πa 2 /2)/0.920 = 10.674 by fixing the electric field criterion at Ec = 1 µV cm−1 . Such a procedure was also used in order to analyse the magnetic relaxation measurements of an HgBa 2 CuO4+δ single crystal [29]. Therefore the critical current jc and pinning potential U0 are calculated from the fitting parameters. Different values of Ec = 1–103 µV cm−1 change the estimated jc , U0 values by no more than 7% owing to the logarithmic relation between Ec and jc . This means that the jc values are well defined even if Ec was arbitrarily chosen. The estimated jc values must be considered as extrapolated ones at t = t0 , assuming the same U (j ) relation, i.e. without any crossover from one pinning regime to another. Furthermore, from the estimated values jc = 1.534×107 A cm−2 and U0 = 625.2 K for Ha = 10 kOe and R = 90.9 Oe s−1 , we estimate τ ' 0.028 s and t0 ' 0.0023 s. Since the time for the first measurement is ti ' 74 s τ , we cannot observe the initial transient regime 0 < t < τ in our measurements, 684 described by equation (6). We observe only the steadystate relaxation regime described by equation (7), in which the nonlinearity of U (j ) manifests itself in the logarithmic relaxation of m(t). Furthermore, nonlogarithmic relaxation of m(t) described by the so-called interpolation formula has been observed in HTSs for very-long-time (of the order of days) relaxation measurements [30]. In fact this deviation from the logarithmic relaxation can be obtained in the above model using the exact relation for ∂j/∂E in the integrand of equation (2) [17]. Figure 2 shows the dependence of the estimated critical current jc values on the applied magnetic field at T = 10 K. In the same plot the current density j calculated from the relaxed magnetic moment at t = 103 s is shown. Both can be fitted better using a power law relation of the form j (B) = aH −b with b ' 0.48–0.49, in the examined field range from 5 to 50 kG. This means that the same field dependence of the critical current jc is kept during the relaxation process. Furthermore, the same field dependence of the critical current with b ' 0.47–0.53 has been obtained for different temperatures in the range T = 10–60 K. Figure 2 shows also the variation of the estimated U0 with the applied magnetic field at T = 10 K. The estimated values do not show any significant dependence on the magnetic field in the field range from 5 to 50 kG. Based on the above results the relaxation rate dm/d ln t = (kT /U0 )mc as a function of the magnetic field decreases as the field increases mainly as a result of the decrease of jc . A power law with b ' 0.3 was also found by Griessen et al [31] to describe the field dependence of the estimated critical current jc up to B = 70 kOe in YBa2 Cu3 O7−δ thin films. This much weaker field dependence than the exponential decay jc ∝ exp(−B 3/2 ) or jc ∝ B −3 predicted by theory [1, 21] in the small-bundle and large-bundle regimes was considered as further evidence in favour of the singlevortex regime. However, one expects that the single-vortex depinning current density does not depend on the magnetic field. Figure 3 shows the variation of the estimated jc values with temperature. Full lines represent fits with an exponential law jc (T ) = jc (0) exp(−T /T0 ), with about the same values of T0 ' 25 and 24 K for H = 20 and 40 kG respectively. This apparent linear variation of ln jc versus T was also observed by other authors [32, 33] for the examined temperature range, and it may be considered as a manifestation of collective pinning and creep effects. It must be noted that quantum creep effects which are important at low temperature do not influence the data in the examined temperature range T = 10–60 K, as pointed