Rep. Prog. Phys. 60 (1997) 1581–1672. Printed in the UK PII: S0034-4885(97)41466-5 Open questions in the magnetic behaviour of high-temperature superconductors L F Cohen† and Henrik Jeldtoft Jensen‡ † Blackett Laboratory, Imperial College, Prince Consort Road, London SW7 2BZ, UK ‡ Department of Mathematics, Imperial College, 180 Queen’s Gate, London SW7 2BZ, UK Received 10 March 1997 Abstract A principally experimental review of vortex behaviour in high-temperature superconductors is presented. The reader is first introduced to the basic concepts needed to understand the magnetic properties of type II superconductors. The concepts of vortex melting, the vortex glass, vortex creep, etc are also discussed briefly. The bulk part of the review relates the theoretical predictions proposed for the vortex system in high temperature superconductors to experimental findings. The review ends with an attempt to direct the reader to those areas which still require further clarification. c 1997 IOP Publishing Ltd 0034-4885/97/121581+92$59.50 1581 1582 L F Cohen and H J Jensen Contents 1. Introduction 2. The vortex system and its behaviour 2.1. Type I and type II superconductors 2.2. Isotropic ideal type II superconductors 2.3. H c1 and H c2 2.4. Disturbances of the ideal hexagonal flux-line lattice 2.5. Fluctuations in the order parameter 2.6. Anisotropy 2.7. Thermal equilibrium and non-equilibrium properties 2.8. Symmetry of the order parameter 3. Interpreting vortex behaviour 3.1. Transport and magnetization measurements 3.2. Transport E–J curves 3.3. Flux creep 3.4. Magnetic measurement analysis 3.5. Critical scaling applied to E–J curves 4. Experimental observation of vortex behaviour 4.1. Reversible properties 4.2. The irreversibility line 4.3. In the vicinity of H irr 4.4. The order of the melting transition 4.5. Below the irreversibility line—the vortex solid 5. Summary of the questions at the brink of resolution Acknowledgments References Page 1583 1584 1584 1584 1585 1586 1593 1594 1594 1596 1596 1596 1598 1599 1602 1603 1604 1604 1608 1618 1630 1649 1663 1665 1665 Magnetic behaviour of superconductors 1583 1. Introduction The justification for a predominantly experimental review of the magnetic behaviour in hightemperature superconductor (HTS) materials is simply the recognition that an introductory text, as this sets out to be, may yet help shed light on the behaviour of vortices in the presence of disorder. Furthermore, if we are to utilize HTS materials in the form of magnets, power cables or high-frequency filters, summarizing our understanding of the pinning properties of vortices and the form of the HT diagram in equilibrium or otherwise—is still of paramount importance. Several excellent reviews have recently appeared (Farrell 1994, 1995, Fischer 1993, 1994, Blatter et al 1994b, Brandt 1995) and inevitably there will be some overlap. Interpretation of the experimental evidence will probably not stand the test of time in the same way as a theoretical review because as new data appears it sheds light on all that has gone before. Nevertheless, we feel it is important to brave the unknown, take a snap shot in time, examine the current position critically and ask whether various types of novel behaviour which have been predicted have indeed been observed. Our task is made more difficult because observed vortex behaviour is linked to underlying static disorder and the general classification of material purity and quality is still incomplete. In an attempt to simplify matters we have restricted the discussion to single crystals and, where necessary, to thin films. We limit the discussion further to YBCO 123 and BSCCO 2212 to illustrate the range of properties in systems with very different anisotropy. In reality the magnetic properties of HTS is a labyrinth. One can easily get lost and confused. The sirens’ song sounds seductively in the form of wonderful sparkling theoretical inventions: Bose glass, vortex glass, quantum creep, melting, entanglement, disentanglement, pancakes, dimensional crossovers, plastic flow. The list goes on and on. Only a strong and clearly directed guide will enable one to make it through the maze. Even more so if one not only wants to survive the expedition, but in fact has the ambition to refresh one’s mind and gain overview and understanding from the quest. In section 2 we develop the basic notions used to describe magnetic properties of superconductors. We then introduce various concepts used to describe HTS behaviour which are novel at least to superconductivity, and highlight those ideas which are simply an extension of concepts discussed previously. We discuss similarities and differences between conventional and HTS. The main difference between the new and old superconductors is the relevance of thermal fluctuations and strong anisotropy.We have tried to address in general terms the connection between the new theoretical developments and quantities measured in experiments. We pay special attention to the importance of distinguishing between thermodynamic equilibrium quantities and non-equilibrium experiments. In section 3 we review some concepts related to the interpretation of experimental data which we feel would otherwise weigh down the discussion in the next section. We place significant emphasis on the understanding of transport and magnetization measurements as these provide the bulk of experimental evidence. In section 4 we really start our expedition into the wilderness. For consistency sake we have tried to divide the enormous body of experimental information into broad headings which reflect the novel phase diagrams which have been proposed for HTS materials. We 1584 L F Cohen and H J Jensen review evidence for the behaviour of Hc1, Hc2, the form and meaning of the irreversibility line, the melting line and glassy behaviour. We discuss the different regions of behaviour in the field (H ) and temperature (T ) plane emerging from transport and magnetization data, in order to explore whether different experimental angles of approach lead to a consistent picture. The phenomenology derived from this exploration is brought into contact with various the theoretical pictures introduced in section 2. In section 5 we summarize the open questions which remain. 2. The vortex system and its behaviour In this section we will run through the basic concepts of how a magnetic field behaves inside a superconductor. Details are elaborated in many books on superconductivity. Two excellent classics are Tinkham (1995) and de Gennes (1966). We will also expand on these ideas in very general terms to include the novel aspects of HTS in magnetic fields. A further reference which gives a good idea of the complexity of the problem is the extensive review by Blatter et al (1994b). 2.1. Type I and type II superconductors Superconductors exist in one of two types. In the first kind an external magnetic field cannot penetrate into the bulk of the sample without destroying the superconducting condensate. We are not going to deal more with this kind. The second kind of superconductors, of which HTS are prominent members, are able to remain superconducting over a range of fields H in the interval Hc1 < H < Hc2 . At the lower critical field Hc1 the first magnetic flux starts to enter the bulk of the superconductor. The field does not penetrate the bulk in a homogenous way. Had this been the case, the magnetic properties of type II superconductors would have been much simpler. The mixed state which exists for fields between Hc1 and Hc2 is spatially inhomogeneous. Both the local magnetic induction and the local density of superconducting electrons are position dependent. The magnetic field penetrates the bulk of the superconductor in the form of quantized flux tubes or magnetic vortices. 2.2. Isotropic ideal type II superconductors Figure 1 illustrates a vortex line and the important lengths λ, the penetration depth and ξ , the coherence length. At zero temperature in an isotropic ideal superconductor containing no inhomogeneities in the superconducting matrix the mixed state is threaded by straight vortex lines running parallel to the direction of the external magnetic field. They are called vortex lines because they consist of vortices in the superfluid of Cooper-paired electrons. These vortices were discovered by Abrikosov using the phenomenological Ginzburg–Landau theory (Abrikosov 1957). The diverging circulation velocity as one approaches the centre of the vortex drives the density of superelectrons to zero. At the axis of the vortex the superconducting order parameter is equal to zero. It increases as one goes radially out from the vortex core and reaches its asymptotic limit over a distance given by the Ginzburg– Landau coherence length ξ . For HTS, ξ ≈ 10–20 Å at zero temperature (in the direction parallel to the superconducting planes, see below). At the centre of the vortex the magnetic induction is maximum. This field is screened by√the circulating supercurrents. As a result the magnetic induction decreases as exp(−r/λ)/ r as one goes away from the vortex axis. The field and circulating currents decrease rapidly to zero beyond the London penetration depth λ. Each vortex carries one quantum of magnetic flux φ0 = h/2e where h is Planck’s Magnetic behaviour of superconductors 1585 Figure 1. An illustration of a vortex line and the important lengths, λ the penetration depth and ξ the coherence length. constant and e is the elementary charge. Therefore the number of flux lines inside the sample is approximately proportional to the external field. For HTS, λ ≈ 1500 Å at zero temperature. The circulating supercurrents (or equivalently the magnetic induction) give rise to an interaction between the flux lines, extending out to a distance of order λ. The depletion of the order parameter at the vortex axis also leads to a short-range attraction between vortices. The long-range magnetic interaction is repulsive for straight parallel vortex lines and attractive for antiparallel lines. Due to this interaction the minimum energy configuration for a flux-line system in an ideal isotropic superconductor consists of parallel vortex lines arranged in an hexagonal lattice in the plane perpendicular to the field direction. 2.3. Hc1 and Hc2 The transitions at Hc1 and Hc2 can be thought of as follows. At the lower critical field the (Gibb’s) free energy of the state without a flux line is equal to the state with one flux line (or many non-interacting flux lines). It uses energy to keep the field out of the bulk of the superconductor. As soon as the field becomes a tiny bit larger than Hc1 flux lines will flow into the bulk. As long as they do not interact, the free energy is independent of the number of flux lines within the bulk. Hence, the flux lines will rush in unhindered until they start to have an average separation of order λ. Accordingly the magnetic field at Hc1 will approximately be one flux quantum within the area of a circle of radius λ. The precise expression of Hc1 is Hc1 = φ0 ln(λ/ξ )/(4πλ2 ). As the external field is gradually increased, the flux lines are squashed together. Eventually their cores, in which the superconducting order parameter is equal to zero, will begin to overlap and the whole bulk becomes normal. This is what happens at Hc2 , at least at the simplest mean-field Ginzburg–Landau level of description. We expect the value of Hc2 to be given by one flux quantum through an area of the size of the core. This simple picture becomes more complicated as the effect of thermal fluctuations is included. Fluctuations always become important close to the temperature where the system undergoes a (continuous) phase transition. The width of the fluctuation regime depends on the ‘stiffness’ of the order parameter close to Tc (H ). Mean-field theory is applicable as long as the length scale of spatial variations is longer than or equal to the coherence length ξ of the mean-field theory. When the thermal energy kB T becomes of the order of the free energy within a correlated volume ξ 3 f , where f is the free-energy density, fluctuations make the mean-field theory inappropriate. Since ξ ∼ (Tc − T )−1/2 and 1586 L F Cohen and H J Jensen f ∼ (Tc − T )2 close to Tc , in mean-field theory ξ 3 f/kB T ≈ 1 as Tc is approached (Landau and Lifshitz 1969). Precisely how close to Tc fluctuations become important depends on the size of ξ and Tc . For conventional low-temperature superconductors ξ is large and Tc is small. The transition in these superconductors is therefore described well by mean-field Ginzburg– Landau theory except for an unresolvably narrow region of width about 10−4 K around Tc . In the HTS the situation is reversed. The coherence length is short and the transition temperature is high. Thermal fluctuations are accordingly much more important over a broad regime around the Hc2 (T ) line. In conventional superconductors the experimentally observed transition from a resistive phase for T > Tc (H ) to a phase with zero resistivity below Tc (H ) occurs very sharply as the temperature is lowered through the meanfield transition temperature Tc (H ). In HTS the resistivity only vanishes slowly while a superconducting state gradually builds up. The true superconducting phase transition takes place at a temperature significantly lower than the mean-field value for Tc (H ). Below we shall return to the question: What replaces the mean-field Hc2 (T ) line? The very nature of the Hc2 (T ) line is unclear when one goes beyond mean-field considerations. It is not even known if fluctuations change the nature of the transition from a continuous transition (as in mean field) to a first-order transition. Some calculations suggest that the transition in the pure system is first order and that disorder replaces the transition by a gradual crossover from the vortex liquid to a vortex solid (Moore and Newman 1995). Also the nature of the mixed state in the neighbourhood of the Hc2 (T ) is much more subtle than hitherto anticipated. This is also true for conventional superconductors. However, for the low-temperature superconductors these questions are mainly of non-observational academic interest. In HTS this subtle neighbourhood is much broader and of greater practical importance. 2.4. Disturbances of the ideal hexagonal flux-line lattice Let us again return to the simple mean-field Ginzburg–Landau description of superconductivity. We need to consider the extent to which the ideal hexagonal line lattice can be disturbed by defects and thermal fluctuations. 2.4.1. Defects. Inhomogeneities in the superconducting material can lead to a local suppression of the superconducting order parameter. An example is a void or a hole in the superconductor. This will lead to an interaction between the vortex core and void. In order to minimize the suppression of the order parameter it will be beneficial to locate the vortex core on top of the void, thereby only depleting the order parameter once. This mechanism leads to core pinning. The void attracts the vortex line, hence it tends to trap or pin the line. The spatial variation in the flow pattern of the supercurrents can also lead to pinning, especially to surface pinning. Core pinning is in general the most important bulk pinning mechanism. A random arrangement of material inhomogeneities will lead to a random potential and random forces acting on the vortex lines. This will disturb the positional order of the flux lattice. The interaction between the inhomogeneities in the superconducting matrix and the flux lines is of crucial importance for the ability of the superconductor to support a dissipation free electric current when penetrated by vortices. The reason is as follows. As an electric current is passed through the superconductor the Lorentz force will act between electrons and the (localized) magnetic field carried by the flux lines. In this absence of pinning this force will move the flux lines with the result that a time-dependent local magnetic Magnetic behaviour of superconductors 1587 Figure 2. An illustration of the defect structure and the soft and stiff vortex lattice ‘trying’ to take advantage of the defects. induction is established. As a result, an electric field is induced which then acts on the electrons thereby leading to a voltage and corresponding dissipation. The details of this scenario are controlled by the Josephson relation (Josephson 1965, Tinkham 1995). The only way dissipation can be avoided when a magnetic field threads the superconductor is by preventing the flux lines from moving. Since inhomogeneities attract the flux lines they are able to prevent this motion up to a certain pinning force Fp . The degree to which the flux lines are pinned determines the maximum Lorentz force one can apply without dissipation. The Lorentz force (per volume) is given by Bj , where B is the magnetic induction locally averaged over the flux lines and j is the current density. The maximum dissipation free current—called the critical current—is given by jc = Fp /B. It is important to keep in mind that even if one could make the pinning centres infinitely strong there would still be an upper bound for the dissipation free current. Dissipation will then occur when the Cooper pairs start to break up due to the induced current. This happens at the depairing current where the kinetic energy of the Cooper pairs equals to the superconducting condensation energy which binds the electrons together in Cooper pairs (see Tinkham 1995). However, in a magnetic field the depinning critical current is always found to be smaller than the depairing current. The efficiency of the pinning centres depends indirectly on the strength of the vortexvortex interaction. A very stiff vortex system will not be able to adjust to the random pinning potential and can therefore not relax deeply into the pinning potential. A soft vortex system on the other hand will be able to adjust itself to the random pinning forces and thereby sink deep down into the pinning potential. This leads to a more strongly pinned configuration than in the case of a stiff vortex system. Figure 2 illustrates point defect structure and soft and stiff vortex lattices attempting to fit to it. The interaction between flux lines can, as a good approximation, be described by a two-body potential between flux-line elements. For an isotropic interaction, the interaction energy between two flux-line segments dl1 and dl2 separated by the distance r12 is given by the sum of two exponentially screened Coulomb-like contributions 0 0 Um + Uc = dl1 · dl2 e−r12 /λ − |dl1 ||dl2 |e−r12 /ξ . (2.1) The first term has its origin in the magnetic field carried by the flux lines. This interaction 1588 L F Cohen and H J Jensen √ is screened beyond an effective field-dependent magnetic penetration depth λ0 = λ/ 1 − b. Where b = B/Bc2 is the ratio between the actual induction B and the induction Bc2 corresponding at the upper cirtical field Hc2 . The second term describes an attraction between √ the core of the flux lines and is very short ranged ξ 0 = ξ/ 2(1 − b). The interaction between flux-line elements in anisotropic superconductors is of the same nature although in more complicated detail due to the dependence of the interaction on the orientation of r12 with respect to the symmetry axis of the material. An excellent discussion of these important details is given by Brandt (1995) in his recent review. The interaction between the flux lines is responsible for the elastic rigidity of the flux lattice. The elastic properties are described by a shear C66 modulus, a tilt modulus C44 , and a compression modulus C11 . These moduli have been calculated for isotropic superconductors as well as for anisotropic superconductors, see again Brandt (1995). Here we list the expressions of elastic moduli. An essential point to bear in mind is that these moduli are field dependent such that C11 ∼ C44 ∼ b2 and C66 ∼ b(1 − b)2 , where b = B/Bc2 . The shear modulus vanishes at the upper critical field giving rise to a softening of the flux lattice and thereby a more effective pinning close to Hc2 . The increase in the pinning force close to Hc2 is known as the peak effect (Pippard 1969). Another important point is that the tilt and compression moduli depends strongly on the wavelength of the imposed elastic deformation. A short wavelength tilt deformation uses significantly less energy than homogeneous tilt. This has to be taken into account when making quantitative estimates of the deformations of the ideal flux lattice. The interaction between a single pinning centre and a segment of a vortex line is difficult to calculate and depends on the nature of the pinning interaction. However, it is useful to bear in mind an estimate of the pinning energy obtained from the excluded volume effect. If the defect depresses the superconducting order parameter in a volume of size V (smaller than the core volume) the energy gained by positioning the core of the vortex line on top of the defect will be of order VHc2 ,√since the superconducting condensation energy density is given by the square of Hc = φ/2 2πλξ . Close to Hc2 the condensation energy vanishes as (1 − b) (Thuneberg 1984). Since C66 vanishes as (1 − b)2 the pinning energy may dominate over the elastic energy in this field regime. This is the explanation of the peak effect mentioned above and discussed in sections 4.3.2 and 4.5.1. Pinning of the flux lines is not only induced by random disorder in the bulk. Any spatial inhomogeneity in the superconducting properties may make the free energy of a vortex line depend on position. The energy of the supercurrents circulating the vortex line will vary close to the sample surface. This effect leads to the Bean–Livingston surface barrier which a flux line parallel to a plane surface has to surmount in order to enter into the bulk of the sample (see e.g. de Gennes 1966). Another barrier to flux entry relates to the shape of the sample and is denoted by a geometric barrier (see e.g. Zeldov et al 1994). The barrier is again an energy barrier that the flux line will have to overcome in order to move between the interior and exterior of the sample. The barrier is estimated from two contributions. One is the energy needed to create a flux line of the length of the sample thickness. The other contibution is the energy extracted when the Lorentz force induced by the circulating supercurrents, perform work on the flux line while these currents attempt to move the flux line towards the centre of the sample. If one chose an appropriate shape of the sample this barrier can by made to vanish. See figure 16. For further discussion see sections 4.2.2 and 4.5.1. Finally we must mention that the layered structure of the cuperate HTS leads to a significant intrinsic pinning. When the flux lines are arranged parallel to the superconducting copper-oxide planes energy is gained when the normal core of the flux line is positioned Magnetic behaviour of superconductors 1589 in the less strongly superconducting region in between the superconducting planes. The flux lines have to overcome an essential energy barrier in order to move across the superconducting planes. This effect leads to dramatic peaks in the critical currents when field alignment is nearly parallel to the copper-oxide planes. See figure 17. 