out in [31, 32]. According to collective pinning theory, the effect of pinning by randomly distributed weak pinning centres leads to a finite correlation length Lc parallel to the external magnetic field which is aligned in the c-axis direction, in our case. The appropriate length is given by [1] Lcc ∼ = εξab j0 jc 1/2 (8) where the anisotropy ratio is ε = (mab /mc )1/2 and the depairing current is j0 = c80 /12 √ Ginzburg–Landau 2 3π 2 λ2 ξ with λ2 = λ2L /2, ξ 2 = 0.54ξBCS (clean limit) Flux creep in a thin disc of YBa2 Cu3 O7−δ 50 -0.4 40 30 -0.5 20 m (emu) -0.6 -0.7 -0.8 10 -0.9 -1.0 -1.1 5 10 2 10 3 10 4 t (sec) Figure 1. Semilogarithmic plot of the relaxation of the magnetic moment versus time for the YBCO thin disc at T = 10 K and for perpendicular magnetic fields Ha = 5, 10, 20, 30, 40 and 50 kG. Full lines are fits to the data as explained in the text. 800 2.0 600 1.0 U 0(K) 1.5 7 2 J c (10 A/cm ) 700 500 0.5 400 0 10 20 30 40 50 H(kG) Figure 2. Dependence of the critical current density on the applied field at T = 10 K. The estimated critical current values are represented by full squares and the relaxed ones at t = 1000 s by open squares. In addition, the current density values calculated from the hysteretic moment (Bean model) for sweep rate R ≈ 20 Oe s−1 are shown by crosses. The same power law dependence j (H ) ≈ H −b with b = 0.48–0.50 was found to fit the data better, as shown by full curves. The estimated U0 (right-hand axis) as a function of the applied magnetic field is shown by open circles. The dotted curve is a guide for the eye. The error bars in both cases are typically much less than the size of the centred symbols. the characteristic lengths at T = 0. Taking the anisotropy ratio ε ' 1/6, the corresponding coherence length ξBCS = ξab (0) ' 15 Å, the penetration depth λL = λab (0) ' 1400 Å and the estimated critical current from the extrapolation of jc (T → 0) ≈ 1.1 × 107 A cm−2 for H = 40 kG, one obtains Lcc (T = 0) ≈ 21 Å. This value is higher than the interlayer distance of CuO2 layers d ' 12 Å and satisfies d < Lcc < εa0 ' 36 Å for H = 40 kG, where a0 = (80 /B)1/2 is the measure of flux line lattice cell constant, which indicates single-vortex line pinning. This value is very close to Lcc ≈ 28 Å derived by Wen et al [32] for YBCO thin films, while Thompson et al [30] derived Lcc ≈ 14 Å for proton-irradiated YBCO crystals and Civale et al [34] derived Lcc ≈ 40 Å for ‘clean’ YBCO crystals. Furthermore, the crossover field at T = 0 between single-vortex and small-bundle regimes is given by [1] Bsb = βsb Hc2 (jc /j0 ), with βsb ≈ 5, and this gives a value of Bsb ' 8.4 T for Hc2 ≈ 120 T in our case. As the temperature increases thermal fluctuations of vortices lead to a smoothing of the disorder potential and thereby pinning will be reduced. This phenomenon is known as thermal depinning, and it must be seen not as an abrupt transition but rather as a continuous crossover from a pinned to an unpinned situation. The extent of the single-vortex pinning regime is limited by the single-vortex depinning s which is given by (weak pinning condition) temperature Tdp s Tdp ≈ 0.7(jc /Gi j0 )1/2 Tc where Gi ≈ 10−2 is the Ginzburg s ≈ 73 K using the number for YBCO [1]. We estimated Tdp values of jc and j0 defined above. For parameters typical for YBCO, Blatter et al estimated that the single-vortex regime is realized within the B–T plane for B < Bsb ≈ 6 T and s ≈ 60 K [1]. Our relaxation measurements for T < Tdp fall in this field and temperature range, and the above results may be in favour of this picture, although the possibility of a vortex glass state for T < 60 K cannot be excluded. It should also be noted that the examined field and temperature range is below the irreversibility line which is defined primarily by the deviation from reversible to irreversible magnetic behaviour. Our measurements of the irreversibility line from the onset of irreversibility in hysteresis loops for the YBCO disc give a power law dependence Hirr ∝ (1 − T /Tc )3/2 which gives an estimated value Hirr (T = 60 K) ' 6.6 T. The glass transition describing a second-order transition from a vortex glass to a vortex liquid phase is higher than the examined field and temperature range. A scaling approach of E versus j measurements according to the vortex glass model for YBCO films gives a transition field BG (70 K) ' 5 T, as shown by other authors [35, 36]. Figure 4 shows the variation of the estimated U0 values with temperature for fields H = 20 and 40 kG. In the same plot the variation of the so-called normalized relaxation rate S = d ln m/d ln t = (1/m) dm/d ln t, calculated at t = 120 s, is also shown. The temperature dependence of S steeply increases after a plateau. The plateau values S = 0.025–0.030 fall in the ‘universal’ plateau range observed and discussed by Malozemoff and Fisher [23, 37]. The estimated pinning potential U0 shows a maximum at the temperature where the normalized relaxation rate S = d ln m/d ln t starts to increase steeply, which shifts to lower temperatures for higher fields. It is expected that for temperatures higher than 60 K the pinning potential rapidly decreases as the critical temperature Tc is 685 E Moraitakis et al H =20 kG H =40 kG 7 2 Jc(A/cm ) 10 10 6 10 20 30 40 50 60 T(K) Figure 3. Variation of the critical current density with temperature. The estimated values are shown by full and open squares for fields H = 20 and 40 kG respectively. Full lines represent fits with an exponential law jc = jc (0) exp(−T /T0 ) with T0 ≈ 25 and 24 K respectively. H=20 kG 2500 H=40 kG U0(K) 2000 1500 1000 500 a 10 S=-(1/m)*(dm/dlnt) 0.07 20 30 40 50 60 30 40 50 60 H=20 kG H=40 kG 0.06 0.05 0.04 0.03 0.02 b 10 20 T(K) Figure 4. (a) The variation of the estimated U0 with temperature for the YBCO thin disc. The estimated values are shown by full and open squares for the fields H = 20 and 40 kG respectively. The errors bars are typically much less than the size of the centred symbols. (b) The variation of the normalized relaxation rate S = −(1/m) dm/d ln t at t = 120 s. reached. The same behaviour was also observed for melttextured YBCO samples [38]. Similar behaviour was also observed by Wen et al [32] for a dynamical relaxation rate Q = d ln js /d ln(dB/dt) in YBa2 Cu3 O7−δ and YBa2 Cu4 O8 thin films. 686 m(t) = 1 π a3 j (a, t)d c 3 (9) where j (a, t) = j (t) = jc d[1 − (kT /U0 ) ln(t/t0 )] as deduced by direct comparison of equation (9) with equation (7). The electric field at the circumference of the disc for t τ is calculated using equation (4) for t τ and dm/dt = −m1 /t as E(a, t) = 3000 0.01 From magnetic relaxation measurements, E(j ) characteristics can be obtained for the disc geometry, using the analysis of section 2. A sufficiently nonlinear E(j ) relation gives for t τ a nearly uniform current density profile in the disc (except for a small region near the centre of the disc) which can be approximated in most of the disc’s area by a constant value, corresponding to the value j (a, t) at the circumference of the disc [17, 18]. This value is correlated with the hysteretic magnetic moment of the disc in the full flux-penetrated state (Ha > Hp = 2πjc d/c) according to 6 fdisc (1) dm cπ a 2 dt (10) where fdisc (1) can be approximated with fdisc (0.999 989) ≈0.832 046 [17]. In this way the E(j ) relation at the circumference of the disc r = a can be derived. Magnetic relaxation measurements in a well-defined geometry probe the E(j ) characteristics