2.4.2. Thermal fluctuations. Melting of the flux lattice. As discussed in section 2.3, thermal fluctuations are much more important in HTS than in low-temperature superconductors. Thermal fluctuations can also perturb the flux-line configuration. Like in an ordinary crystal lattice kept at a constant temperature, elastic forces in the vortex lattice will result in displacements, u, of the vortex lines away from the ideal configuration to reach an average distance given by hu2 i ∝ T . The proportionality constant is determined by the elastic moduli of the flux lattice. If there is sufficient thermal energy available so that u becomes of order 10–30% of the average flux-line separation, one can expect that the flux-line lattice might melt (Nelson 1988). This is called the Lindemann melting criterion U = cL a0 , here cL is somewhere around 0.1–0.3 and a0 denotes the average flux-line separation. The criterion is a phenomenological principle that is known to work for ordinary crystals. The limitation of this principle is that it does not explain what type of fluctuations in the lattice structure (dislocations, disclinations etc) actually causes the melting. One should of course find a way to calculate the temperature at which the shear modulus describing homogeneous shear goes to zero (the short wavelength shear modulus remains non-zero in the liqiud). This has not yet been done (for any three-dimensional system in fact). Nevertheless, the melting of the flux-line lattice is thought of as a melting in the traditional sense, namely as a softening of the flux lattice leading to a phase unable to support any shear. The curve Tm (H ) in the T –H phase diagram at which the flux lattice melts is called the melting line. In principle for fields very close to Hc1 the flux system should always be a ‘liquid’ since the flux lines are separated more than λ leading to a vanishing interaction between the vortices. In practice, flux lines enter the sample at Hc1 very rapidly and the region where the separation is larger than the interaction length is hardly accessible. However, very recently this ‘re-entrant’ melting effect has been observed experimentally (Ravikumar 1997) in the low-temperature superconductor Nb2 Se. The fact that the melted flux-line lattice has lost its rigidity (shear modulus equal to zero) has been used more or less intuitively to suggest that dissipation, i.e. flux flow, is more easy in the melted phase than in the solid phase. It does not need to be so. One should keep in mind the fact that the solid Abrikosov lattice will flow subject to the slightest applied driving force if no pinning potential is present to hinder the motion. Furthermore, the efficiency of the pinning potential is reduced by the stiffness of the flux system. In order to follow the pinning centres, the flux lattice has to distort. This uses elastic energy. Non-interacting flux lines of infinite flexibility would be able to take full advantage of the pinning potential. A decrease in the elastic moduli is a move towards this optimum situation. The peak in the pinning force observed close to Hc2 in conventional superconductors was once connected with the softening of the elastic coefficients of the flux-line lattice with field (Pippard 1969, Larkin and Ovchinninkov 1979). If the shear modulus of the flux lattice vanishes upon thermal melting the pinning centres could act more efficiently leading to a reduced mobility of the flux lines and therefore a reduced dissipation. This scenario assumes a density of pinning centres much higher than the flux-line density. If there are only a few very strong pinning centres the situation would be reversed. The melted flux liquid will be able to flow in between the pinning centres leading to an increased dissipation above the melting temperature. In any case, one assumes 1590 L F Cohen and H J Jensen that the energy scale of the pinning potential is larger than the available thermal energy kB Tm at the melting transition. Otherwise the pinning centres would already have become unable to trap the flux system at a lower temperature. In this case one would observe a depinning line (see section 2.4.3) rather than a melting line. 2.4.3. Irreversibility line. Real systems always contain pinning centres. This leads to yet another energy scale. The properties of the flux system are determined by the relationship between the thermal energy Eth , the vortex–vortex energy Evv , and the pinning energy Epin . The competition between these three energy scales is complicated because the effective pinning energy depends on the vortex–vortex interaction energy. Pinning centres become more effective if the interaction between vortices is small. The competition between Eth and Epin leads to the existence of the irreversibility line in the H –T plane. This line is determined as follows. For temperatures below this line the pinning is strong enough to be able to trap flux lines as they are pushed in and out of the sample. If one sweeps the external field from zero up to a value H > Hc1 and down again to zero some flux lines will remain pinned inside the sample even after the external field has returned to zero. The magnetization of the sample behaves in an irreversible way. Above the line the only contribution to the magnetization comes from the reversible Meissner effect. The thermal energy now dominates the pinning energies so that the flux is no longer trapped in metastable configurations. The irreversibility line is identified in current–voltage experiments as the line in the H –T plane that separates the region of fields and temperatures above the line where the voltage depends linearly on the applied current from a region below the line of a nonlinear current–voltage characteristics. Figure 3 shows the rough effective pinning potential acting on a flux bundle. Eth = kB T is the scale of the thermal fluctuation. Flux bundle at position A is easily thermally activated out of the local potential well. The flux bundle at B is trapped. In theory, there is no reason for a specific relationship to exist between the irreversibility line and the melting line. The melting occurs when Eth ≈ Evv . Reversible behaviour is separated from irreversible behaviour when Eth ≈ Epin . Depending on the accidental relation between Epin and Evv (accidental since Epin depends on the properties of the defects of the Figure 3. The rough effective pinning potential acting on a flux bundle. Eth = kB T is the scale of the thermal fluctuation. The flux bundle at position A is easily thermally activated out of the local potential well. The flux bundle at B is trapped. Magnetic behaviour of superconductors 1591 material, whereas Evv is an intrinsic flux lattice property) the melting line and irreversibility line can be very close together, or melting can occur at a lower or higher temperature than where reversibility sets in. All three possibilities have been identified in experiments although the different lines were originally assumed to be the same. 2.4.4. Vortex glass. The effect of the pinning potential (which results from the underlying static disorder) is at the heart of the approach to the phases of the flux system that has become known as the vortex glass scenario (Fisher 1989). The vortex glass approach tries to incorporate the pinning potential from the beginning. This is in contrast to the melting theory described above (see section 2.4.2). The melting theory focuses on the properties of the pure system. The effect of the pinning potential is then treated as a perturbation of the pure system. The high-temperature phase of the vortex glass model is considered to be a liquid of mobile flux lines moving unhindered over the pinning potential. The low-temperature phase is an immobile amorphous phase. Because the pinning potential is supposed to disorder the flux system. The essence of the vortex glass picture is that collective effects are predicted to be able to produce infinite energy barriers leading to a strictly zero linear flux-flow resistance as J approaches zero. Different workers have emphasized different aspects of the physics of the vortex glass. Fisher (1989) introduced the term vortex glass. Fisher was especially concerned with the transition between the high-and low-temperature phase. Fisher argued that the transition is a true phase transition. Furthermore, Fisher assumed the transition to be continuous and worked out a scaling theory for the voltage–current characteristics in the vicinity of the transition (see section 3.5 below). Feigel’man et al (1989) formulated a theory of the voltage–current characteristics inside the low-temperature ‘glass’ phase. They generalized the collective pinning approach developed by Larkin and Ovchinnikov (1979) to discuss the collective behaviour of the flux system. Their model is known as the ‘collective creep’ theory. Fiegel’man et al (1989) calculated from elasticity theory the effective energy barriers set up by the competition between the elastic vortex–vortex interaction and the pinning potential. They considered how the flux bundles creep over these barriers and they derived power laws for the logarithmic time dependence of the electric current inside the vortex glass regime. Experimental observation of the vortex system deep in the vortex solid is discussed in section 4.5. Various viewpoints are advocated concerning the thermal stability of the vortex glass. One school (Nelson and Vinokur 1992, 1993) claimed that a thermodynamically stable glass is only possible if strong disorder is present in the form of columnar defects or twin planes. Randomly positioned point defects might induce distortions of the flux lattice but point defects will not be strong enough to establish a stable glass in the thermodynamic sense. The vortex glass idea was originally suggested in connection with point defects. It was suggested (Fisher et al 1991) that point defects are able to produce the diverging barriers associated with the vortex glass. Experimentally this question is delicate since the response of a system with large but finite barriers might easily look like the response from a system with infinite barriers. The situation is similar to the one encountered in connection with ordinary glasses. As one goes through the glass temperature the viscosity changes by 15 orders of magnitude. However, so far no one has been able to show that this change is related to a genuine phase transition (see e.g. Nagel 1993). The vortex glass is supposed to be a consequence of diverging energy barriers (Fisher et al 1991, Feigel’man et al 1989). That such a divergence may in principle exist is most simply seen from the following argument. The relaxation of the flux-line structure is always 1592 L F Cohen and H J Jensen caused by a driving (Lorentz) force which is proportional to the local current density j . This current arises due to the existence of a gradient in the density of the flux lines. According to the Maxwell relation ∇ × B = j , i.e. j goes to zero as the flux structure relaxes to a homogeneous arrangement. Assume that the Lorentz force is able to move a flux bundle of volume V . The total driving force on this volume, Fd , is proportional to Vj . The driving force has to be able overcome the pinning force acting on this volume. The pinning force, Fp , is a sum of V np (here np is the density of pinning centres) individual pinning forces acting in random directions. We estimate the sum of the pinning forcespby measuring the variance of the sum of V np independent random numbers, i.e. Fp ∼ V np (Larkin and Ovchinnikov 1979). Precisely when the driving force is able to make the flux bundle inside the volume V move we have Fd = Fp and therefore V ∼ 1/j 2 . Hence, as j → 0 the volume that will have to move coherently increases. This makes the energy barrier produced by the pinning centres and the compressibility of the flux system (Feigel’man et al 1989) diverge. It is always difficult to establish equilibrium in systems with large energy barriers. This is a well known theme in the field of spin glasses. This must be kept in mind when analysing experiments. The properties observed below the irreversibility line, where the vortex glass is supposed to exist, are hysteretic (by definition) and their relation to genuine equilibrium properties is complicated. Another point to keep in mind is that the diverging energy barrier arises because it is assumed that the coherent motion of a larger and larger volume is necessary in order to induce relaxation. This may not be so. Plastic deformations of the flux structure may be able to break the flux bundle volume up into smaller pieces which can be moved by a finite-energy input. The scenario is easy to visualize in two dimensions. Here the energy of a dislocation moving through the flux lattice is some finite energy given by the shear strength of the system and expected to be proportional to the shear constant C66 . (For a three-dimensional flux system this energy will be proportional to the thickness of the sample.) When the barrier needed to move the increasing coherent volume V becomes larger than the dislocation barrier the volume will break up into sub-volumes separated by boundaries of sliding dislocations. The scenario is more subtle in three dimensions where it is more difficult to visualize the nature of the plastic deformation that may cut up the coherent volume. One possibility is that the flux lines cut through each other. In this case the diverging energy barrier will be replaced by the energy scale of flux cutting. At this point it is important to note that vortex glass scaling behaviour can only be observed in a restricted regime. Namely, in the current regime where the volume V (j ) is increasing with decreasing current density j . However, when the energy barrier associated with this volume becomes larger than the plastic barrier the increase in the energy barrier with decreasing j is cut off for currents below some current scale jplas . Lack of resolution in experiments may make it difficult to probe the current scales below jplas . One can then be misled into the false conclusion that the system exhibits a vortex glass transition. A similar difficulty is encountered in numerical simulations of weak pinning centres where it may be difficult to reach system sizes larger than the volume V (jplas ) associated with the onset of plastic deformations. 2.4.5. Plastic flow. Plastic deformations are also of crucial importance at the depinning transition. The effect is most clearly seen at zero temperature. Consider a pinned flux system under application of a transport current, or driving force Fd = Bj . When the driving force equals the volume pinning force, the flux structures start to break away from the pinning centres. The onset of motion can either take place as a coherent displacement Magnetic behaviour of superconductors 1593 of the entire flux strucure or in the form of incoherent motion of parts of the flux array in between islands of pinned vortices. In the latter case plastic shearing of the flux lattice obviously occur. This scenario has long been observed in experiments (Wördenweber et al 1986, Wördenweber and Kes 1986, Bhattacharya and Higgins 1993, and Yaron et al 1995), and was for instance considered theoretically by Kramer (1973). In principle plastic shearing will always occur if the flux system is large enough. This is clearly seen in finitesize scaling studies of computer simulations (Jensen et al 1988a) where rather picturesque channel-flow patterns were observed. The size dependence of the onset of plastic flow is most easily understood by a clear mean-field argument due to Coppersmith and Millis (1991). Consider the balance between the forces acting on a volume V = Ld . Here L is the linear dimension of the d-dimensional volume. There are three different types of forces acting on the volume. Namely, the applied force Fa induced by the applied current. Secondly, the pinning force Fp produced by the pinning centres within the volume V . And finally, the vortex–vortex interaction force Fb acting across the boundary of the volume between the vortex structure inside the volume V and the rest of the vortex structure outside this volume. When the applied force is tuned precisely to the threshold for depinning (i.e. the applied current equals the critical current) these three forces are exactly at balance with each other Fa = Fb + Fp . The boundary force Fb = Fd − Fp is needed to compensate the mismatch between the globally averaged pinning force, which is the threshold force Fthr and the local pinning force Fp , which fluctuates from one sub-volume to another. The deviation between the actual sum of the random pinning forces inside the volume V and the global average will scale as the square root of the number of pinning centres contained in the volume V , see section 2.4.4. Accordingly we have Fb ∼ Ld/2 . There are Ld−1 bonds across the boundary of the volume. Therefore the force fb that each individual bond has to support will scale as fb ∼ L1−d/2 , i.e. the load on the individual bonds increases with the size of the volume which is assumed to act coherently. Since a bond will only be able to support a stress up to a certain value, the coherent volume will break up into smaller pieces. √ Thus, the threshold for the onset of plastic deformations is expected to scale like 1/ L in one dimension. A logarithmic size dependence is expected in two dimensions. This size scaling agrees qualitatively with simulations in one and two dimensions (Jensen et al 1988a, Jensen 1995). In three and higher dimensions the plastic onset of deformations will also occur. However, one has to go beyond the simple average arguments presented here and consider rare events (Coppersmith and Millis 1991). 2.5. Fluctuations in the order parameter In section 2.3 we mentioned that HTS are much more susceptible to thermal fluctuations than low-temperature superconductors. The specific nature of the fluctuations in the order parameter is not yet completely clear. One type of fluctuation is similar to the fluctuations known to occur in two dimensions. For very thin films (which can be modelled as two-dimensional systems) of conventional superconductors it has been know for many years that the superconducting transition is completely controlled by thermal fluctuations (Minnhagen 1987). Since the sample is very thin, the thermal energy close to Tc is able to create vortex excitations. So instead of inducing vortices by an external magnetic field vortex pairs can be thermally excited. Somewhat like the appearance of bubbles in water just below the boiling temperature. In three-dimensional samples vortex-loop excitations play the role of the twodimensional vortex–antivortex excitations. In fact the temperature dependence of the 1594 L F Cohen and H J Jensen resistivity near Tc led Minnhagen (Persico et al 1996) to conclude that the transition (in zero-magnetic field) in bulk HTS may be controlled by unbinding vortex loops by cutting through single superconducting planes. The role of vortex-loop excitations in a non-zero magnetic field have been intensively studied by computer simulations (Chen and Tietel 1995, Caneiro 1995, Nguyen et al 1996) as well as analytically (Tes̆anović 1995). Even without identifying the nature of the fluctuations one can derive a relation for the temperature dependence of the magnetic field at the phase transition. In zero field the symmetry of the Ginzburg–Landau free energy is the same as the symmetry of the threedimensional XY model. The critical exponents of the XY model are well known. The correlation length, for instance, diverges like ξ ∼ 1/|T − Tc |ν , where ν ≈ 0.66, when Tc is approached. We can now attempt to deduce the shift in the transition temperature produced by an applied magnetic field B. The field B, the flux quantum φ0 and the correlation length ξ , can be combined in a dimensionless combination like Bξ 2 /φ0 . From this we conclude that the field at which the transition occurs, must scale like B ∼ ξ −2 ∼ |T − Tc |2ν . Where Tc is the transition temperature in the zero-magnetic field. It is not clear how large magnetic fields can be applied before this scaling relation for B(T ) may break down. However, we shall see below (section 4.2.2) that a relationship B(T ) ∼ |T − Tc |4/3 is in fact consistent with several experiments. 2.6. Anisotropy Most conventional superconductors are isotropic or only weakly anisotropic. The cuprate superconductors are layered structures (perovskites) and therefore inherently anisotropic. The degree of anisotropy varies enormously for the different types of HTS. We will concentrate on YBCO 123 as an example of the superconductors with the smallest anisotropy and we choose BSSCO 2212 as an example of the strongly anisotropic samples. The anisotropy gives rise to new effects. First it makes the flux system more susceptible to finite wavelengths tilt. More dramatically the anisotropy, which has its origin in the layered nature of the HTS materials, may lead to a dimensional crossover. This crossover can be viewed as a change from a situation where the flux system can be treated as consisting of continuous flux lines to a situation where the layered structure of the materials manifests itself more explicitly. Quantitatively, the crossover takes place when the superconducting coherence length perpendicular to the layers, ξ⊥ , becomes of the order of the distance between the layers (Klemm et al 1975). The energy of the continuous flux lines depends on their orientation relative to the crystal axis. This situation is described by a Ginzburg–Landau free energy in which the gradient term is anisotropic. As the effective anisotropy becomes stronger this description is replaced by a free-energy functional describing a set of superconducting layers coupled together via Josephson coupling (Lawrence and Doniach 1971). The continuous flux lines are replaced by stacks of two-dimensional vortices confined to the superconducting layers but coupled across the layers by to the Josephson effect (for a review see Fischer 1993, 1994). 2.7. Thermal equilibrium and non-equilibrium properties The properties of the magnetic flux system inside the superconductor must be divided into two categories: equilibrium and non-equilibrium properties. Furthermore, it is important to distinguish between static and dynamic properties. Magnetic behaviour of superconductors 1595 Among the equilibrium properties one would first like to establish the phase diagram of the flux system as a function of field and temperature. The melting line in the ideal system without any pinning potential, has attracted much attention (Nelson 1988). The structural character of the flux system, i.e. the order transverse to the field direction, and the order along the field direction, in the different phases should be determined. The dynamic properties of systems without pinning are simple. As soon as a current is passed through the material, the Lorentz force will make the flux system flow with a velocity proportional to the current. This results in a current-independent flux-flow resistance (Bardeen and Stephen 1965, Tinkham 1995). In the absence of pinning the melted flux system flows in the same way as a flux lattice when a constant uniform Lorenz force is applied. Only if one applies a Lorentz force that varies in space will the difference between the finite shear rigidity of the flux solid and the zero shear modulus of the liquid flux system show up in transport experiments. The presence of a pinning potential may dramatically change the situation. The pinning potential disturbs the translational order of the flux-line lattice. Even weak pinning can make it difficult to experimentally access the thermal equilibrium states. One signature of this is the observed history dependence of the established flux structure. Recent highprecision neutron scattering experiment on the flux-line lattice in a clean niobium sample by Gammel et al (1994) found that the best orientationally ordered flux-line lattice was established by entering the superconducting state by slowly decreasing the magnetic field through Hc2 rather than field cooling or zero-field cooling followed by an increase of the magnetic field (Mason 1991). The irreversibility line itself separates non-equilibrium configurations for temperatures below the irreversibility line from those above. The Bean critical state (see Tinkham 1995) is a non-equilibrium state and relaxation of the magnetization in this state occurs as the flux lines creep towards equilibrium, as illustrated in figure 4. The rate of relaxation has been viewed as a signature of the specific phase the relaxing flux system is in (Malozemoff and Fisher 1990, Krusin-Elbaum et al 1992). Since one is dealing with an intrinsically non-equilibrium property one must exercise great care in this approach. A similar situation is encountered when one wants to deduce the nature of the phase of the flux system or the nature of the transition between flux phases from transport measurements. The resistance obtained from voltage–current measurements has been used to conclude that the melting of the flux-line system is a first-order transition in clean samples and that the transition becomes second order as pinning becomes relevant (Safar et al 1992c). It has only recently (Jiang et al 1995) been appreciated that the hysteretic behaviour of the Figure 4. An illustration of the Bean flux profiles across a sample during a sweeping up of the external magnetic field in (a) the absence of thermal activation and (b) the presence of thermal activation. The broken curve indicates the profile after some time has elapsed. 1596 L F Cohen and H J Jensen resitivity might be due to non-equilibrium effects rather than the assumed first-order nature of flux-line lattice melting. The out-of-equilibrium driven flux-line lattice has properties of its own that cannot be directly related to the phase of the non-driven equilibrium system. One example is the tendency to ordering of the flux lattice subject to a driving force. As mentioned in section 2.4.5, the spatial fluctuations in the random pinning force can tear the flux lattice into pieces when the applied current is close to the depinning current. As the current is increased the flux lattice is ‘lifted out’ of the pinning potential. This reduces the effect of the pinning forces. The forces acting between the flux lines may then be able to induce more order into the lattice structure than would be observed in the pinned non-driven system in equilibrium. The ordering due to an applied current was observed long ago in neutron scattering experiments (Thorel 1973). Recently, Koshelev and Vinokur (1994) suggested that the applied current might induce a phase transition from a disordered moving system, for currents close to the depinning current, to a moving flux system with crystalline order at larger currents. Although it is clear that some ordering occurs as the drive is increased much current work attempts to clarify the precise details of the nature of the ordering of the moving system (Giamarchi and Le Doussal 1996, Moon et al 1996, Faleski et al 1996, Spencer and Jensen 1997). 2.8. Symmetry of the order parameter The symmetry of the Ginzburg–Landau wavefunction, which describes the motion of the superconducting charges, may be different to that of conventional superconductors. The standard symmetry is the spherically symmetric s-wave, with the notation of atomic orbital theory. Experimental and theoretical studies suggest that in HTS this spherical symmetry is replaced by a ‘four-leaf clover’ symmetry characteristic of d-orbital. Since the symmetry of the superconducting wavefunction is reflected in the symmetry of the vortex core the microscopic symmetry of the Ginzburg–Landau wavefunction may influence the properties of the vortex system (see, e.g. Berlinsky et al 1995). In these circumstances one finds that the ideal hexatic Abricosov lattice is modified. The structure of the vortex lattice depends on the magnetic field and the degree of assumed mixing between the s-wave and d-wave component of the Ginzburg–Landau wavefunction. The larger the magnetic field (or the stronger the d-wave component) the more the vortex lattice changes structure from a triangular lattice towards a square lattice. Note that this will influence the elastic moduli of the flux lattice. Another important difference between the s-wave superconductor and d-wave superconductor is the difference in the quasiparticle spectrum around the vortex line. The density of states around the vortex in the d-wave has four-fold symmetry (Schophol and Maki 1995). The energy gap for creating quasiparticle excitations in the d-wave vanishes along certain directions of the quasiparticle momentum. As a consequence, the temperature dependence of, say, the magnetic penetration depth and specific heat will be different in the d-wave superconductor compared with the s-wave case. 3. Interpreting vortex behaviour 3.1. Transport and magnetization measurements The two main experimental approaches to studying vortex behaviour are direct transport measurements and magnetic studies. The electric field E within the superconductor is explicitly measured in the former approach, but plays a less obvious though no less important Magnetic behaviour of superconductors 1597 role in the latter. This makes a comparison of the data from each technique difficult. Vibrating reed, AC susceptibility, mechanical oscillator and torque magnetometry all belong in the second category but will not be discussed specifically. In a four-terminal transport measurement it is usual to apply current and to measure a voltage. Electric field (E) versus current density (J ) curves can then be plotted as a function of temperature T or magnetic field B. The local slope ρ = dE/dJ of the E–J curve at a fixed voltage (threshold voltage) can be extracted and plotted as a function of the current density. The limitations of transport measurements are related to the practical constraints of attaching contacts, passing large currents through those contacts and measuring small voltages (10−8 V). Consequently transport measurements are usually made at high temperatures or large magnetic fields, where the loss (voltage) is significant. The magnetic measurement has the advantage of being contactless. The magnetic moment can be translated into a current density, but care is needed to take into account sample geometry effects. The electric field is set by the experimental conditions, most usually the sweep rate of the external magnetic field, or more generally by the rate of change of flux through the sample. The local slope of the E–J curve can be obtained from a magnetization measurement in a number of ways. For example, small changes in the magnetic field sweep rate (or electric field) alter the induced current through the sample and variation of current as a function of sweep rate is a measure of the local slope of the E–J curve. (This is also later referred to as the dynamic creep rate.) In theory, magnetic and transport techniques yield identical information. However, at the same fixed magnetic field the transport measurements are restricted to high temperatures and the magnetization to intermediate and low temperatures. As described by Caplin et al (1994), an E–J –B surface can be constructed which schematically illustrates the regimes which each kind of measurement is capable of accessing. It is extremely rare to find transport and magnetization measurements made in overlapping regions of the E–J –T –B parameter space. The effective well depth for pinning centres Ueff (J ), depends on the current density J such that when J = Jc , Ueff (Jc ) = 0. In a transport measurement, currents in the vicinity of the critical value can be applied and the onset of irreversible behaviour (nonohmic E–J curve), can be examined in detail. The transport measurement is used to study the reversible state close to Tc and it is also frequently used to examine the nature of the reversible–irreversible (or solid-to-liquid) transition. In a magnetization measurement the electric fields are lower than in the transport measurement. The irreversible magnetization (or current density) reflects the Bean profile which is set up across the sample. This means that the current density is always less than the critical value and for a given temperature and electric field the critical value can only be approached by increasing the magnetic field. In order to examine the behaviour and transformations deep inside the pinned vortex solid the magnetization measurement is the obvious choice. Figure 5 illustrates the variation of the effective pinning well with applied current. Transport and magnetic measurements can yield complementary information, but this is not always the case. The reason is associated with the relative strengths of the vortex pinning forces and driving force. When the pinning forces are weak compared with the driving forces, the vortex lattice deforms elastically and flows coherently, as often observed in a transport measurement. When the two forces are comparable plasticity and incoherent motion results. This is more frequently observed in magnetic measurements. Once the pinning potential dominates, thermally activated creep is anticipated and again the system may behave elastically. Only in this case may one expect direct complementarity between transport and magnetic measurements. 1598 L F Cohen and H J Jensen Figure 5. A variation of the effective pinning well Up (r) with applied current I . The well is tilted by the current and for I = Ic , the local minima in Up (r) vanishes. 3.2. Transport E–J curves In this section we discuss the essential mechanisms responsible for the different types of current–voltage (I –V ) curves one can observe in superconductors. According to the Josephson relation (Tinkham 1995) motion of the vortices inside a superconductor gives rise to an electric field between two points A and B proportional to the number of vortices crossing the line connecting A and B per unit of time. Let us first consider the zero-temperature case. The disorder in the superconducting matrix can prevent the motion of the vortices as long as the applied Lorentz force produced by the applied current is smaller than a depinning current jc corresponding to the pinning force. The E–J characteristics are therefore of the following form: ( 0 if J < Jc E= f (J ) if J > Jc . Since eventually all vortices must depin and flow with a velocity proportional to J we must always have f (J ) ∝ J for J Jc . The shape of the function f (J ) for currents in the vicinity above Jc depends on the dimension of the system (Larkin and Ovchinnikov 1986) and on the topology of the induced flux motion (Jensen et al 1988b). The flux system can either depin as a coherent structure moving homogeneously through the pinning potential. The pinning force will deform the elastic flux structure but only elastically. Alternatively, motion can occur inhomogeneously when the flux system depins. In this case plastic deformations will take place when the more mobile parts of the flux array passes by the stronger pinned, and therefore less mobile, regions of flux lines. The latter scenario appears to be the most common (Wördenweber et al 1986, Bhattacharya and Higgins 1993, Yaron et al 1995). In two dimensions f (J ) ∝ J if all vortices depin at the same value of J . A nonlinearity in f (J ) arises due to partial depinning as J is increased leading to successively more vortices being pulled into the flow (Jensen et al 1988b) our two-dimensional simulation. The situation is less simple in three dimensions. The nonlinearity can here contain contributions both from elastic distortions of the flux lattice as it flows and from inhomogeneous plastic flow of the flux lines (Larkin and Ovchinnikov 1986, Bhattacharya and Higgins 1993). The above picture is changed at non-zero temperature due to thermal activation of the Magnetic behaviour of superconductors 1599 vortices over the pinning barriers. This can lead to a non-zero electric field even for J < Jc . The actual shape of the I –V curve for currents J 6 Jc is conveniently discussed in terms of the following ansatz E = Bv (3.1) where v is the flux velocity with the following exponential activation form v = ωd exp[−Ueff (J, B, T )/kB T ]. (3.2) The prefactor ω denotes the ‘attempt frequency’, i.e. the number of times the effective flux bundle tries to overcome the barrier per unit time. The factor d denotes the distance moved by the flux bundle as it jumps over the barrier. The product ωd can be measured, but it is difficult to determine each factor separately. Equation (3.1) follows from the Josephson relation which can be expressed in the form E = φ0 nv where φ0 is the flux quantum and n is the density of the vortices which move with velocity v. The detailed dependence of Ueff on J will determine the shape of the I –V curve. The behaviour of Ueff in the limit of J approaching zero is of special importance. If the barrier remains finite Ueff (J ) → U0 as J → 0 equation (3.1) leads to a linear I –V at small current. To obtain this result one must remember to include the contribution in equation (3.1) of flux bundles jumping in the direction opposite to the applied force as well as those jumping with it. Since the applied current lowers the barrier for jumps in the direction of the force and increases the barrier for jumps in the opposite direction we can write the effective barrier for small currents as Ueff = U0 ± J ∂Ueff (0)∂J . Subtracting the contribution of the jumps up against the applied force from the contribution produced by the jumps in the direction of the applied force leads to the famous result (Tinkham 1995, Blatter et al 1994b) E ≈ v exp[−U0 /T ]2J ∂Ueff (0)∂J. (3.3) We conclude that the I –V curve has three regions of different behaviour depending on the sizes of the current compared with the zero-temperature critical current Jc (see figure 6). When J < Jc the electric field is either linear in J (finite barriers) or E vanishes faster than linear when J → 0 (infinite barriers). In the region of currents J ≈ Jc a rapid increase in E is bound to take place as the barriers vanish. Finally, for J > Jc one enters the free flux-flow regime where the current is able to completely overcome the pinning potential. In this regime E ∝ J . 3.3. Flux creep The specific purpose of studying the vortex solid deep inside the irreversible regime, is to determine the functional dependence of Ueff (B, J, T ), the effective pinning well depth, and compare it with theoretical predictions to understand which barriers are involved in different regions of the H –T plane and for different kinds of static disorder. An electric field is associated with a time-varying supercurrent and all thermally activated forward vortex motion can be described by the simple rate equation. −Ueff (J ) Bd −Ueff (J ) dJ exp ∝E= = Bωd exp (3.4) dt t0 kT kT where t0 is the hopping attempt time for a vortex or vortex bundle, ω is the effective attempt frequency and d is the hop distance. The rate equation can be solved approximately once the form of Ueff (J ) is known. 1600 L F Cohen and H J Jensen Figure 6. A schematic illustration of three regions of the I –V curve shape at (a) T = 0 and (b) finite T . There are many experimental artefacts such as irregular sample shapes, temperature instability and magnetic field stability (transients in the magnet itself), which can influence the form of the decay. Recently, spatial relaxation information using Hall probe arrays, have confirmed that global measurements can only be simply interpreted once the DC magnetic fields is greater than the penetration field and in general at fields where surface or geometric barriers are insignificant (Brawner et al 1993b, Abulafia et al 1995). Anderson–Kim thermally activated flux creep. This simple model assumes thermal activation of uncorrelated vortices or vortex bundles over a net potential barrier which depends linearly on applied current density J : J (3.5) Ueff = Uc 1 − Jc where Uc is the J = 0 barrier and Jc is the current density in the absence of thermal activation. Combining equations (3.4) and (3.5) t kT ln . (3.6) J (T , t) = Jc 1 − Uc t0 To eliminate the unknown Jc in equation (3.5), it is convenient to evaluate a normalized relaxation rate defined as the logarithmic derivative of the magnetization or current density Magnetic behaviour of superconductors 1601 (in the Bean critical state the magnetization is proportional to the current density J ) S=− −d ln(J ) −d ln(M) 1 d(J ) ≈ = . J d ln(t) d ln(t) d ln t (3.7) Long-time relaxation. The most common way to examine flux creep by magnetic measurement is to study the long-time decay of a magnetization signal. The magnetic field is applied to the sample at a certain rate (dH /dt), setting a certain electric field through the sample. The field sweep is stopped at a desired value of magnetic field. As discussed by Gurevich et al (1991, 1993), an initial ‘settle time’ has to be allowed for while the electric field spatially redistributes through the sample as a consequence of the change of driving conditions from the initial voltage driven to the final current driven situation. The decay of magnetization dM/dt, now sets the electric field across the sample and can in principle be translated as a journey down the E–J curve. If the decay is approximately logarithmic in time, the slope of dlnM/dlnt is approximately equivalent to the inverse slope of the ln E–ln J curves over the range of electric fields associated with the experiment. For further details see Zhukov (1992). From equation (3.6), we can write −1 kT kT ln(t/t0 ) . (3.8) 1− S= Uc Uc Abulafia et al (1995, 1997) developed an analysis based on the rate equation which allows a direct model-independent determination of the local activation energy and logarithmic timescale t0 for flux creep. Dynamic relaxation. Dynamic relaxation is a technique developed by Pust (1990) and Pozek et al (1991). It monitors the change in the magnetization signal as a function of the sweep rate of the magnetic field. The normalized dynamic sweep rate Q is defined as Q= d ln M . d ln(dH /dt) (3.9) The equivalence of Q and S has been discussed by Jirsa et al (1993) and Schnack et al (1992). The general case for I –V curves of different shapes has been analysed by Zhukov (1992). Deviations from Anderson–Kim behaviour. HTS materials show giant flux creep effects. Deviations from the straightforward logarithmic decay rate predicted by Anderson–Kim have been reported extensively using a variety of standard and novel techniques. Initial reports of non-Anderson–Kim behaviour came from long-time relaxation measurements. See for example Yeshurun and Malozemoff 1988, Malozemoff (1991), Thompson et al (1991a) and Sengupta et al (1993). Other types of measurement confirmed these observations using for example flux creep annealing (Thompson et al 1991b, Sun et al 1990, Maley et al 1990), short-time relaxation (Gao et al 1992), and dynamic relaxation (Pust et al 1990, Zhukov et al 1995, Perkins et al 1996) to name but a few. Zeldov et al (1990) proposed a logarithmic form for the current dependence of the activation energy which has frequently been observed. In order to explain deviations from Anderson–Kim behaviour, Hagen and Griessen (1989), suggested that a distribution of activation energies should be taken into account. In quite general terms, both the vortex glass (Feigel’man 1989) and collective pinning theories describe a form for Ueff (J ) which diverges as J → 0. Feigel’man suggested 1602 L F Cohen and H J Jensen an interpolation formula between the high-current Anderson–Kim limit and the low-current regime which is frequently used, such that µ Jc Uc −1 (3.10) U (J ) = µ J where µ = (d + 2ς − 2)/(2 − ς ) > 0, d is the dimensionality of the relevant vortex bundle volume and ς is the wandering exponent determined by equating the energy of an elastic deformation to the fluctuation in pinning energy. Depending on the dimension etc, µ can take various values for glassy behaviour, see Feigel’man (1989) and for the collective pinning behaviour of an elastic vortex medium, see Blatter et al (1994b). 