at much lower electric field levels (of the order of 10−13 V cm−1 ) than those experimentally accessible in transport measurements. The results for T = 10 K are shown in a double-logarithmic plot in figure 5. The apparent linearity in the log E versus log j plots means that the E versus j relation can be effectively described by a power law dependence. In fact, it can be shown that the interpolation formula for U (j ) yields for µ 1 a logarithmic expression U (j ) ≈ U0 ln(jc /j ). In this case the E versus j relation is E(j ) ≈ Ec (j/jc )n with n = U0 /kT [8, 39]. This power law E(j ) relation may be considered as a general one which describes different regimes from Ohmic (n = 1) to Bean-like behaviour (n → ∞) [8, 18]. The full lines in figure 5 represent power law fits using the above relation. In the fitting procedure the jc values were fixed to those obtained from the relaxation fits and Ec , n were the free parameters. Successful fits were obtained with values Ec = 5 × 10−7 –1 × 10−6 V cm−1 and n = 49–52 as shown in the inset of figure 5. Such a very steep power law E(j ) relation must be considered actually as an exponential one (given by the same expression as for Anderson–Kim flux creep), which for the narrow interval of current j spanned during the relaxation appears as a line in log E–log j scale. This behaviour is consistent with the assumptions of the theoretical model and the very high values U0 /kT ≈ 60 estimated from the relaxation data. The narrow interval of current density j spanned during the magnetic relaxation in our case does not permit us to test the E versus j characteristics at higher current density j or electric field E levels. Power law E versus j characteristics from magnetic relaxation and field-sweep measurements, in YBCO discs and rings, were also obtained by other authors [26, 40]. In [41] E versus j curves extracted from magnetic relaxation measurements on a YBCO ring have been combined with direct I –V measurements to allow quantitative analysis of Flux creep in a thin disc of YBa2 Cu3 O7−δ 10 -9 H=20 KOe 55 n 50 10 10 -11 10 -12 10 -12 10 -13 -13 4x10 20H(kG) 40 E (V/cm) E (V/cm) -11 45 0 10 10 -10 10 50 40 30 20 50K 45K 40K 60K 5 10 6 2 10 7 10K 10 7 2 Figure 5. Double-logarithmic plots of the E versus j characteristics at the circumference of the disc, estimated from the relaxation data at T = 10 K. The apparent linearity means that a power law relation E(j ) = Ec (j/jc )n fits the data better, as shown by full lines. The inset shows the variation of the exponent n with field. data spanning more than 10 decades in electric field. The data were consistent with a current-density-dependent pinning energy given from the interpolation formula with an exponent µ ≈ 1/3 at T = 70 K. Figure 6 shows the estimated E(j ) characteristics at the circumference of the disc for different temperatures and for a field H = 20 kG. Deviation from the linear variation behaviour in the log E versus log j plots is observed for T = 60 K. It must be taken into account that this temperature is close to the single vortex–small bundle s boundary which is estimated to be Tdp ≈ 60 K [1]. As pointed out by Maley and Willis [19], the true shape of the activation barrier for vortex motion U (j, T = 0, H ) can be determined from the relaxation data. In the following we derive an equation for U (j, T = 0, H ) based on the exact relations for a thin disc. The E versus j relation at the circumference of the disc (r = a) can be written using equation (1) as (11) Using equation (10) and equation (9) we obtain E(a, t) =0.416(4ad/c2 ) dj (a, t)/dt. We finally obtain for a thin disc U (j, T = 0, H )/k = −T [ln(dj/dt) − C]g −1 (T ) 6 20K J (A/cm ) J (A/cm ) U [j (a, t)]/k = −T [ln E(a, t) − ln Ec ]. 