3.4. Magnetic measurement analysis The description of irreversible phenomena in the mixed state of HTS has to be based upon a self-consistent relation between the current density J and the flux density B, taking into account the nonlinear current dependence of the creep activation barrier. Several related methods have been developed to treat this problem. Hagen and Griessen (1989) developed a model making it possible to calculate a distribution of activation energies for flux motion from magnetic relaxation data using an exact inversion scheme. See also further work by Griessen (1990, 1991). 3.4.1. The Maley method. The Maley method was introduced in 1990 (Maley et al 1990). The basic idea was that one could extract Ueff (J ) directly from the time dependence of M. Based on the general nonlinear form for Ueff (J ) from Beasley et al (1969), Maley et al (1990) wrote down an expression for Ueff (J ) such that Bdω (3.11) Ueff (J ) = −kB T ln |dM/dt| + kB T ln τπ where kB is the Boltzmann’s constant, τ is the thickness of the sample, d is the flux bundle hopping distance and ω is the attempt frequency (see Sengupta et al 1993). In this approach one first calculates T ln |dM/dt| at a given field. Then the data can be directly plotted as Ueff versus (M − Mequ ), at different temperatures. Finally, by adjusting the constant C = ln(Bωd/τ π) all of the data can be made to fall on the same smooth curve and the Ueff − J relationship is obtained. The Maley method assumes that the characteristic current and energy scales are temperature independent. A very useful example of the application of the method is given in Sengupta et al (1993). Maley observed a logarithmic functional form for Ueff (J ) in YBCO agreeing with the earlier Zeldov et al (1990) observation. 3.4.2. The generalized inversion scheme. Schnack et al (1993) introduced the generalized inversion scheme which separates the effective activation energy into an energy term and a current density term so that the rate equation can be written Ueff (J, B, T ) = U0 (B, T )F [J /J0 (B, T )] = kT ln(Bdω/E) (3.12) where U0 is an energy scale, J0 (B, T ) a current density scale and F is a function which describes the J dependence of Ueff (J, B, T ). Within this definition U0 and J0 are closely associated with the pinning mechanism. The Schnack method is model dependent because it requires that Ueff (T ) = [J0 (T )]p . Using this assumption, both Ueff (J, T , B) and the parameter C = (ln(Bdω/E) can be directly extracted from relaxation and current measurements. Magnetic behaviour of superconductors 1603 3.4.3. The magnetic scaling analysis. The concept of magnetic scaling originates from the observation that M–H loops at different temperatures can be brought to lie on top of each other, producing a unique curve, if both the M- and H - axis are normalized appropriately. This reflects one dominant physical process determining the behaviour in the temperature regime of interest. Magnetic scaling has been reported frequently. See for example Zhukov et al (1993), Kobayashi et al (1993), Oussena et al (1993) or Klein et al (1994). Scaling relationships are also observed for the creep rate (Zhukov 1994) and the pinning force density, J B (Civale et al 1991a). The magnetic scaling analysis, introduced by Perkins (1995), builds on the Schnack et al (1993) formalism and further considers the geometrical restrictions of moving through the four-dimensional parameter E–J –T –B space. No assumptions are made about the form of Ueff (J ) but the existence of magnetic scaling has certain implications. Incorporating equation (3.12) into the rate equation and under the condition that M(H ) exhibits scaling, it can be shown that U0 (B, T ) and J0 (B, T ) must take the following forms: U0 (B, T ) = 9(T )B n (3.13) J0 (B, T ) = λ(T )B m . (3.14) Differentiation of the rate equation with respect to ln B and ln E leads to the relationship between χln = d ln J /d ln B, and the dynamic normalized creep rate S(B) = [(ln J )/(ln E)]B,T at constant T : χln = m + (nC − 1)Q (3.15) where C = ln(Bωd/E), and Q is the dynamic creep rate. Both χln and Q can be taken directly from magnetization data. 3.5. Critical scaling applied to E–J curves It is believed that in very pure crystals the transition between vortex solid and liquid will be sharp and first order. In the presence of static disorder, it is further believed that this transition broadens and become second order or continuous. If the transition is continuous then the general rules which apply to all critical behaviour can be applied. Fisher et al (1985) gave a general formulation of the scaling at and near a continuous transition. The basic idea is that physical quantities near the transition can be expressed in terms of the appropriate powers of a diverging coherence length ξ and coherence time τ , such that as the transition temperature Tg is approached from above ξ ≈ (T − Tg )−ν and τ ∝ ξ z. (3.16) A current scale Jsc for linear ohmic response is then defined as Jsc = ckB T /φ0 ξ 2 (3.17) here c denotes the speed of light, φ0 is the flux quantum, and kB is Boltzmann’s constant. A non-ohmic response at all current scales is a signature of a critical transition. The E–J characteristic for a d-dimensional sample is predicted to be a power law of the form ln E ∝ [(z + 1)/(d − 1)] ln J. (3.18) For T greater than Tg the characteristic changes from ohmic behaviour for small current densities where ρ(T ) ∝ (T − Tg )ν(z+2−d) = (T − Tg )s (3.19) 1604 L F Cohen and H J Jensen to power law at large current densities. For T less than Tg the E–J characteristic is always nonlinear and for small current predicts a resistivity given by ρ = E/J = ρ0 exp[−(Jc /J )µ ] (3.20) this implies a negative curvature on a ln E–ln J plot. At large current densities it is predicted to again have critical power law behaviour. The crossover current vanishes as (T −Tg )ν(d−1) . These scaling ideas can be applied to any continuous transition. An example is the Kosterlitz–Thouless transition appropriate to two-dimensional thin films where d = z = 2 and E = J 3 (Hebard and Fiory 1982). These scaling laws have been applied to the E–J curves in HTS and the values of the exponent ν, the dimension d and the dynamic exponent z, have been used extensively as evidence for the existence of melting and a transition to a vortex glass phase. The validity of these claims are reviewed in more detail in section 4. Alternative scenarios exist, see for example Kiss and Yamafuji (1996). For T less than Tg critical scaling is not observable at high driving forces. The influence of disorder on the vortex system as it freezes is strongly tuned by the magnitude of the drive current in a manner not addressed by scaling theory. Koshelev and Vinokur (1994) and Koshelev (1996b) investigated this theoretically. They predicted that in the presence of static disorder, freezing will take place into a perfect lattice if the lattice is moving sufficiently fast and that plastic or glassy behaviour will be observed otherwise, depending on the magnitude of the drive current with respect to Jc . In this scenario, a current-driven transition into a perfect lattice can be visualized as occurring well below the freezing transition. 4. Experimental observation of vortex behaviour 4.1. Reversible properties 4.1.1. Hc1 . Measurement of Hc1 is usually made by examining the point of departure from linearity on the initial slope of the magnetization curve. Without pinning there is of course a sharp cusp at Hc1 . In HTS this measurement is difficult because pinning causes only a subtle departure from linearity and deviation from the M = −H is small at Hc1 . For H //c, the sample geometry (which is usually plate-like) and self-field effects contribute to the shape of the initial slope. Surface barriers to flux penetration result in enhanced values of Hc1 and measurements yield an upper bound only (Umezawa et al 1988). Microwave measurements of the change in the penetration depth as a function of DC magnetic field, yield clean data of the first effective field at which vortices penetrate single crystals (Wu (0) = 200 Oe. and Sridhar 1990). For YBCO, Hcc1 (0) = 800 Oe, Hcab 1 Not surprisingly perhaps, given these complications, Hc1 has sometimes been reported ) geometry, not to follow the expected GL form for either the H //c(Hcc1 ) or H //ab(Hcab 1 particularly for (T /Tc ) 6 0.5 (Krasnov et al 1991). It has been reported that within a few Kelvin of Tc , Hcc1 disappears. In YBCO crystals this is interpreted as a thermally induced excitation over the surface barrier (Safar et al 1990, Pastoriza et al 1994b). The situation for BSCCO 2212 from Brawner et al (1993b), is illustrated in figure 7. Brawner et al interpreted the disappearance the penetration field Hp (which can be considered as an upper bound for Hc1 ) as evidence for thermally induced vortex–antivortex pairs (Kosterlitz– Thouless-type transition) or alternatively as being related to order parameter fluctuations within vortices. (Blatter et al 1993). When the field is aligned within 6◦ from the ab planes, a sharp change in the initial magnetization in BSCCO has been observed by Nakamura et al (1993). They interpret this Magnetic behaviour of superconductors 1605 Figure 7. (a) The temperature dependence of the critical current density Jc with an inset showing the field-cooled (5 Oe) magnetization of the crystal. Also the penetration field Hp versus T with the inset showing a suggested phase diagram, both for a BSCCO 2212 crystal with H //c. Below 83.5 K, Hp extrapolates linearly to Tc = 86 K. Hp is absent above 83.5 K. (b) Shows Jc and Hp for a YBCO 123 crystal. Both persist to Tc in this case. From Brawner et al (1993b). as an effective three-dimensional to two-dimensional transition in the shielding current path because the c-axis coherence is lost across the thickness of the crystal. They attribute this to the fact that the Josephson current is suppressed by the entrance of flux between the ab in TlBaCuO-2201 (Hussey et al planes. Similar observations have been reported for Hcab 1 1994, 1996). Changes in the reversible screening current path for H //ab and T close to Tc have not been reported for YBCO crystals. The values of Hc1 in the two field orientations can be used to determine the GL superconducting mass anisotropy ratio γ . The problem of internal sample flatness and external sample alignment when the field is aligned parallel to the sample surface, throws doubt on the precision of many measurements. Nevertheless some consensus has been lies between 4 and 7 for well oxygenated YBCO and between 30 reached. Hcc1 / Hcab 1 and 300 for BSCCO 2212 (Martinez et al 1992). λab and λc are determined from the measurement of Hc1 , from AC susceptibility and from microwave measurement. Typical values are λab = 140 nm, λc = 600 nm for YBCO and λab = 185 nm, λc = 7.5 µm for BSCCO 2212. 4.1.2. Hc2 . The mean-field approximation for Hc2 is not valid in the presence of large thermal fluctuations. Only specific heat measurements will determine unequivocally whether there is a first- or second-order transition at Hc2 . For LTS materials, Hc2 (T ) acts as a field scale for all elastic properties of the vortex system. Whether Hc2 (T ) or some other field line such as the melting line plays this role in HTS vortex behaviour is still unclarified. Two non-calorimetric methods are used to determine Hc2 . A line in the H –T plane has been extracted from resistivity measurements (Mackenzie et al 1994, Smith et al 1994, 1606 L F Cohen and H J Jensen Osofsky et al 1993, 1994) which place a lower bound on Hc2 by identification of the field at which the resistivity reaches 90% of its extrapolated normal state value at that temperature. These measurements either require huge magnetic fields to be applied of the order of 140 T, or HTS samples which have been doped in such a way that their Tc and Hc2 values are greatly depressed. (Note that such manipulation of the charge carrier state in the material either by ‘overdoping or underdoping’ can change normal state properties, by opening up a pseudogap in the normal density of states. The pseudogap has been studied by specific heat capacity, NMR, inelastic neutron scattering and thermo-electric power. The anisotropy of the superconducting properties are also modified by doping.) The form of the line extracted from the resistivity curves is similar to the irreversibility line and indeed probably reflects the irreversibility line rather than the Hc2 . These resistive measurements are also hampered by the large broadening of the transition in a magnetic field due to flux flow. Even as a very approximate lower bound it is nevertheless clear that Hc2 (T = 0) must indeed be very large. Another procedure used to estimate Hc2 is to find the temperature at which the reversible magnetization approaches zero for a given field. This can be a problem for HTS materials because many samples have rare earth ions which are paramagnetic. Hc2 (T = 0) can been estimated from the reversible magnetization close to Tc (Welp et al 1989), using either a linear analysis (see figure 8) or a more complex procedure (Hao and Clem 1991). According to the theory of Werthamer et al (1966), δHc2 . (4.1) Hc2 (0) = 0.71Tc δT Tc This gives Hc2 (T = 0) = 140 T for well oxygenated YBCO 123. There are problems in using this method for BSCCO, because of broadening of the transition in magnetic field. is used to estimate γ of the order of 5 in YBCO 123 and 30 in BSCCO The ratio Hcc2 /Hcab 2 2212. Measuring Hc2 = φ0 /2πξ 2 yields ξ . In YBa2 Cu3 O7 , for H //c, ξab = 15.4 Å. As oxygen is removed ξab remains at this value initially and then it increases significantly (Ossandon et al 1992, Zhukov et al 1994). M(T , H ) is at least an order of magnitude smaller when H //ab and therefore Tc (H ) is rather ill-defined. This means that even in the least anisotropic material, measurements of Hcab (0) are not particularly accurate. For 2 well oxygenated YBCO, ξc = 3–5 Å. This value is to be compared with the distance between the superconducting layers which in YBCO 123 is 8 Å or the total unit cell length of 11.7 Å. It is thought that because the Cu–O chains of fully oxygenated crystals are also superconducting (Kresin and Wolf 1992), and it is the chain–plane separation 4 Å which should be compared with ξc = 3–5 Å. In BSCCO 2212 ξab = 10 Å implying from the measurement of γ that the value of ξc is much smaller than the interplanar spacing. In this case the GL anisotropic three-dimensional model breaks down and must be replaced by a quasi-two-dimensional model (Lawrence and Doniach 1971). 4.1.3. Influence of columnar defects. Thermal fluctuation effects are particularly pronounced in the most anisotropic materials. Kes et al (1991) measured a weak fluctuation contribution to the reversible magnetization up to 20 K above the field-dependent transition temperature Tc (B). The reversible magnetization at different magnetic fields applied along the c-axis, becomes field independent at some temperature T ∗ < T ∗ , where T ∗ is known as the ‘crossing point’ (Kes et al 1991, Tesanovic and Xiang 1991). Below T ∗ , vortex positional fluctuations (phase fluctuations) modify the field dependence of the magnetization and suppress the pinning critical current density (Feigel’man and Vinokur 1990). At T ∗ the Magnetic behaviour of superconductors 1607 Figure 8. (a) Temperature dependence of the reversible magnetization in YBCO in magnetic fields of 50 gauss and 5 T with the field applied parallel to the planes. (b) Temperature dependence of the magnetically determined upper critical field. The slopes of linear extraction are indicated. The broken lines represent the upper critical points taken from resistive zero points. After Welp et al (1989). magnetization becomes field independent because the logarithmic field dependence of the mean-field magnetization is cancelled by the same logathimic dependence in the entropy contributions of pancakes decoupled along the c-axis (Bulaevskii et al 1992, Tes̆anović et al 1992). The observation of the crossing point is theoretically considered to be the strongest support for the discrete nature of the pancake vortices. In columnar defected BSCCO 2212 crystals this behaviour is modified, see Bulaevskii et al (1996). A maximum is observed in M and the crossing point phenomenon is suppressed 1608 L F Cohen and H J Jensen for fields 0.2Bφ 6 µ0 H 6 Bφ . Bφ = nφ φ0 is known as the matching field associated with a certain density of columnar tracks, nφ . See van der Beek et al (1996), Qiang Li et al (1996) and Pradhan et al (1996). It is still unclear exactly what role columnar defects play and whether they effectively enhance coupling between layers and correlations along the c-direction (Bulaevskii et al 1996). The absence of the crossing point phenomenon suggests that the random distribution of columnar defects suppresses the interaction between fluctuations in the critical regime (van der Beek 1996). The crucial message at this point is that it is clear that vortex pinning not only affects irreversible magnetic properties but can also affect its thermodynamic properties. 4.2. The irreversibility line In 1988, Yeshurun and Malozemoff had already recognized the existence of giant flux creep in HTS. It was quickly established that a line existed well below Hc2 that separated reversible from irreversible magnetic behaviour. Below it, because irreversible magnetization was observed, vortices were thought to be pinned and this pinning was associated with a vortex solid. Above the line, the vortices were considered to be unpinned and under the influence of a driving force—flux flow or a liquid state existed. As we discussed in section 2, these initial conclusions have since been re-examined because pinned liquids might exist below the irreversibility line and an unpinned perfect Abrikosov lattice might exist above it. Over the past eight years there has been an active discussion as to whether the Hirr line represents simple depinning or whether it is coincident with melting or decoupling. Cooper et al (1997) presented experimental evidence which appears to show that thermodynamic fluctuations play an important role in determining the irreversibility line and reversible magnetizatisation in YBCO 123 crystals. Cooper et al (1997) found that the form of Hirr (T ) is consistent with the three-dimensional XY model which is based on fluctuations of the order parameter. As vortices depin (or become pinned), dissipation associated with their movement generates an increase (decrease) in voltage. The change in the vortex properties at Hirr may reflect the existence of a phase transition. However, the disappearance (or appearance) of an irreversible signal is intrinsically non-equilibrium and depends on experimental method and criteria. For clarity, we discuss the evidence for melting and decoupling in separate sections. 4.2.1. The position of the Hirr line in the H –T plane. Optimally doped rare earth (Re)BCO 123 and BSCCO 2212 crystals have similar Tc values and yet the position of their Hirr lines in the H –T plane are strikingly different, as shown in figure 9. Assuming that Tc reflects the strength of the in-plane superconductivity, then the Hirr line is clearly affected by other factors such as the superconducting anisotropy ratio γ , and the c-axis coupling strength. At T /Tc = 0.75, Tallon (1994) found that the position of the Hirr line is exponentially dependent on the inter-plane or plane–chain separation for many families of optimally doped materials as indicated in figure 10. The exponential relationship strongly suggests a weak link plane–plane or plane–chain coupling mechanism. (Implying that the chain is also superconducting). The Tallon group have also shown that by careful cation substitution and oxygen doping, Tc can be kept constant while the c-axis coupling strength is weakened. In these samples Tc is constant but Hirr (T /Tc = 0.75) falls, strongly supporting the premise that this is the main difference between the YBCO and BSCCO Hirr lines. The position of the Hirr line in the H –T plane, also depends on the vortex and defect dimensionality and the defect density. Columnar defects produced by heavy ion irradiation, Magnetic behaviour of superconductors 1609 Figure 9. The irreversibility field versus reduced temperature T /Tc for a series of deoxygenated TmBaCuO7 d (123) crystals (+) with increasing oxygen from left to right across the graph, a YBaCuO8 (124) sample () and a BiSrCaCuO 2212 crystal (*), using a criteria of 10 A m−1 cut-off in the closing of the M–H -loops. After Cohen et al (1994b). enhance the Hirr line, as illustrated in figure 11, for the relatively anisotropic Tl-2223 and Tl-1223 systems. It is energetically favourable for pancake vortices to line up along the columnar track, and by doing so phase fluctuations between planes are suppressed. In this way, although the coupling between the planes is not necessarily altered physically, the system can mimic the vortex behaviour in more three-dimensional systems. In YBCO 123, unidirectional enhancement of the Hirr line along the direction of the columnar was interpreted as a reflection of the line nature of the pre-existing vortex structure (Civale et al 1991a, b). When 200 nm thick YBCO films were irradiated with 5 GeV Pb ions to produce columnar defects, both the screening current and the Hirr line were enhanced preferentially when the vortices were aligned parallel to the columns (Prozorov et al 1994, Fischer 1992). BSCCO 2212 crystals have significant uniaxial enhancement of Hirr due to columns above 40 K as shown by Klein et al (1994, 1993b), and Thompson et al (1992). This is illustrated in figure 12(a). It is generally accepted that below 40 K the vortices become more two-dimensional-like so that isotropic point-defect pinning becomes more favourable. However, columnar defects are thought to suppress phase fluctuations and it is possible that at lower temperatures this is simply no longer significant. The Hirr line and the screening current density J are greatly enhanced over the unirradiated crystal down to 15 K (Klein 1993b). Below 15 K the pre-existing point defects are more effective than the columns at pinning defects. Zech et al (1995) have demonstrated that the unidirectional enhancement also disappears within 10 K of Tc , suggesting that thermal fluctuations eventually reduce the usefulness of the columns. Note that columnar defects influence the behaviour of phase fluctuations as reflected by reversible magnetization properties at temperatures approaching Tc . Gray et al (1996, 1997) discussed the contradition of three- 1610 L F Cohen and H J Jensen Figure 10. (a) Temperature dependence of the irreversibility field H ∗ for various materials. (b) H ∗ at T /Tc = 0.75 versus separation of CuO2 planes (di ) for various materials. Magnetic behaviour of superconductors 1611 Figure 11. The Hirr line enhancement resulting from columnar defects of different densities, in Tl-2223 and Tl-1223 systems. After Brandstätter et al (1995). dimensional-like uniaxial enhancement due to columns at temperatures within the twodimensional thermally activated limit (see sections 4.1.3, 4.3.1 and 4.4.2). The interplay between different types of defects which may be present in a crystal such as point defects (e.g. oxygen vacancies) and planar defects (twinning planes) makes it impossible to summarize all possible behaviours. Twin boundaries are reported to act as strong pinning sites in some situations. However, Oussena et al (1995, 1996) reported that in unidirectional microtwinned crystals at low magnetic fields the twin boundaries act as vortex channels. Figure 13 illustrates the case highlighted by Jahn et al (1995) who found that in YBCO 123, point-like defects can depress the Hirr line without significantly lowering Tc , whereas correlated disorder such as intrinsic pinning of the planes or twin boundaries enhance the Hirr but lower the screening current density. Recently, Flippen et al (1995) showed that even surface damage can significantly alter the pinning properties of YBCO single crystals which have very low twin boundary densities. 4.2.2. The temperature dependence of the Hirr line (H //c geometry). Almasan et al (1992) suggested that the Hirr (T ∗ ) line obeyed a scaling relationship universal for all HTS, whereby at temperatures above T ∗ /Tc > 0.6 the Hirr (T ∗ ) exhibited a power law dependence of the form Hirr (T ∗ ) = (1 − T ∗ /Tc )m (4.2) where m = 32 and T ∗ (H ) is the temperature at which the resistance R(H, T ) drops to 50% of its normal state value, at a fixed field, as shown in figure 14. At lower temperatures a crossover to a more rapid dependence occurs as shown in figure 14 (see the break away on the right-hand side of the figure). Although such scaling exists, it is not as general as first implied by Almasan et al (1992). For example the Hirr line of optimally doped YBCO 123 shows no upturn down to the lowest temperatures measured. The Hirr (T ) of BSCCO 2212 is illustrated in figure 15, after Schilling et al (1993). It follows a power law form with an exponent m = 2 at high temperatures and a much stronger upturn at low 1612 L F Cohen and H J Jensen Figure 12. (a) Magnetization curves at 40 K and 60 K for a BSCCO 2212 crystal which has been irradiated with Pb ions at 45◦ to the c-axis. The field is applied at +45◦ and −45◦ to the c-axis. After Klein (1993b). (b) The effective enhancement of the Hirr line in BSCCO 2212 crystals for continuous columnar defects produced from Pb ions which are effective below about 60 K and Xe ions which create cluster defects. temperatures than found in the intermediately anisotropic materials. The significance of the low temperature upturn in Hirr , has been interpreted as a softening of the elastic parameters of the vortex system making the existing pinning sites more effective and enhancing the Hirr line. Schilling related this softening to a dimensional crossover in the vortices from three-dimensional at higher temperatures to two-dimensional at lower temperatures. The Hirr line as measured by torque magnetometry (refer to Farrell 1994) coincides with sharp features in the resistive transition as measured by Safar et al (1992c), in untwinned YBCO crystals with very low point-defect density. It appears that in this case Hirr coincides with the melting line, i.e. that pinning disappears at the solid-to-liquid transition. Once static disorder is introduced by irradiation or the underlying static disorder becomes effective because the system is cooled, the melting line and the bulk-depinning line separate. This is also consistent with the predictions of the three-dimensional XY model as discussed in a brief review by Cooper et al (1997). The three-dimensional XY model Magnetic behaviour of superconductors 1613 Figure 12. (Continued). is based on fluctuations of the order parameter, and predicts a melting line of the form described in equation (4.4) with α = 1.3. Cooper et al found that both the Hirr line and the reversible magnetization scale according to the predictions of the three-dimensional XY model. Several groups have shown that the electronic specific heat (Overend et al 1996 and references therein), magnetization (Hubbard et al 1996, Liang et al 1996) and electrical resistivity (Howson et al 1995) obey the three-dimensional XY scaling laws associated with critical thermodynamic fluctuations. However, the coincidence of melting and depinning is strongly disputed for BSCCO crystals (Farrell et al 1995, 1996). Zeldov et al (1994) have shown that in perpendicular applied magnetic fields, vortex penetration is delayed significantly in disk-shaped samples due to the presence of a potential barrier of geometric origin. The geometric barrier produces hysteretic magnetization in the absence of bulk pinning. Majer et al (1995) used local Hall probe measurements to show that when a BSCCO 2212 crystal is shaped so that its surface gradient is greater than that of the ellipse making up its extremal dimensions then the Hirr line due to the geometric barrier is reduced to zero. This is illustrated in figure 16. The jump in magnetization marked in the figure by Hm is associated with the melting line. This study suggests that in BSCCO 2212 an Hirr line (although not the one associated with bulk pinning) can be positioned quite separately on the H –T plane to the melting line. However, the topic is controversial and may only be resolved when there is an improvement in experimental sensitivity. 4.2.3. The Hirr line as a function of magnetic field orientation. Remembering that the G– L parameters are anisotropic, one can nevertheless learn about vortex dimensionality and the influence of defect dimensionality on vortex behaviour, by varying the magnetic field orientation with respect to the ab planes. In transport measurements the magnetic field and current directions can be controlled separately. In a magnetization measurement this is not the case and the screening current is induced in a direction either perpendicular to 1614 L F Cohen and H J Jensen Figure 13. Magnetization loops for two YBCO crystals at 77 K. W1 is a twinned crystal with very low density of point defects and W2 is doped with Sr and has a high density of point defects. W2 shows a large J but a depressed Hirr line compared to W1. After Jahn et al (1995). the applied field or in a direction controlled by sample or defect geometry. The induced screening current should always be perpendicular to the field, a field tilted away from the c-axis will produce two components of current (Gyorgy et al 1989). However, geometry effects in flat samples (Zhukov et al 1997) and linear or planar defects (Küpfer et al 1996), can prevent the magnetic moment from tracking the applied field. Making magnetic measurements in this case requires more care. When the field is close to the ab planes, quantitative analysis is difficult because the vortex system may take a new structure related to the extreme angle. For example, in YBCO crystals at low fields and angles close to the ab planes, high resolution Bitter patterns reveal that the flux-line lattice which lies in the tilt plane takes the form of vortex chains, as a result of attraction within the tilt plane (Gammel et al 1992, Grigorieva et al 1993). In BSCCO similar attractive forces results in the coexistence of two vortex species, oriented parallel to the ab planes and the c-axis (Grigorieva et al 1995). When the field is aligned accurately in the planes, a lock-in transition occurs and has been observed in untwinned YBCO crystals ab (T ) occurs at around 80 K which (Ossandon et al 1992). A crossover in the form of Hirr has been associated with the disappearance of the coherence in the c-axis current (Lacey et al 1994, Ossandon et al 1992). Such a crossover is predicted to occur when the thermal energy washes out the inter-plane coupling (Fischer)- formulae. The variation of Hirr (T ) with θ, the angle between H and the ab plane, can be used to determine the dimensionality of the vortex structure at different temperatures. For example, Kes et al (1990) proposed, that the field component parallel to the c-axis, H sin θ determines Magnetic behaviour of superconductors 1615 Figure 14. Log(H /H + ) versus log(1 − T ∗ /Tc ), where H + is a normalizing field and T ∗ (H ) is the temperature at which the normal state resistance falls to 50% of its normal state value. The full curves in all four figures are power-law behaviour with the exponent m = 32 . (e) is a composite plot of all the data from figures 2(a)–(d ). From Almasan et al (1992). Hirr for quasi-two-dimensional pancake vortices. Various angular scaling relationships have been suggested: Tachiki and Takahashi (1989), Klemm and Clem (1980), Tinkham (1963). These scaling models breakdown in a variety of ways when line-like and planar defects dominate the pinning behaviour or when Josephson kink structures are created at angles close to the planes (Zhukov et al 1996). In general, anisotropic G–L theory suggests that H = HA (sin θ 2 + (1/γ 2 ) cos θ 2 )−1/2 . (4.3) Once γ > 15 this expression is, for small θ indistinguishable from Kes scaling (Iye et al 1992). In YBCO crystals containing a homogeneous point-defect distribution, the Hirr line moves systematically upwards in the H –T plane as the field is swung away from the c-axis. It is found to be about five times higher when it lies approximately in the plane, in agreement with equation (4.3) and measured values of γ (see Angadi et al 1991). 1616 L F Cohen and H J Jensen Figure 15. The Hirr line, interpreted as a melting line, is plotted on a logarithmic field scale for BSCCO 2212 crystals. The arrows indicate the temperature range where the data has been fitted to expressions for two-dimensional (low temperature) and three-dimensional melting lines. The full curve corresponds to a description in terms of a Josephson coupled layered superconductor (JCLS). The broken curve is a quadratic fit to the data near Tc . After Schilling et al (1993). Schmitt et al (1991) showed that for BSCCO films measured in the maximum Lorentz force geometry (H ⊥ J ), the maximum critical current in the planes is obtained when H //ab and it is field independent due to strong intrinsic pinning by the planes. Equation (4.3) does not describe the behaviour well because γ reflects the mass anisotropy of the superconducting electrons and not the strongly anisotropic pinning and flux creep effects found in BSCCO (see Kobayashi et al 1995a, b). The difference between the pinning (or current) anisotropy and γ has been explored in YBCO crystals as a function of oxygen by Thomas et al (1996). Nelson and Vinokur (1993) predicted that correlated disorder such as columnar defects, twin or intrinsic planes, can transform the vortex solid into a Bose glass. The Bose glass is predicted to have a distinctive Tirr (θ) cusp-like dependence when the field component along the disorder is fixed, and is less than or equal to the matching field (one vortex per column). The predicted behaviour is shown in figure 18. Tirr increases with increasing ab plane field component (increasing tilt) for a vortex glass and decreases with increasing tilt for a Bose glass. The behaviour of Tirr can be explained more simply. In the unirradiated crystals a simple vector model argument along the lines of equation (4.3) can be used. In the irradiated crystals, the reduced effectiveness of the columns to pin the vortices as the field is tilted away from the line-like defect, would produce the cusp-like feature in Tirr . The cusp-like feature has been observed in Tirr in columnar-defected YBCO crystals (Jiang et al 1994, Reed et al 1995), intrinsic planes of YBCO (Kwok et al (1992) as shown in figure 17(b)) and columnar-defected BSCCO (Seow et al 1996) crystals. The latter is illustrated in figure 28. The enhanced Tirr (θ) is only observed at high temperatures. Zech et al (1995) have shown how Kes scaling of the Hirr line breaks down in irradiated BSCCO Magnetic behaviour of superconductors 1617 Figure 16. Local magnetization loops B–Ha versus Ha in BSCCO crystals of (a) platelet and (b) prism shapes at T = 80 K. The platelet crystal shows hysteretic magnetization below the irreversibility field HIL . In the prism sample the geometric barrier is eliminated and a fully reversible magnetization is obtained at temperatures above 76 K. After Majer et al (1995). crystals as the field is brought into line with the columnar tracks. Küpfer et al (1996) clarified the relationship between the variation of the screening current density J (T , θ, B) and the Hirr (θ, T ) as a function of angle in a series of YBCO crystals with varying degrees of random point-like disorder (vortex glass-like) and twin plane disorder (Bose glass-like). In the case of flux channelling along twin planes, minimum pinning and maximal Hirr line is produced when the field is aligned to the c-axis, as illustrated in figure 19. Considerably more complex behaviour can result when crystals have different types of disorder. 1618 L F Cohen and H J Jensen Figure 17. (a) The transport current of epitaxial BSCCO 2212 films as a function of the angle between the magnetic field and the c-axis. There is a maximum in the critical current when H //ab is field independent. After Schmitt et al (1991). (b) The angular dependence of the vortex lattice melting temperature Tm (θ ) of a twinned YBCO crystal. The inset shows the plot of Tm versus the angle displaying the cusp near H //c. The line is a guide to the eye. After Kwok et al (1992). 4.3. In the vicinity of Hirr If the Hirr line coincides with a well defined melting line Hm , the vortex system will be a pinned solid below the line and an unpinned liquid above it. If melting and decoupling occur simultaneously with the Hirr line the pinned solid will transform above the line into a two-dimensional lattice or gas (as determined by the density of vortices). These pictures are straightforward to visualize but experimental differentiation between them is not easy. How does one set about doing this? At the transition between an unpinned vortex lattice and an unpinned vortex liquid the ab plane resistivity ρab should register a change in shear viscocity (see section 4.5.1). Within each regime ρab is ohmic, with a value which is some fraction of the normal state resistivity. More usefully perhaps, the c-axis Magnetic behaviour of superconductors 1619 Figure 17. (Continued) correlation (or line integrity) is lost in the liquid state when the vortices are entangled as a result of thermal disorder (Nelson 1988). Hence, a non-Lorentz geometry with J and H //c, ρc will be finite in the vortex liquid (where phase slip processes associated with vortex cutting and rejoining will create dissipation). For a three-dimensional vortex lattice ρc = 0. A quasi-two-dimensional vortex lattice also has only weak c-axis correlation. Differentiating between a quasi-two-dimensional solid and an entangled three-dimensional liquid, is yet more subtle. 4.3.1. The disentangled vortex liquid and the c-axis correlation. Fendrich et al (1995) examined the question of pinning in the liquid state by looking at the effect of pointdefect electron irradiated on clean untwinned YBCO crystals. In the unirradiated crystals a linear free flux flow resistivity was found to follow the simple Bardeen Stephen model ρff = (B/φ0 Bc2 )ρn , where ρn is the normal state resistivity. In the irradiated crystals, the linear resistivity was depressed relative to the unirradiated at the same reduced temperature, below a certain temperature denoted Tp . As the defects in this sytem were uncorrelated, individual vortex pinning was found to be too small to explain the decrease of ρ in the postirradiated liquid. The authors attributed the depression of ρ to an increase in viscosity as a result of increased liquid entanglement in the irradiated crystals. They took into account the viscous shear processes that take place in liquid flow and related this to a plastic energy proportional to the energy involved in vortex cutting and recombination. They also found that the resistive transition was broadened as a function of the point disorder (see section 4.5.1) and the Hirr line was depressed. Correlated disorder will disentangle, localize (or pin) the vortex liquid as claimed by Civale et al (1991a, b) in the case of columnar defects, and by Flesher et al (1993), in the case of twin boundaries. The correlation along the c-axis or degree of entanglement can be determined by measuring ρc directly. A pseudoflux transformer set up along the lines of Giever (1965), measures vortex line tension which also reflects the strength of the c-axis correlation. In the pseudoflux transformer configuration (figure 21), eight electrical contacts are attached to the crystal. A driving current is made to enter and leave the sample through leads at the top surface of the sample and external magnetic field is applied perpendicular 1620 L F Cohen and H J Jensen Figure 18. (a) The phase diagram in the (T , Hperp ) plane with the field Hz along the direction of the fixed correlated disorder. The crystalline phase is an Abrikosov lattice for fields tipped away from a single family of parallel twins or it represents a smectic-like phase for columnar pins or a mosaic of twin boundaries. (b) The phase diagram in the limit of strong correlated disorder. Interactions are important in determining the localization length and transport at intermediate current scales above the broken crossover curve B ∗ . After Nelson and Vinokur (1993). (c) The resistively determined irreversibility line Tirr before (BI) and after irradiation (AI). Also shown is the Bose glass temperature TBG . (d ) Angular dependence of Tirr in an irradiated crystal (lefthand axis) compared to an unirradiated crystal for fixed Bz parallel to the c-axis when the field component parallel to the ab-planes Bperp is increased. After Seow et al (1996). Magnetic behaviour of superconductors 1621 Figure 18. (Continued) to the superconducting CuO planes. The voltage drop at the surface Vtop and at the bottom Vbot of the sample is measured simultaneously. The system is studied in the linear I –V region at T > Tirr as well as in the nonlinear I –V region. (The irreversibility line is defined here simply as the temperature below which limJ →0 dE/dJ = 0.) Lopez et al (1994a, b, 1996) used the DC flux-transformer configuration to compare the behaviour in twinned and untwinned YBCO crystals, and there are quite remarkable differences between the two. In the twinned crystals, Vtop 6= Vbot for temperatures T > Tth (th for thermal cutting of the vortices). The dependence of Tth is studied as function of the external magnetic field. The electric field induced by the motion of vortices are according to the Josephson relation (see e.g. Tinkham 1995) given by E ∝ nv where n denotes the density of the moving vortices and v is the velocity with which they move. Since the same number of vortices are induced in the top of the sample as in the bottom, the authors conclude that the velocity of the segments of the flux lines in the top of the sample must be 1622 L F Cohen and H J Jensen Figure 19. Angular dependence of the normalized hysterisis width δm(φ)/δm(φ = 0) taken at the peak position (open squares). Measurements are made at (a) 80 K for crystal number 1 and 77 K for all other crystals. The full squares represent the corresponding values of the normalized irreversibility fields Birr . The crystal in (a) has strong twin-plane pinning, (b) and (c) have twin planes and point defects, (d ) has point defects only. After Küpfer et al (1995). different from those in the bottom when Vtop 6= Vbot . The current voltage curves (IVCs) are linear down to temperatures Ti which is associated with the irreversibility line. Above Tth the coherence across the thickness of the sample is lost and flux cutting must occur. Below Tth the flux lines are coherent across the sample thickness. However, between Ti and Tth , at large enough currents, denoted Ic (the cutting current), Vtop 6= Vbot . Ic is dependent on the magnetic field and thickness of the crystal. Figure 22 shows a flux transformer in twinned and untwinned crystals. Twin planes acting to disentangle vortex liquid. After Lopez et al (1996). For twinned crystals, the region between Ti and Tth is thought to be a disentangled vortex liquid, where the twin boundaries stabilize the disentanglement (i.e. they hold the vortices straight), but only effectively up to a finite shear force associated with Ic . In the untwinned crystal a sharp resistive transition is observed at a temperature Tm . Below Tm , Vtop = Vbot at all drive currents, and the IVCs are nonlinear for any current drive. Above Tm , Vtop 6= Vbot and the IVCs are linear at all currents. In the untwinned crystal, the authors attribute the behaviour above Tm , to an entangled vortex liquid and below Tm , to plastic motion of a vortex solid which is correlated in the c-direction. Interpretation of flux transformer data is complicated by the question of whether non-local resistivity has to be taken into account. Safar et al (1994) claimed that in heavily twinned YBCO, the line-like nature discussed above implies that a simple local electrodynamic picture (that the local electric field is simply determined by the local current), breaks down. However, Eltsev and Rapp (1994, 1995a, b) successfully explained their Magnetic behaviour of superconductors 1623 Figure 20. (a) Shows the resistivity of an untwinned YBCO single crystal versus normalized temperature, at 4 T for H //c. The sharp resistive transition at Tm occurs in the virgin crystal. The transition broadens when the crystal was irradiated with 1 MeV electrons. (b) Shows voltage–current characteristics before and after electron irradiation at several different reduced temperatures T /Tc for H //c. After Kwok et al (1993). 