30K (12) where C = ln(Ec c2 /1.664ad) and j ≡ j (a, t). The crossover electric field between flux flow and creep can be estimated from Ec ≈ (1/c)H ν0 X, where ν0 and X are a characteristic attempt frequency and hopping distance respectively. The function g(T ) contains the separated temperature dependence of the activation energy Figure 6. Double-logarithmic plots of the estimated E versus j characteristics at the circumference of the disc for H = 20 kG and for different temperatures in the range T = 10–60 K. Deviation from linearity is observed for T = 60 K. U (j, T , H ) = U (j, T = 0, H )g(T ). We tried the functions g(T ) = 1 − (T /Tc )2 suggested by Tinkham [38, 42, 43] and g(T ) = [1 − (T /Tc )2 ]1.5 used by others [44, 45]; the best results were found for the second one for the determination of U (j, T = 0, H ). First, one has to plot the second part of equation (12) as a function of j for the relaxation data at different temperatures and, secondly, the unknown parameter C has to be determined by the condition that all points have to lie on one smooth curve. Rough estimations of the parameter C for different Ec values are C ' 19 and 26 for Ec = 1 and 1000 µV cm−1 respectively. Figure 7 shows the U (j, T = 0, H = 20 kG) relation determined from the relaxation data at different temperatures by using the function g(T ) = [1 − (T /Tc )2 ]1.5 ; a variation of the C value changes slightly the obtained U (j ) curves as shown in the inset of figure 7. The value C = 30 was found to put the data better on the same curve with a slight deviation of the data at T = 60 K. The full curve is a fit of the multiple data sets with the interpolation formula U (j ) = (U0 /µ)[(jc /j )µ − 1]. In the fitting procedure the value jc (H = 20 kG, T → 0 K) ' 1.6 × 107 A cm−2 from the extrapolation of the jc (T ) curve in figure 3 was fixed, while U0 , µ were the free parameters. Our estimated values U0 ' 342 K and µ ' 0.68 are similar to those reported by Wen et al [32, 33] for YBCO thin films. However, the value µ ' 0.68 deviates from the corresponding one for the single-vortex regime µ = 1/7 and is intermediate between single-vortex and small-bundle regime µ = 5/2 according to collective pinning theory. One explanation of this deviation is that the transition from single-vortex to small-bundle regimes should be considered as a rather wide crossover instead of a sharp boundary and so the application of Maley’s method in our relaxation data gives an intermediate value for the 687 E Moraitakis et al 4 0.9 2.13 G/sec 60K (U/k)/g(T) 3000 50K 2000 2 1.5 2 1.5 c=30,g(T)=[1-(T/Tc) )] c=25,g(T)=[1-(T/Tc) )] 45K 2 c=30,g(T)=[1-(T/Tc) )] 40K 47.2 G/sec 2 H=20 kG M (emu) (U/k)/g(T)(K) 20.0 G/sec 1000 20 30 40 84.6 G/sec 0.3 1 10 100 dH a/dt (G/sec) 0 100 1000 2 6 2 g(T)=[1-(T/Tc) )] 1.5 2x10 9x10 30K 20K 0 6 6 j(A/cm ) 1x10 H=20kG,c=30 6 4x10 15K 6 2 6x10 -2 10K 6 8x10 0 10 Figure 7. Plot of the U (j, T = 0, H = 20 kG) relation determined from the relaxation data at different temperatures according to Maley’s method for the case of a disc. The value C = 30 was used. The inset is a double-logarithmic plot of the U (j ) relation for different C values and different functions g(T ) as discussed in the text. The full curves are fits to the data with the interpolation formula. exponent µ. Although there is a slight deviation of data at T = 60 K in figure 7, the relaxation of the magnetic moment does not clearly show any change of the observed logarithmic time dependence within the examined time window. This discrepancy cannot be easily resolved. Figure 8 shows hysteresis loops of the YBCO disc at T = 10 K, collected for different field sweep rates R = dHa /dt by varying the field increment. Sweep