1624 L F Cohen and H J Jensen Figure 21. Sketch of the Giaever DC flux transformer and pseudo DC flux transformer contact configuration. In the latter, due to the anisotropy in the resistivities, injecting current in the top face produces an inhomogeneous current distribution. After de la Cruz et al (1994b). Figure 22. Normalized resistance versis reduced temperature of twinned and untwinned single crystals measured in a flux transformer geometry. After Lopez et al (1996) who argue that twin planes act to disentangle vortex liquid. Magnetic behaviour of superconductors 1625 transformer measurements in twinned YBCO by a local anisotropic superconductor. There is a contradiction here. Interpretation of flux transformer data hinges on the temperature dependencies of ρab and ρc . In the twinned YBCO crystals Safar et al (1994), found that ρc goes to zero at Tth . (This data is only shown on a linear scale so results may be inconclusive.) In a local picture, if ρc is finite, the current is distributed across the crystal leading to flux cutting and Vtop 6= Vbot . If ρc = 0, then the same current runs along each ab plane and one would expect that Vtop = Vbot . Those that believe that the local resistivity picture is correct would say that twin planes reduce the thermally induced phase fluctuations, so that ρc drops more rapidly in crystals with correlated defects. BSCCO crystals, unlike YBCO are not naturally twinned, so that the only static disorder is point disorder. The flux transformer data is much more straightforward to interpret (Safar et al 1992a, Busch et al 1992, Wan et al 1993). The in-plane dissipation associated with an ab plane electrical transport current is found not to be correlated over the sample thickness over wide portions of the H –T plane. The flux transformer data shows that Vtop and Vbot are never equal and in increasing field, the difference between them increases (see figure 23). It is concluded that the vortices, must cut and reconnect during transport, suggestive of twodimensional vortices. A peak is seen in Vbot which is depressed in low fields. The peak is probably associated with the temperature at which Josephson coupling becomes important. Related to this, Fuchs et al (1996) showed that Vtop = Vbot at the melt temperature in BSCCO. Doyle et al (1996) made transport measurements in the flux transformer and c-axis geometries in heavily columnar defected BSCCO crystals. They found that as in twinned YBCO, there is a range of fields and temperatures where Vtop and Vbot show close correspondence. They successfully described their transformer data using a local anisotropic electrodynamics picture only and compared with directly measured ρc data. In the fieldtemperature region where the top and bottom voltages match they show that ρc vanishes faster than ρab . To support the case where columnar defects enhance the c-axis correlation Doyle et al also looked at the angular dependence of extracted ab and c-axis resistivities as shown in figure 24. Both components are reduced for B// defects, as expected for strong uniaxial pinning and finite-line tension. Apparent activation energies are obtained from linear fit to Arrenhenius plots of the resistivities as a function of temperature. They found that Ueff is field independent below the matching field, suggesting that vortices are indeed ab c ' Ueff . If columnar defects enhance localized on defects. Above the matching field Ueff the line-like nature of BSCCO 2212 as suggested by uniaxial enhancement above 40 K (as discussed in section 4.2.1) this appears to be inconsistent with the local electrodynamic picture proposed by Doyle et al . The coincidence of the three-dimensional-like behaviour reflected by uniaxial enhancement and two-dimensional-like scaling (Kes et al 1990), has been discussed by Gray et al (1996, 1997) (see section 4.4.2). They concluded that vortex– vortex interactions have not yet been taken into account properly and that the contradiction can be reconciled by considering that the vortices are not strictly two-dimensional, but that the CuO2 layers remain very weakly coupled, up to the highest temperatures. 4.3.2. Peak effects in I (T ). Kwok et al (1994) reported a peak in J (T ) below the sharp resistive transition at Tm in twinned YBCO crystals, as illustrated in figure 25. Tang et al (1996) reported similar behaviour when the field is aligned parallel to the ab plane. This behaviour is interpreted as a softening of the shear modulus C66 which indicates a precursor to melting (of Larkin domains, Larkin and Ovchinnikov (1979)) in the presence of correlated disorder (Larkin et al 1995). The low-Tc -layered superconductor 2H-NbSe2 1626 L F Cohen and H J Jensen Figure 23. The flux transformer data from two high-quality BSCCO 2212 single crystals. The temperature dependence of the top and bottom voltages Vt and Vb respectively. After Safar et al (1992a). shows possibly related results for free flux-flow Hall effects, see Bhattacharya et al (1994). Recently, Yaron et al (1996) reported that small-angle neutron scattering (SANS) from the flux-line lattice in high-quality niobium crystals reveals drastic structural disordering near the peak effect seen in the transport critical current. The flux-line lattice appears to disorder as a function of applied field in a two-step process characterized first, by a complete loss of long-range translational (hexatic glass) followed by a subsequent loss of orientational order (vortex glass). 4.3.3. Vortex slush—melting as a function of drive current. The observation of multiple regimes of behaviour caused by the interplay of pinning and current drive may relate to the most recent discussions concerning plastic flow deep in the solid state. Worthington et al (1992) first introduced the term vortex slush while investigating the effect of disorder. Three YBCO crystals were examined. A ‘clean’ crystal with low screening current density which was irradiated with 3 MeV protons with fluence of 1016 cm2 , a crystal which was heavily twinned and point defected, with much higher current density, and finally a crystal which Magnetic behaviour of superconductors 1627 Figure 24. Flux transformer data from a columnar-defected BSCCO crystal with a 0.5 T matching field. The top V23 /I and bottom V67 /I resistances (see figure 21) are marked as points on the graph. The calculated secondary apparent resistance from the top and c-axis voltages at 1 T, are marked as curves on the graph. The inset shows the temperature dependence of the ratio of V67 /V23 and V37 /V23 at 0.5 T for the same crystal. Note that in this experiment the numeration for the bottom voltage probes is reversed compared with that indicated in figure 21. After Doyle et al (1996). was irradiated with 1 GeV Au ions creating columnar defects. For these ‘intermediately disordered’ YBCO crystals, two transformations (or a double shoulder) were observed in the resistivity as a function of current, for small magnetic fields 0.1 T, as illustrated in figure 26. The first being associated with the remnants of the first-order melting transition Tm and the second, at lower temperatures associated with the transition to zero linear ρab at Tg . At higher magnetic fields this double transition was washed out. At temperatures between Tg and Tm , Worthington et al (1992) suggested an intermediate vortex ‘slush’, which had finite-linear resistivity greatly reduced from the flux-flow liquid state above Tm . The authors demonstrated that the upper transition Tm is dependent on the current value or is a non-equilibrium ‘current-induced’ melting transformation. The possibility of additional heating due to collision of vortices with pinning centres was considered by Worthington et al (1992). They assumed that the disorder-induced heating effect grows with increasing current drive. Note that this is in contrast to the Koshelev et al (1996b) model which assumes that the effect of the fluctuating component of the pinning force which produces ‘shaking’ of vortices and an associated ‘shaking’ temperature Tsh is inversely proportional to the drive current and is only a well defined effect for drive currents less than the critical current. Later publications by Safar et al also referred to the vortex slush in YBCO as a region where finite transverse correlation has set in as indicated by a sharp resistive jump, but longrange order has not been established throughout the system. Many crystals show a sharp resistivity jump but ρab remains finite at lower temperatures. This is more likely to occur in irradiated or twinned crystals and is probably not significant in very clean crystals where the onset of long-range order coincides with the resistive jump as discussed in section 4.5. The term vortex slush may describe the same type of inhomogeneous onset of flux flow as studied by Bhattacharay and Higgins (1993) and Yaron et al (1995). 1628 L F Cohen and H J Jensen Figure 25. (a) The Temperature dependence of the resistivity below Tm at different angles of the magnetic field with respect to the c-axis and the twin planes. (b) Temperature dependence of the critical current for different orientations of the magnetic field with respect to the twin boundaries, showing the peak effect just below Tm . The peak is maximum when the field is aligned to the twin planes. From Kwok et al (1994). 4.3.4. Dissipation in highly anisotropic systems. The question has been raised, whether the origin of dissipation in the highly anisotropic systems is Lorentz force determined. Iye et al (1989) and Woo et al (1989) reported that in BiSrCaCuO and TlBaCaCuO 2212 thin films down to 15 K, the DC resistivity was independent of the angle between the magnetic field and the transport current when both lay in the ab plane. The implication was that the resistivity was independent of the Lorentz force. The angular dependence of the resistivity could also be explained by the Kes model (refer to section 4.3.3). In contrast, Palstra et al (1988, 1989) who were studying YBCO 123 single crystals, found a distinct resistance anisotropy, implying that in the more three-dimensional systems the dissipation appears to be Lorentz driven. Silva et al (1995) have shown that the high-frequency surface resistance of YBCO behaves similarly. Kadowaki et al (1994) investigated c-axis transport measurements with the field orientation both parallel and perpendicular to the c-axis. It is found that the current carrying ability in the c-axis is hindered most when the field is aligned parallel to the Magnetic behaviour of superconductors 1629 Figure 26. (a) Typical resistivity versus current-density isotherms for a YBCO crystal after 3 MeV proton irradiation at a dose of 1016 cm−2 . (b) The linear resistivity extracted from the ρ versus J isotherms in (a) at low current. (c) The melting current in MA m−2 versus temperature defined as the current where δ ln ρ/δ ln(J ) is maximal. After Worthington et al (1992). 1630 L F Cohen and H J Jensen Figure 26. (Continued) c-axis and least, when it is aligned along the ab planes. Figure 27 illustrates the huge caxis resistivity broadening in BSCCO in finite field for the Lorentz force-free geometry. It might be supposed that the magnetic field between the planes would decouple the planes and ultimately destroy the Josephson coupling and that the force-free geometry would support the larger current. This is the opposite of what is observed. The experimental observations are understood to result from the fact that when the field is aligned along the c-axis the phase fluctuations are more deleterious to the current. However, it is not completely clear to what extend the G–L anisotropy masks the Lorentz force dependence. The pronounced broadening of the resistive transition is a unique phenomenon. Although there are several explanations based on the traditional idea of vortex motion (Gray and Kim 1993), several novel approaches have been introduced such as vortex–antivortex excitations, thermal fluctuations of flux lines (Fastampa et al 1993) and superconducting fluctuations (Tsuneto 1988, Ikeda et al 1991, Kadowaki et al 1994). See also Koshelev (1996a) for a phase slip model for c-axis resistivity. On a cautionary note, the validity of the Lorentz force–free resistive broadening for the H //c//I geometry will depend on the purity of the current path. Any current deviating along ab planes will generate a force geometry. 4.4. The order of the melting transition It is suggested that the order of the melting transition is a function of the static disorder in the crystal. In clean samples with negligible pinning the vortex solid–melting transition is expected to be a first-order transition between a vortex liquid and a well defined Abrikosov lattice. It is thought that disorder drives the transition second order as assumed in the vortex glass scenario (section 2.4.3). The influence of defects and their dimensionality on the order of the transition and the nature of the glassy or solid phase are key questions still undergoing clarification. If the disorder is correlated in one or more dimensions then it is predicted that different kinds of glasses may occur such as a Bose or smectic glass (see Blatter et al 1994b). Magnetic behaviour of superconductors 1631 Figure 27. The logarithmic high-field c-axis resistivity behaviour of a BSCCO 2212 single crystal for the Lorentz force-free geometry, B//I //c-axis. After Kadowaki et al (1994). A second unresolved issue is whether the melting of vortex lattice occurs by means of a single or two-stage process. A two-stage process could take the form of lattice decomposition into a liquid of vortex lines, followed by a decoupling transition, when thermal excitations destroy long-range correlation parallel to the c-axis. The order of the events depends on field and temperature and the form of the decomposition and decoupling lines. Jagla and Balserio (1997) discussed the circumstances under which c-axis correlation or ab plane long-range order disappears, including how the anisotropy of the system affects the order in which they are lost. 4.4.1. Evidence for a first-order transition. Evidence for a first-order thermodynamic melting line, denoted Tm or Hm originates from coincidence of a sharp, hysteretic, resistive transition, changes in latent heat, a jump or stepwise change in the reversible magnetization M(B) and a frequency-independent peak in AC susceptablility. The resistive transition. There are three features of interest in the resistive transition. The sharp but hysteretic behaviour in low fields, the appearance of one or more shoulders (vortex slush) and finally the broadening of these features in applied magnetic fields (Lorentz forcedriven dissipation). As shown in figure 28, untwinned YBCO 123 crystals with very low disorder, were reported to show sharp hysteretic resistive transitions by Safar et al (1992c, 1993), Kwok et al (1992, 1994) and Charalambous et al (1993). The transformation line Tm (B) inferred from these observations for H //c geometry coincides with the irreversibility line Tirr 1632 L F Cohen and H J Jensen Figure 28. (a) Normalized linear resistance versus temperature for an untwinned YBCO crystal, using a SQUID picovoltmeter. Note the hysteresis. (b) As in (a) but over a wide temperature range. (c) Hysteresis width as a function of the field. After Safar et al (1992c, 1993). extracted from oscillator experiments in similarly clean crystals. Tm (B) fits rather well to some of the melting criteria models (Farrell 1994). There are various theoretical derivations of the vortex-lattice melting line based on the Lindemann criterion, see for example Blatter and Ivlev (1994) and Brandt (1989). Houghton et al (1989) predict a power law form (which is the form followed by Hirr in fully oxygenated YBCO) that is best approximated by Bm (T ) = B0 (1 − T /Tc )α where α 6 2. (4.4) Magnetic behaviour of superconductors 1633 Figure 28. (Continued) Note that this equation has the same form as equation (4.2) which described the observed temperature dependence of the irreversibility line in YBCO 123. This is also consistent with the predictions of the three-dimensional XY model as discussed by Cooper et al (1997). In clean YBCO crystals, it looks as though the Hirr line and the melting line coincide. It is suggested that the resistive hystersis seen in figure 28(a), indicates superheating and supercooling found at a first-order melting transition. Of course it is unclear whether a non-thermodynamic quantity such as resistivity should follow the same hysteretic behaviour as the internal energy. A compelling experimental paper (Jiang et al 1995) addressed this point and concluded that resistive hysteresis is neither a sufficient nor necessary condition for first-order melting. Part of the Jiang experimental results are shown in figure 29. Superheating and supercooling would imply specific sub-loops which were not observed. Waiting at point B0 or C0 indicated on the figure, the resistance was not seen to relax to a new value which would be expected as a result of equilibrating the temperature. These and other observations described in this paper provide counter evidence against the hysteresis width being directly related to the latent heat. Safar et al (1993) found a critical value of magnetic field in YBCO crystals, at which the slope of the apparent phase boundary in the H –T plane changed. At fields greater than the critical value, the sharp resistive transition was broadened and the hysteresis width narrowed, as shown in figure 28(c). By examining the onset of nonlinear resistance Safar et al also suggested that the glass transition Tg lies below the melting transition in crystals which show the sharp resistive transition. Between these two phases the vortex system had properties denoted ‘vortex slush’ where finite linear resistivity existed but was greatly reduced from the flux-flow liquid state above Tm (Worthington 1992). The vortex slush is discussed further in section 4.4.3. Various possible H –T phase diagrams were suggested by Safar et al (1993) as shown in figure 30. Remarkably, a sharp and hysteretic resistive drop in BSCCO crystals has been reported 1634 L F Cohen and H J Jensen Figure 29. The history and time dependence of the resistivity hysteresis. The data points are for partial heating and cooling cycles, the full curves are full heating and cooling data curves. The inset is a schematic hysteresis and the corresponding subloops based on the assumption of a first-order phase transition. After Jiang et al (1995). Figure 30. A composite phase diagram for untwinned YBCO 123. The full circles are the hysteretic melting temperatures Tm . Open squares are the vortex glass melting temperatures Tg . Also shown are Hc2 and contours of constant resistance. The insets show three posssible phase diagrams. After Safar et al (1993). by Keener et al (1997) as shown in figure 31. Sharp features in ρ(T ) have also been reported by Watauchi et al (1996) and Kadowaki (1996). The drop in BSCCO occurs at a much lower resistance than in YBCO (0.02% of the normal state resistance versus about 20% in YBCO) and is much harder to observe. Keener et al (1997) further claimed Magnetic behaviour of superconductors 1635 Figure 31. The temperature dependence of Vab with I = 0.1 mA in magnetic fields of 0, 10, 20, 30, 40, 50, 70, 100, 120 Oe and higher labelled in the figure. The inset shows the electrode configuration. After Keener et al (1996). to observe a two-stage melting transition interpreting a second sharp feature in the liquid phase as suggestive of thermal inter-layer decoupling of vortex lines. In the limit of low current only, the decoupling and melting lines merge. Evidence of a sharp resistive drop in BSCCO 2212 has also been discussed by Fuchs et al (1996), who found that the onset to the resistive drop occurs concurrently with the equilibrium magnetization step/jump discussed in the next section. In the BSCCO 2212 system similar to YBCO 123, there is strong evidence suggestive of a critical point, although it is unclear whether in fact data merely reflect a crossover in pinning properties. The field at which it occurs in BSCCO varies from 30–2000 mT and is sensitive to oxygen content (Ando et al 1995, Khaykovich et al 1996) which may be tuning the intrinsic anisotropy (Kishio et al 1994) or the disorder. Pastoriza et al (1994a), reported a frequency-independent Hirr line extracted from AC suceptability measurements, which became frequency dependent above 36 mT. The reversible magnetization jump also disappears at the same field (as discussed below). In agreement with this, Keener et al (1997), found that the sharp hysteretic resistive drop broadens in fields greater than 70 mT. The crossover field is independent of current, possibly reflecting thermodynamic behaviour. However, the temperature Tm at which the sharp drop in resistance occurs, increases as the current is decreased. This suggests that the resistance drop is associated with a currentdependent shearing mechanism rather than a sudden disappearance of C66 expected at a melting transition. More direct evidence for a loss of shear viscosity has been measured at Tm in a rather novel experiment by Pastoriza and Kes (1995) where parallel tracks of columnar defects were introduced in BSCCO crystals. In this configuration, the restoring shear force of vortices situated in the weak-pinned channels between tracks, were measured by a simple resistance measurement with the applied current perpendicular to the tracks. Within a continuum approximation, the current density Js , which initiates the flow of vortices in these channels can be expressed as Js = 2AC66 /W B where W is the width of the channel, 1636 L F Cohen and H J Jensen A is a constant and B is the field. Below a certain field of the order of 30 mT, a finite shear current density is identified in the weakly pinned material. The shear current for fields less than 10 mT is indicated by an arrow in figure 32(b). No shear current is supposed to exist for higher fields because it is assumed that the system is in the liquid state at higher fields. Js is interpreted as the force needed to overcome the interaction of the vortex lattice with the channel boundaries. For J greater than Js , the vortex system will comprise of pinned vortices and flowing vortex channels. In fact this describes certain kinds of plastic motion very well. In a sense, this is an observation of current-induced melting (see section 4.43). Measurements of entropy change. For a sample exhibiting a first-order melting, there should be a jump in the latent heat and in the magnetization 1M associated with a change of entropy, such that 1M = [dTm /dH ]1S (4.5) where 1S is the entropy jump per unit volume. The latent heat L per unit volume and the magnetization jump 1Mm are related to the entropy change by L = H /sφ0 Tm 1Sm (4.6) 1Mm = H /sφ0 (dTm /dH )1Sm (4.7) where s is the spacing of the CuO planes, Tm is the melting temperature and H is the applied field in Oe and δSm is the entropy change per unit volume. For more extensive discussions of this topic see Farrell et al (1995, 1996) and Rae (1996). Magnetization. Experimental detection of the jump in magnetization is difficult because the background magnetization change over the same temperature interval of the jump, is about eight times greater than the magnetization jump itself. Nevertheless, a reversible magnetization jump associated with an entropy change of 0.06kb , was first reported in BSCCO crystals in 1994a by Pastoriza et al using DC SQUID magnetometry. Majer et al (1995) repeated these experiments using local Hall sensor arrays. The data is extremely clean, the discontinuous jump in magnetization is shown in figure 33. Also shown are the estimated entropy change per vortex as a function of temperature. Close to Tc , δs appears to increase rapidly, which the authors suggest could be attributed to critical fluctuations. Farrell et al (1996) raised concern regarding the estimate of the entropy change in the Zeldov paper, once the demagnetization factors have been taken into account. This has been considered further by Rae (1996). If associated with a first-order transition, the Zeldov experiments indicate that the density of vortices increases in the liquid state, resembling the water–ice transition. Pastoriza et al (1994a) carried out AC susceptibility using a SQUID as an amplifier. The resulting low-field phase diagram showing a frequency-independent irreversibility line, is shown in figure 34. As first discussed by Pastoriza et al (1994a) and confirmed by Zeldov et al (1990) the transition line terminates at a critical point in low applied fields of the order of 40 mT at 40 K. The phase diagrams for BSCCO 2212 that emerge are shown in figure 34. This resembles the YBCO phase diagram (see for example figure 30) but shifted down to lower temperature and field scales, as one might anticipate in the more anisotropic system (Koshelev et al 1996). Farrell et al (1995) repeated the Zeldov experiment using global SQUID magnetometry. The magnitude of the δM change was found to correlate with the strength of the irreversible signal as determined by varying the field orientation. This suggests that the change of Magnetic behaviour of superconductors 1637 Figure 32. (a) The Arrhenius plot of voltage versus temperature at a current density of 106 A M−2 in a 10, 20, 30, 40 and 50 mT field (from left to right). Full symbols; before irradiation, open symbols; after irradiation. The inset shows the zero-field transition. (b) I –V characteristic for the irradiated samples at different magnetic fields in mT, at 80 K. The arrow marks the estimated shear current. After Pastoriza and Kes (1995). magnetization is due to the sudden disappearance of pinning, more likely related to a secondorder decoupling transition than a discontinuous entropy change. These results contradict local AC suceptiblility measurements by Schmidt et al (1996) who claimed that the size of 1638 L F Cohen and H J Jensen Figure 33. (a) A step in local B on crossing the melting line by decreasing the temperature at a constant applied field of 50 Oe. The full curve is a guide to the eye. (b) Entropy change per vortex pre-layer at the melting transition as a function of Tm . The inset gives an expanded view near the critical point. The full curves show linear fit to the data. After Zeldov et al (1995a). the jump in δM is independent of the angle of the magnetic field with respect to the c-axis and also independent of frequency. This issue has yet to be resolved. Zeldov et al (1995a) reported expressions (see references therein), for δs assuming the transition at Tm to be either vortex lattice melting or decoupling. For melting, δs can be calculated from the internal energy difference between a vortex solid and a vortex liquid per unit volume, δU ' cL2 c66 where cL is the Lindemann number. Using an expression for the melt temperature Tm ' 10.8cL2 c66 a03 where a0 is the intervortex spacing and = 1/γ reflects the anisotropy, the entropy change per vortex per layer δs ' (0.1d/)(Bm /φ0 )−1/2 . (4.8) Magnetic behaviour of superconductors 1639 Figure 34. The low-field phase diagram of BSCCO. From DC magnetization (full circles), from the peak in the in-phase part of the differential susceptibility at selected frequencies. After Pastoriza et al (1994a). (b) The first-order phase-transition line in BSCCO as measured by field (circles) and temperature (squares) scans. The full curve is a fit to (1 − T /Tc )α vortex lattice melting behaviour. The broken curve is a fit to (Tc − T )/T decoupling transition. The inset shows the phase transition line Bm in the vicinity of the Tc . After Zeldov (1995a). Hanaguri et al (1996) and Khaykovich et al (1996) explored the melting line as a function of oxygen doping which reduces the anisotropy. Both found that the melting line becomes steeper as the anisotropy γ is reduced, as illustrated in figure 35. Hanaguri et al estimated δs from the size of the δM step. They found that by increasing the oxygen content, the temperature δs increases but with increasing magnetic field, δs decreases. All of these are contradictory to the simple vortex lattice melting picture and these results need further explanation. Welp et al (1996) reported clean results for the δM jump in untwinned YBCO crystals using a global SQUID magnetometer. The change of M in fixed field (temperature) sweeping the temperature (field) are consistent with the local slope of the melting line, in agreement 1640 L F Cohen and H J Jensen Figure 35. (a) Magnetic phase diagrams of over-doped and optimally doped BSCCO 2212 single crystals. Open diamonds indicate the position of the second peak. The inset shows the temperature dependence of δM. (b) The magnetic field dependence of δs. The curves are guides to the eye. The inset shows the temperature dependence of δs. After Hanaguri et al (1996). with equation (4.7). Figure 36 shows δM and δs versus temperature. Note that the rise in δs close to Tc , reported in the BSCCO crystals is apparently absent in YBCO crystals. As yet no existing theory can fully describe the temperature dependences or amplitudes of Bm or sm in the YBCO and BSCCO system. Magnetic behaviour of superconductors 1641 Figure 36. (a) The temperature dependence of the magnetization in 4.2 T. The dotted curve represents a linear extrapolation of the low-temperature variation. The inset shows the magnetization jump in 4.2 T and 2.9 T fields. (b) Top panel: the temperature dependence of magnetization and entropy jump. Bottom panel: the phase diagram of the melting transition from resistivity and magnetization measurements. After Welp et al (1996). Specific heat. Specific heat directly measures a change in latent heat. Energy must be supplied to the crystal as a whole, in order to drive the vortex assembly through the transition. In YBCO this is more than two orders of magnitude greater than the expected latent heat. Schilling (1996) looked at untwinned YBCO crystals using a differential thermal analysis 1642 L F Cohen and H J Jensen Figure 37. (a) The temperature difference δ(Ts − Tr ) as a function of the sample temperature Ts between the untwinned YBCO single crystal and a copper reference, measured in a magnetic field of 5 T at various heating rates. The data is shifted vertically for clarity after corrections for the smooth background differences in Cr − Cs which are qualitatively similar to the 80–85 K segments displayed in the right inset. The right inset shows corresponding data taken in the zero magnetic field around Tc . The left inset shows the experimental configuration with the heat links ks and kr respectively. (b) The entropy change δS per vortex per layer. Full circles are independent estimates from δM shifts at Hm on the same sample. The inset shows the first-order phase boundary and the Hc2 (T ) crossover region which separates the normal and vortex fluid states. After Schilling et al (1996). technique (Schilling 1995) with a resolution in C better than 1 mJ/mole/K2 and in latent heat L ≈ 40 µJ kB−1 . Their results are shown in figure 37. They reported an entropy change δs ' 0.45 kB /vortex/layer, in agreement with the observed changes in 1M reported by Welp et al (1996) on similar crystals. Based on the criteria set up at the beginning of this section, these results appear to demonstrate rather clearly that a first-order phase transition takes place in the vortex state of untwinned YBCO crystals. However, Moore (1997), has drawn attention to the very different temperature-dependent form for δs in the BSSCO and YBCO systems which is difficult to understand. Moore proposed that the changes in magnetization, entropy and resistance discussed in this section, may not be about a first-order transition but may reflect an underlying crossover from threedimensional to two-dimensional behaviour when the phase correlation length l along the field direction in the vortex liquid becomes comparable with the sample dimension. Central to the Moore picture is the idea that at all non-zero temperatures the system is in a liquid state and therefore correlation lengths continue to grow at the crystal is cooled. Moore Magnetic behaviour of superconductors 1643 Figure 37. (Continued) calculated that l grows very rapidly as the temperature is lowered and the crossover in dimensionality is quite sharp and comparable with the width of the drop in magnetization. 4.4.2. Decoupling transition. Various questions still need to be addressed concerning whether the vortex lattice melts via a two-stage process. The influence of disorder (thermal, current induced, static, etc) on this process and the role of electromagnetic rather than Josephson coupling across the planes (Blatter et al 1996a, b, Nordborg et al 1996, Aegerter et al 1996, Lee et al 1997). Decoupling as a result of thermal disorder. Theoretically there are many predictions which suggest that the vortex lattice melts via a two-stage process. First into a line liquid and then at higher temperatures decoupling into a system where long-range correlation has been lost along the c-axis. Well defined thermal decoupling of the planes associated with thermal fluctuations of pancake vortices has been discussed theoretically Jensen and Minnhagen (1991), Daemen et al (1993), Glazman and Koshelev (1991), Ikeda (1995), Li and Teitel (1994) and Blatter et al (1994). It has been predicted that a change in c-axis transport or C44 should occur at this thermally induced crossover. A crossover field is suggested B0 which takes a similar form in each theory. B0 ≈ 4φ0 /γ 2 d 2 (4.9) where d is the interplanar spacing and the temperature dependence of such a decoupling line is BD = B0 (Tc − T )/T . (4.10) Evidence for decoupling in the liquid state comes from ρc and flux transformer transport data as discussed in section 4.4.1. There is strong evidence from combined ρc (T ) and AC susceptibility (Pastoriza et al 1994a) or miniature two coil c-axis transmissivity (Doyle et al 1995b) and mutual inductance techniques, (Ando et al 1995), which suggests that in 1644 L F Cohen and H J Jensen pure unirradiated BSCCO crystals melting and decoupling occur simultaneously. At high fields and low temperatures, Pastoriza’s AC susceptibility data show two dissipation peaks (which are frequency dependent) in the transition to the reversible state. This was originally interpreted as a two-step transition to an incoherent liquid phase, attributed to first a loss of c-axis long-range correlation and then a depinning of pancakes at a field associated with the DC irreversibility line, i.e. inter-plane and intra-plane dissipation occurring consecutively (see Arribere et al 1993 and references therein). More recently these peaks have been associated with the resistivity being different in the c-axis and the ab plane and the matching of the skin depths (associated with the resistivity and measuring frequency) to some sample dimension (Supple et al 1995, Steel and Greybeal 1992). Pastoriza et al (1994a) claimed that the peak in the AC susceptibility measurement which occurs when the skin depth δ matches a sample dimension D for a given frequency ω allowed the author to estimate the c-axis resistivity at this transition, by using the following relationship D = δ = ρ(T )c2 /2φ0 ω. (4.11) Pastoriza et al (1994a) proposed that the frequency-independent Hirr line below 36 mT is a true first-order phase transition ending in a critical point at 36 mT (as shown in figure 34). From the high electrical resistance along the c-axis they also claimed that this transition coincides with a three-dimensional to two-dimensional crossover. Doyle et al (1995b), explored this further by making miniature two-coil c-axis transmissivity measurements. Figure 38 shows the imaginary part of the transmitted voltage. The real part measures ρc and shows a sharp drop as a function of temperature or magnetic field. (Such sharp drops in ρab are interpreted as melting, refer to figure 28(a).) Two loss peaks are also observed in the imaginary transmitted voltage as a function of temperature and are attributed to ab (the higher temperature peak) and c-axis currents, as before. As the magnetic field is varied evidence of a sharp transition appears and moves through the c-axis current peak. This is strongly suggestive of a transition which is associated with a sudden change in the local c-axis resistivity, i.e. a decoupling transition. In contrast to the above experiments, Wan et al (1994) flux transformer data taken within a few Kelvin of Tc was suggestive of a two-step transition surviving up to Tc . It was interpreted as in-plane dissociation by a Kosterlitz–Thouless-type process first and then a Josephson decoupling transition. However, validity of these conclusions is doubtful because data which was taken at very different effective electric field was compared directly. Cho et al (1994), showed that c-axis transport measurements may also indicate a two-step transition where the decoupling line lies above the melting line, in agreement with the Wan et al result. Kadowaki et al (1994) investigated c-axis transport measurements with the field parallel and perpendicular. From the very nonlinear and hysteretic I –C curves in applied magnetic field, a field scale (B ∗ T ) is identified and represents a measure of the Josephson coupling strength between adjacent pancake vortices. The authors claimed that at high temperatures a short-range Josephson coupled–vortex liquid exists above the melting line. However, a note of caution is required here because large transport currents were used in many of the experiments which claim to observe a two-step transition and these currents may have induced heating and other complications. Gray et al (1996, 1997) addressed the influence of correlated disorder on the thermal decoupling transition. In columnar-defected thallium 2212 thin films the films appear to behave quasi-two-dimensional-like when the field is not closely aligned to the tracks. When the field is aligned along or close to the tracks, directional suppression of the ab plane resistivity is observed, implying vortex-line-like behaviour. The authors suggested that there is always weak coupling between the plane even at temperatures above the thermal Magnetic behaviour of superconductors 1645 Figure 38. The imaginary transmitted voltage at 10 kHz with DC fields of (a) 70 mT, (b) 60 mT, (c) 40 mT and (d ) 10 mT applied parallel to the c-axis. The peak at higher temperatures is associated with in-plane resistivity and that at lower temperatures with out-of-plane currents. After Doyle et al (1995b). decoupling temperature and they also implied that vortex–vortex interactions have not been properly considered in the case of columnar-defected materials. Decoupling in the presence of current-induced disorder. Many experiments which might be considered to be in a high-current limit, are suggestive of a two-step transition process. See for example, Wan et al (1993) and Cho et al (1994). Keener et al (1996) showed evidence that the two-step process may be current induced. The statement concerning experiments conduced at high current, discussed in the last section is also valid here. 1646 L F Cohen and H J Jensen Decoupling in the presence of static disorder. There have been many experiments discussed throughout the text which could be described in this section. Here we focus on the debate concerning the fate of the melting/decoupling line as the crystal is cooled such that underlying static disorder pin vortices and destroy the perfect Abrikosov lattice. Note that magnetically induced decoupling is not a dimensional crossover as such but will occur in anisotropic systems if the pinning in each layer is sufficiently strong. In this case, the correlation between the pancake vortices in the c-direction is destroyed above a crossover field B2D , because the magnetic repulsion between pancakes in the same layer becomes stronger than the attraction in adjacent layers. From various experiments described in sections 4.6.3 and 4.7 it is established that in unirradiated BSCCO, bulk pinning becomes effective at around 40 K and at a crossover field H ∗ (see for example Zeldov et al 1995b, Cohen et al 1997). Khaykovich et al (1996) found that below about 40 K the transition to bulk pinning occurs very sharply at the local field B ∗ which is approximately temperature independent and terminates at the critical point (discussed in section 4.5.1). Neutron diffraction data from Cubitt et al (1993) suggests that an ordered vortex lattice (recently denoted Bragg glass), exists in the entire low-field phase below H ∗ . This has led Khaykovich et al to suggest that although Bm is a simultaneous first-order melting and decoupling transition, in the presence of pinning, this becomes a sharp second-order decoupling transition at B ∗ . This area has to be explored further. Josephson weak-link behaviour. Passing a transport current along the c-axis would appear to be the most straightforward method to examine the dimensionality issue. From caxis transport measurements Kleiner et al (1994a, b) have shown that Josephson weaklink characteristics and Shapiro steps can be obtained in every system at 4.2 K except well oxygenated YBCO 123 crystals. Measurements have not been made at low magnetic fields because of severe heating effects. It is quite probable that Josephson behaviour can only occur once the vortices are pinned, because depinned vortices are susceptible to flux flow which must introduce large phase fluctuations between planes. It is difficult to draw firm conclusions when Josephson behaviour is not observed because of the number of possible explanations. An absence of characteristic Josephson behaviour as a function of temperature could imply a complete loss of c-axis correlation, an onset of very strong caxis correlation, depinning or melting into a highly entangled flux liquid (where flux cutting and reforming must also introduce large plane–plane phase fluctuations). The temperature at which Josephson behaviour disappears depends on the family of crystals and the angle of the external magnetic field relative to the crystal planes. However, the absence of Josephson-like features in well oxygenated YBCO crystals implies that quasi-two-dimensional behaviour, or decoupling, does not occur in the irreversible regime. No exploration of the effect of twinning on the observation of weak link behaviour in YBCO crystals, has been reported. The c-axis conductivity in BSCCO 2212 crystals shows that there is a well defined Josephson current at the field of the order of 2T for H //c for a temperature range between 20 and 10 K but that this current shows an anomalous re-entrant behaviour at lower temperatures and a gradual smoothing out of the Josephson characteristic (Rodriguez et al 1993, Doyle et al 1995a). The re-entrant behaviour has a dependence on magnetic prehistory and might be attributed to an interference between shielding and transport currents (Gordeev et al 1994). 