rates in SQUID measurements are not actually constant but can be approximated with some average values. It is obvious that the m versus Ha curves for these sweep rates almost coincide with each other. In fact the inset of figure 8 shows in a double-logarithmic plot the hysteretic moment 1m = m+ − m− = πa 3 j d/3c, calculated using the Bean model, as a function of the sweep rate R. The hysteretic moment 1m shows a rather weak dependence on R, indicating that we scan a very steep part of the E(j ) curves for these sweep rates. The current density values as a function of the applied field or R = 20 Oe s−1 calculated from the hysteretic moment (Bean model) are shown in figure 2. A measurement of the magnetic moment with a SQUID magnetometer when a hysteresis loop is formed lasts approximately t ' 60 s after the applied field has stabilized to the desired value. The corresponding current density value is not the critical current jc but some relaxed value reached during this time interval. The same power law dependence as for the relaxation data, j (H ) ∝ H −b with the same exponent b ' 0.5, was found to fit the data better. The power law dependence of jc needs modification for low fields owing to a singularity at H = 0. It should be noted that the characteristic field above which the critical current jc flows in most of the disc’s area and the magnetic moment is saturated is Hc = 2πjc d/c. This field is typically no more than 600 G for YBCO films at T = 5 K [9, 10]. This means that, for the field range from 5 to 50 kG examined above, H Hc and the disc should be considered in the full flux-penetrated state. As shown in [10] the critical state of a thin disc, incorporating 20 30 40 50 H(kG) j(A/cm ) 688 10 kG ∆ m(emu) 4000 Figure 8. Hysteresis loops of the YBCO disc collected at T = 10 K for different sweep rates R = dHa /dt by varying the field increment. The inset shows the hysteretic moment 1m, calculated using the Bean model as a function of the sweep rate R. the dependence of jc on the applied field through a Kimlike expression jc (Ha ) = jc (0)/(1 + |Ha |/B0 ) and taking into account self-field effects, successfully reproduces the experimental hysteresis loops up to 15 kG for YBCO thin films. 5. Conclusions The flux-creep behaviour of a YBa2 Cu3 O7−δ thin disc for the temperature range T = 10–60 K and for a perpendicular applied field Ha = 5–50 kG was analysed in the framework of the analytical model described in [17], as discussed in the text. Within experimental accuracy, the relaxation of the magnetic moment shows a logarithmic dependence on time as predicted for a sufficiently steep E(j ) relation in the above model. The critical current jc values show an exponential dependence on temperature of the form jc (T ) = jc (0) exp(−T /T0 ) with T0 ' 25 K, which may be considered as a manifestation of collective pinning and creep effects. The temperature dependence of the pinning potential U0 shows a maximum which shifts to lower temperatures for higher fields and may indicate a field-dependent crossover of the vortex state. Furthermore, jc values show a power law dependence on field of the form jc (H ) ∝ H −b with b ' 0.50, which holds during the relaxation process. Using the above analysis the E versus j characteristics for the disc geometry are quantitatively obtained from the relaxation data, and show a very steep E(j ) relation described by a power law E(j ) = Ec (j/jc )n while a change in this behaviour is observed for T = 60 K. Furthermore, the application of Maley’s method for the extraction of the activation barrier U (j, T = 0, H ) based on the exact relations for a thin disc gives an exponent µ ' 0.68 for H = 20 kG which lies between those for the single-vortex (1/7) and the small-bundle (5/2) regimes. References [1] Blatter G, Feigel’man M V, Geshkenbein V B, Larkin A L and Vinokur V M 1994 Rev. Mod. Phys. 66 1125 [2] Bean C B 1964 Rev. Mod. Phys. 36 31 [3] Mikheenko P N and Kuzovlev Yu E 1993 Physica C 204 229 Flux creep in a thin disc of YBa2 Cu3 O7−δ [4] Zhu J, Mester J, Lochhart J and Turnearure J 1993 Physica C 212 216 [5] Zeldov E, Clem J R, McElfresh M and Darwin M 1994 Phys. Rev. B 49 9802 [6] Brandt E H, Indenbom M V and Forkl A 1993 Europhys. Lett. 22 735 [7] Brandt E H and Indenbom M V 1993 Phys. Rev. B 48 12 893 [8] Brandt E H 1997 Phys. Rev. B 55 14 513 [9] Moraitakis E, Pissas M and Niarchos D 1995 Supercond. Sci. Technol. 8 647 [10] Moraitakis E, Pissas M and Niarchos D 1995 Physica C 241 63 [11] Frankel D J 1979 J. Appl. Phys. 50 5402 [12] Daeumling M and Larbalestier D C 1989 Phys. Rev. B 40 9350 [13] Theuss H, Forkl A and Kronmuller H 1992 Physica C 190 345 [14] Schuster Th, Kuhn H, Brandt E H, Indenbom M V, Klaser M, Muller-Vogt G, Habermeier H U, Kronmuller H and Forkl A 1995 Phys. Rev. B 52 10 375 [15] Xu X N, Sun A M, Qin M J, Ding S Y, Jin X, Yao X X and Yan S L 1997 Physica C 291 315 [16] Brandt E H 1994 Phys. Rev. B 49 9024 Brandt E H 1994 Phys. Rev. B 50 4034 [17] Gurevich A and Brandt E H 1994 Phys. Rev. Lett. 73 178 [18] Brandt E H 1996 Phys. Rev. Lett. 76 4030 [19] Maley M P and Willis J O 1990 Phys. Rev. B 42 2639 [20] Anderson P W 1962 Phys. Rev. Lett. 9 309 Anderson P W and Kim Y B 1994 Rev. Mod. Phys. 36 39 [21] Feigel’man M V, Geshkenbein V B, Larkin A L and Vinokur V M 1989 Phys. Rev. Lett. 63 2303 Feigel’man M V, Geshkenbein V B, Larkin A L and Vinokur V M 1991 Phys. Rev. B 43 6263 [22] Fisher M P A 1989 Phys. Rev. Lett. 62 1415 Fisher D S, Fisher M P A and Huse D A 1991 Phys. Rev. B 43 130 [23] Malozemoff A P 1991 Physica C 185–189 264 [24] Kes P H, Aarts J, van der Beck J and Mydosh J A 1989 Supercond. Sci. Technol. 1 242 [25] Zhukov A A 1992 Solid State Commun. 82 983 [26] Zhukov A A et al 1993 Cryogenics 33 142 [27] Gurevich A and Kupfer H 1993 Phys. Rev. B 48 6477 [28] Moraitakis E, Anagnostou M, Pissas M, Psyharis V, Niarchos D and Stratakos G 1998 Supercond. Sci. Technol. 11 686 [29] Pissas M, Stamopoulos D, Moraitakis E, Kallias G, Niarchos D and Charalambous M 1999 Phys. Rev. B 59 12 121 [30] Thompson J R, Sun Y R, Christen D K, Civale L, Marwick A D and Holtzberg F 1994 Phys. Rev. B 49 13 827 Thompson J R, Sun Y R and Holtzberg F 1991 Phys. Rev. B 44 458 [31] Griessen R, Wen H-h, van Dalen A J J, Dam B, Rector J and Schnack H G 1994 Phys. Rev. Lett. 72 1910 [32] Wen H-h, Schnack H G, Griessen R, Dam B and Rector J 1995 Physica C 241 353 [33] Wen H-h, Zhao Z X, Wijngaarden R J, Rector J, Dam B and Griessen R 1995 Phys. Rev. B 52 4583 [34] Civale L, Krusin-Elbaum L, Thompson J R and Holtzberg F 1994 Phys. Rev. B 50 7188 [35] Dekker C, Eidelloth W and Koch R H 1992 Phys. Rev. Lett. 68 3347 [36] Deak J, McElfresh M, Clem J, Hao Z, Konczykowski M, Muenchausen R, Foltyn S and Dye R 1994 Phys. Rev. B 49 6270 [37] Malozemoff A and Fisher M P A 1990 Phys. Rev. B 42 6784 [38] Reissner M and Lorenz J 1997 Phys. Rev. B 56 6273 [39] Zeldov E, Amer N M, Koren G, Gupta A, McElfresh M W and Gambino R J 1990 Appl. Phys. Lett. 56 680 [40] Mawatari Y, Sawa A, Obara H, Umeda M and Yamasaki H 1997 Appl. Phys. Lett. 70 2300 [41] Sandvold E and Rossel C 1992 Physica C 190 309 [42] Tinkham M 1988 Phys. Rev. Lett. 61 1658 [43] McHenry M E, Simizu S, Lessure H, Maley M P, Coulter J Y, Tanaka I and Kojima H 1991 Phys. Rev. B 44 7614 [44] Xiao Z L, Haring J, Heinzel C and Ziemann P 1995 Solid State Commun. 95 153 [45] Ossandon J G, Thompson J R, Christen D K, Sales B C, Sun Y and Lay K W 1992 Phys. Rev. B 46 3050 689