4.4.3. Critical scaling. The general description of critical lengths and timescales at a continuous phase transition were set out in section 3.5, as were the current–voltage characteristic (IVC) predictions. Evidence for glassy behaviour can perhaps be better Magnetic behaviour of superconductors 1647 Table 1. Critical exponents. v(z + 2 − d) from ρL (T ) ξvg = (ckT /ϕ0 J )1/2 1.7 6.6 1.6 6.4 1.8 6.4 1.8 ± 0.2 6.2 6.5 6.5 6±2 0.3 µm 2±1 6.5 15 µm z from I –V v from curve J0+ (T ) v(z + 2 − d) from z, v Sample Comments Field T YBCO Film (1989) Koch T ↓ Tg 2//c 3//c 4//c 1//c 4.9 5.0 4.7 1–600 MHz T ↓ Tg 1–3//c 6//c 3.7 ± 0.46 3.4 ± 1.5 T ↓ Tg > 10//c ρ = (T − T ∗ )S s =6±1 T ↓ Tg 6//ab s = 1.35 ± 0.15 e-irradiated T ↓ Tg 1–8//c s = 1.2–3.0 depends on field and is not glassy-like Film Dekker (1992) T ↑ T g Film Hui Wu (1993) Twinned crystal Gammel (1991) Untwinned crystal Safar et al (1993) Untwinned YBCO crystal Kwok et al (1994) Untwinned YBCO crystal Fendrich (1995) BSCCO BSCCO crystal Doyle et al (1996) BSCCO crystal Zech et al (1995) BSCCO films Miu et al (1995) BSCCO crystal Safar et al (1992b) 4.3 ± 1.5 Below 20 K T ↓ Tg columnar defected T ↓ Tg columnar defected T ↓ Tg T ↓ Tg c-axis transport µ depends on field smoothly s = v(z − 1) s = 8.5 ab-plane transport 8.5 ab-plane transport 9 2–6//c 6.7 examined well below this transition line. We discuss the evidence for glassy behaviour deep inside the solid, in section 4.6. It is appropriate here to mention that there are problems associated with the scaling experiments which have been long understood from computer simulations of continuous transitions. If the experimental window is too small conclusions might well be misleading. If critical exponents can be extracted reliably then this suggests that there is a thermodynamic transition and it is second order. The critical exponents must be independent of magnetic field. Critical exponents which appear to vary smoothly with magnetic field are more suggestive of plastic behaviour. From exponents determined from below the transition, such as Dekker et al (1992), or in electron-irradiated crystals, there is definitely evidence of plastic behaviour. In table 1 we examine the consistency of extracted exponents between different samples. There is some agreement between YBCO films, twinned and untwinned YBCO crystals (at low temperatures and high fields in the latter case). The exponents change systematically for different kinds of static disorder, such as columnar defects or intrinsic planes. Note that in the case of intrinsic planes theoretical studies of inter-layer vortex melting in two-dimensional layered systems predict that an intrinsically pinned vortex lattice cannot melt via a second-order transition (Mikheev and Komomeisky 1991, Korshunov and Larkin 1992). Usually IVCs can only be scaled in a range of temperatures about 2 K wide around the transition for YBCO and about 5 K wide for BSCCO. Occasionally scaling of 1648 L F Cohen and H J Jensen transport IVCs are reported to occur over a wider temperature range which counts against the existence of a critical region (Koch et al 1989). Scaling also occurs over a narrow range of currents only. For small currents, it appears that the equilibrium properties are lost and the vortices become pinned. Large currents have been shown to induce non-equilibrium effects when measuring the resistivity (Liang et al 1996). Four representative experiments are examined in more detail below. YBCO films–H //c. Scaling was first shown to occur in YBCO thin films (Koch et al 1989) although the strength of the scaling argument was criticized by Coppersmith et al (1990) and Griessen (1990). In order to be in the critical regime, restrictions on length-scales were first that ξ > l, where l is the average distance between vortex lines, which is only satisfied in high fields (but below Hc2 ), and secondly that ξ 6 t, where t is the film thickness, otherwise the three-dimensional assumption would also break down. Koch claimed that the I –V curve shape for T 6 T g was inconsistent with the standard flux creep model, which predicts that V ∼ sinh(I /I0 ) resulting in an IVC with a positive curvature on a log I –log V plot. Figure 39 shows that the curvature is negative at low temperatures, with a value of µ ∼ 0.4 ± 0.2, where µ is defined in equation (3.20). Untwinned YBCO crystals H //ab. Kwok et al (1993, 1994) showed that the sharp resistive transition associated with first-order melting is suppressed when the magnetic field is aligned within 0.5◦ of the ab planes. For the H //ab geometry critical exponents are extracted by plotting [d(ln ρ)/dT ]−1 versus T , where the slope of the straight line is 1/ν(z − 1) and the intercept defines the transition temperature Tg as shown in figure 40. The exponents are consistent with a smectic transformation as found in liquid crystals. Unirradiated BSCCO crystals H //c. The IVCs of clean unirradiated BSCCO crystals are expected to be linear above the decoupling field H ∗ (see sections 4.5.2 and 4.6.3). By examining the temperature dependence of the linear resistivity Tg can be predicted from equation (3.17). Safar et al (1992b) observed Arrhenius behaviour (i.e. ln ρ ∝ 1/T ) at high temperatures and a critical scaling regime at low temperatures. Figure 41 shows the mechanism by which the Tg line was extracted for pure BSCCO crystals. This is the same as just described for the unirradiated YBCO crystals. Safar et al defined a temperature T ∗ below which the behaviour entered the critical regime. The critical exponent agrees with vortex glass prediction. Columnar-defect BSCCO crystals H //c. For correlated disorder such as columnar defects, the low-temperature vortex solid is proposed to be a Bose glass (Nelson and Vinokur 1992). Critical scaling analysis has been applied to the in-plane resistivity at high temperatures by Miu et al (1995) and the c-axis resistivity by Seow et al (1996). The results from the two experiments are consistent and produce critical exponents which agree with Bose glass scaling. Seow et al found that for B parallel to the tracks, the linear resistivity along the c-axis, ρc , does not show Arrenhius-like behaviour, whereas when B is misaligned with the tracks, the resistivity becomes Arrenhius-like below about 1% of the normal state resistivity. The behaviour of ρc , when the field is parallel to the tracks returns to Arrenhiuslike behaviour for B > Bφ . As shown in figure 42, when the field is aligned parallel to the tracks and B < Bφ , ρc can be replotted in the critical scaling form, producing a value for Tg = 66 K which is greatly shifted up in temperature compared with the virgin crystals. The effective exponent s = ν 0 (z 0 − 2) = 8.5, agrees with Bose glass predictions. Magnetic behaviour of superconductors 1649 Figure 39. The I –V curves at constant T for (a) H = 0.5 T and (b) H = 4 T. The curves differ by intervals of 0.1 and 0.3 K, respectively. After Koch et al (1989). 4.5. Below the irreversibility line—the vortex solid The conclusion drawn from the transport measurements (reviewed in section 4.4.3) are that in the presence of disorder IVCs become nonlinear in a way that suggests transitions into glassy-like solids. This is true for both YBCO and BSCCO. Exotic solids such as smectic glasses, Bose glasses and Bragg glasses reflecting the dominant source of static disorder were identified. The focus of this section is to examine evidence for the continuity of behaviour below the irreversibility line. Vortex behaviour deep inside the vortex solid can only be explored with magnetic measurements, where low electric fields can be accessed 1650 L F Cohen and H J Jensen Figure 40. (a) The plot of 1/[(1/ρab )(dρab /dT )] versus T , for a YBCO crystal in a 6 T field. The kink associated with the first-order melting is present for small misorientation angles of 0.5◦ , but disappearing when H is aligned to the planes. (b) The plot of H //ab versus T ∗ . The inset shows the field dependence of the dynamic scaling exponent s obtained from (a). After Kwok (1993). easily. For reviews on magnetic relaxation see Yeshurun et al (1996), for thermally activated motion see Schnack (1995), and for quantum creep see van Dalen (1995). As discussed in section 3, the aim of monitoring dynamic behaviour is to determine the functional form of the effective pinning barrier on current Ueff (J ). 4.5.1. Peak effects in J (B). Peaks in J (T ) were discussed in section 4.4.2 and they are not associated with the phenomena discussed in this section. Kobayashi et al (1995a, b) showed that crystals which show large J (B) peaks with a straightforward monotonic temperature dependence, do not show the J (T ). Crystals with much weaker pinning with a nonmonotonic temperature dependence, show both a J (T ) peak and a J (B) peak, but they lie in different positions in the H –T plane. The two peaks are not related but may coexist. Magnetic behaviour of superconductors 1651 Figure 41. (a) The plot of the inverse logarithmic derivative of the resistance. The full curve represents a fit to the vortex glass theory with Tg = 20.2 K. Deviations from the vortex glass theory are observed above 28 K. The inset shows the critical exponents for different fields. (b) The H –T -plane showing the positions of Tg and T ∗ lines. After Safar et al (1992b). Peak effects were first discussed by Le Blanc and Little (1960) who observed an anomalous peak in J (T , B) in LTS. Pippard (1969) and Larkin and Ochinnikov proposed that this peak effect was related to softening the vortex lattice. (A soft lattice can pin more strongly that a more rigid one and hence can produce current enhancement.) In fact there are many mechanisms which can generate a peak in J , see Cambell and Evetts (1972) for a summary. Most YBCO crystals show an anomalous second peak in the magnetization loop, known as the fishtail peak, illustrated in figure 43. The peak position has a strong temperature dependence and in deoxygenated YBCO 123 and in YBCO 124 crystals it exists down 1652 L F Cohen and H J Jensen Figure 42. The plot of ρc (dρc /dT )−1 and normalized ρc against temperature in a 0.7 T applied field. The linear regime is clearly seen below 75 K. The inset shows the exponent n of the Bose glass phase extracted from the resistivity data. After Seow et al (1996). Figure 43. The M–H -loop in an untwinned YBCO crystal at 77 K, showing the second peak feature denoted the fishtail peak. The field at which the peak maximum occurs has a strong temperature dependence (1 − T /Tc )3/2 . to the lowest measured temperatures. The origin of this feature has been much discussed and there is still a lack of consensus in the literature. Explanations include the effects of macroscopic granularity and underlying defect structure thought to be associated with oxygen vacancies (Daümling et al 1990, Yeshurun et al 1994b, Osofsky et al 1992, Erb et al 1996); simple dynamic effects associated with creep (Cohen et al 1993, Delin et al 1992, van Dalen 1995); Krusin-Elbaum et al (1992) discussed the J (B) peak in terms of a a crossover from single vortex pinning to a pinning of vortex bundles; Perkins et al (1995) suggested that it is related to the interplay between the field dependence of the characteristic Magnetic behaviour of superconductors 1653 Figure 44. The M–H -loop in a BSCCO crystal at 30 K, showing the second peak feature denoted the arrowhead peak. The peak only occurs over a limited range of temperatures and is approximately independent of the field, as indicated by the line labelled Bsp in figure 51. energy and current scales; Zhukov et al (1995) speculated that it is associated with plasticity or with softening C66 ; and Abulafia et al (1996) implied that it is related to a crossover from elastic to plastic vortex behaviour. Very pure untwinned YBCO crystals do not show this feature and also highly disordered thin films do not show it. A rough measure of the local static disorder in YBCO crystals is the value of the screening current density J , at 77 K and 1 T. Typical values for untwinned, twinned and proton irradiated YBCO are J = 102 , 103 and 104 A cm−2 , respectively. Werner et al (1994) reviewed the effect in many different samples and concluded that it is caused by an interaction between the flux-line lattice and the defect structure and may not be related to a specific defect structure itself. Erb et al (1996) showed how the peak could be reversibly induced by introducing oxygen vacancies in untwinned YBCO crystals. The effect of point defects and twin planes on the shape of the fishtail feature was elucidated by Küpfer et al (1996). In BSCCO 2212 crystals, an anomalous second peak is observed in J (B), and because of its shape it is known as the arrowhead feature, as shown in figure 44. Unlike the fishtail peak found in YBCO crystals, the arrowhead feature occurs between approximately 20 K and 50 K only and the field at which it occurs is almost temperature independent. It can be altered in size and position by increasing the number of point defects through electron irradiation (Chikumoto et al 1992), by high pressure (Yang et al 1994) or low-temperature oxygen annealing (Kishio et al 1994), by partial doping of lead onto the barium sites (Cai et al 1994), and by introducing structural defects (Yang et al 1993b). As discussed in sections 4.5.2 and 4.6.3, the field at which the arrowhead peak occurs, is associated with magnetic decoupling of the vortex lattice. Zeldov et al (1994) showed that there is a strong interplay between surface and bulk pinning effects at the peak field. It was first suggested by Chikumoto et al (1992), and later by Yeshurun et al (1994a), Cai et al (1994) and Cohen et al (1997), that the arrowhead feature results from an interplay between static and dynamic effects. Kishio suggested that crystals which do not show the arrowhead feature (and there are many which do not), may be so anisotropic that the peak field is unmeasurably low. 1654 L F Cohen and H J Jensen Alternatively they may have an inhomogeneous distribution of properties resulting in a very gradual change rather than a sharp crossover in behaviour as a function of field. 4.5.2. Unirradiated YBCO 123. From transport measurements we learn that above the Hirr line, critical scaling suggests that second-order transitions take place in various elastic vortex solids. The vortex glass exponent µ (defined in equation (3.20)) has been measured in YBCO thin films by Dekker et al (1992) and Berghuis et al (1996), from below the Hirr line. In both papers it was reported that the µ value, rather than change abruptly it slowly varied between 0.19 and 0.94 as a function of temperature, magnetic field or current. The Dekker results are shown in figure 45. Both the Dekker and the Berghuis observations are important because they imply that the pinned vortex system does not necessarily appear glassy, close to the Hirr line when determined from below it. That the so-called glass exponent varies continuously with field implies some kind of plastic behaviour. Turning to magnetic measurements, the first general discussion of evidence of glassy behaviour in YBCO, came from Malozemoff and Fisher (1990). They drew attention to a temperature-independent plateau in the normalized creep rate S(T ). As shown in figure 46, the plateau appeared to have a universal value of 0.03 at fixed fields of the order of 1 T. Expressing the normalized creep rate S = 1/[µ ln(t/t0 )] and substituting µ = 1, at an attempt time of the order of 10−10 s the authors obtained a value of S = 0.033. The value for µ and the attempt time are consistent with vortex glass and collective pinning theories. Malozemoff (1991) also attributed the linear behaviour of S(T ) at temperatures below the plateau to Anderson–Kim-type thermal activation and above it, to a softening of the glass, possibly suggestive of plastic behaviour in agreement with the transport measurements described above. Note that Caplin et al (1995), gave a useful explanation of why S = 0.03 could be so frequently observed simply as a result of the similarity in experimental conditions in which the measurements are made (such as sample size electric and magnetic field ranges etc). In the collective pinning model, many regimes of behaviour are possible in the vortex solid (refer to Blatter et al (1994b)). Krusin-Elbaum et al (1992) presented evidence for many of these regions from flux creep data in twinned YBCO crystals. The same regimes of behaviour set out by Malozemoff have been further explored as a function of magnetic field and temperature in twinned and untwinned 123 and 124 crystals (see Cohen et al 1994b, Perkins et al 1995, Zhukov et al 1995). The Malozemoff ‘plateau’ in S(T ) observed at fixed field actually occurs over a wide range of temperatures and fields as shown in figure 47, as region 1. Region 1 indicates where a glassy-like solid exists in the H –T plane of well oxygenated YBCO. This region ‘shrinks’ as the crystals are deoxygenated and made more anisotropic as discussed by Cohen et al (1994c). Perkins et al (1996) discussed that by using equation (3.12), the dynamic normalized creep rate Q = (d ln J /d ln E)B,T can be expressed as d ln Ueff (4.12) S = 1/C d ln J B,T where C = ln(Bωd/E). A power law Ueff (J ) ∼ J −µ with µ independent of B and T automatically results in constant S(B, T ). This implies a convex ln E–ln J curve, as observed in transport measurement. The form of the ln E–ln J curve in this regime has been confirmed over a large electric field window by a mixture of relaxation and transport measurements, by Gordeev et al (1994), as illustrated in figure 48. Interestingly the authors confirmed that the fishtail feature survives in transport measurements at high electric fields. The region below the Malozemoff plateau in S(T ), is often analysed unreliably because it only occurs at low applied field and self-field effects dominate. The region above the Magnetic behaviour of superconductors 1655 Figure 45. (a) The temperature and (b) the field dependence of the critical exponent µ, on a variety of thin-film YBCO samples. The temperature scale is normalized to the glass temperature in each case. After Dekker et al (1992). plateau in S(T ) was identified by Malozemoff as a softening of the glass. This region is marked on the H –T plane in figure 47 as region 2. It has also been shown by Cohen et al (1994b) that S(B) is linear in this region. Using the magnetic scaling analysis Perkins et al (1995) showed that the linear S(B) is related to a logarithmic law Ueff (J ) dependence. Logarithmic Ueff (J ) implies a power law IVC of the form E = J n , where in this case n is inversely dependent on B and T . The behaviour of this regime has been compared with various theoretical predictions from collective pinning as shown in figure 49. It is found that the field dependence of U0 and J0 are not in agreement with that theory in its present form. Abulafia et al (1996) interpreted the behaviour, in terms of plastic creep resulting from dislocation flow. This is consistent with the Dekker et al (1992) and Berghuis et al (1996) transport results discussed at the beginning of this section. To summarize, except for regime 1, which survives up to high temperatures at low 1656 L F Cohen and H J Jensen Figure 46. Normalized relaxation S versus T for a variety of YBCO samples at 0 T, 1 T and 2 T fields, illustrating the universality of S = 0.03. After Malozemoff et al (1990). fields (as indicated in figure 47), most dynamic magnetization techniques probably measure plastic behaviour associated with the the static disorder and inhomogeneity of a particular crystal. In general it is the field dependence of the creep rate S(B) or Q(B) which fails to fit into the framework of collective pinning. This is a reflection of the fact that most magnetization measurements are set up such that the interplay between vortex–vortex and vortex pin energy is conductive to plastic flow over most of the H –T plane in YBCO. This is discussed in detail by Zhukov et al (1995) and Abulafia et al (1996). 4.5.3. Unirradiated BSCCO 2212. Using transport techniques to carry out critical scaling analysis, Safar et al (1992b) suggested that glassy behaviour occurred below 20 K. However, from the magnetization measurements there is no evidence for the Malezemoff plateau in S(T ) or Q(T ) in BSCCO at these temperatures. It is generally agreed from the symmetry of the M–H loop shape and the size of the irreversible signal that below 20 K, bulk pinning dominates over surface or geometric barrier effects. Magnetic scaling analysis can be performed below 20 K in BSCCO 2212 and the results implying power law E–J curves as found in region 2 in YBCO. Totty et al (1996) found that m and n, the field dependence of the characteristic current and energy scales, are also similar to YBCO but with a stronger temperature dependence. Given the similarity to YBCO, the observed behaviour in BSCCO Magnetic behaviour of superconductors 1657 Figure 47. An experimentally derived H –T diagram showing regimes of flux creep behaviour. Hp is the critical state penetration field, below which the sample is not fully penetrated, Hd is the field at which S begins to rise linearly off the 0.03 plateau, Hs is the field at which much faster creep occurs and Hirr is the irreversibility line. After Cohen et al (1994b). Figure 48. Current and voltage characterisitics of a YBCO crystal at 87 K at several magnetic fields. The higher electric field data is compiled from direct electric transport and the lower electric field data is from magnetization measurements. After Gordeev et al (1994). 2212 over most of the H –T plane is also probably some kind of plastic response. van Dalen et al (1996) explored the dynamic creep rate Q variation as a function of an angle in unirradiated BSCCO crystals at 20 K and found that the measured current density, Q and Uc , the characteristic pinning energy, scale with the c-axis component of the external field. Nideröst et al (1996), observed three regimes of flux creep behaviour measured by long-time relaxation over seven decades of time, as a function of temperature. Using the 1658 L F Cohen and H J Jensen Figure 49. (a) The predicted values of the exponents m and n for the power-law field dependence of J0 ∝ B m and U0 ∝ B n corresponding to each of the regimes in figure 3, where sb, lb and CDW denote small bundle, large bundle and charge density wave. See Blatter et al (1994) for definitions of these regimes. The data for a twinned Tm 123 crystal is indicated by the arrow. After Perkins et al (1996). Maley method they found a logarithmic Ueff (J ) function below 20 K and a power law Ueff (J ) function above 40 K. Between 20 K and 40 K no unique functional dependence could be found. The low temperature behaviour is attributed to individual two-dimensional pancake vortex pinning. Several papers from van der Beek et al (1992), and Vinokur et Magnetic behaviour of superconductors 1659 Figure 50. The B–T phase diagram where the full curve is the theoretical three-dimensional melting line (Houghton et al 1990), the circles are the melting transition from the neutron diffraction intensity and the squares show the boundary between the reversible and irreversible magnetic behaviour in hysteresis loops. After Cubitt et al (1993). al (1995), have also addressed the dynamics in BSCCO and concluded that dislocation mediated creep rather than two-dimensional collective pinning provides a good description of the magnetic relaxation. So in this respect there is some consensus about the behaviour of BSCCO below the irreversibility line. Other information has been obtained about the form of the H –T diagram in BSCCO 2212. The vortex lines can be regarded as two-dimensional pancake vortices confined to the Cu2 O layers by Josephson and/or magnetic coupling. At low fields, such coupling results in essentially three-dimensional flux lines. Josephson coupling ensures phase locking or phase coherence between pancakes on adjacent layers. At high fields the in-plane repulsion between pancakes exceeds their inter-plane attraction. Uncorrelated pinning in different Cu2 O layers breaks up the flux lines in the field direction, leading to so-called flux-line decomposition, or decoupling. Such decomposition has been inferred from small angle neutron diffraction experiments by Cubitt et al (1993) and µSR experiments by Lee et al (1993, 1997). The signature for a correlated three-dimensional lattice disappears above 60 mT. Vinokur et al (1990) predicted this decomposition to occur at B2D = φ0 /(sγ )2 (4.13) where s is the spacing and γ is the anisotropy factor. The field associated with the decomposition line coincides with the arrowhead peak. 1660 L F Cohen and H J Jensen Figure 51. The phase diagram of BSCCO showing the penetration field Hp , the onset of irreversible shielding HIS , the bulk irreversibility line BIR and the low-field phase transition at the onset of the second peak (arrowhead peak) Bsp . The HIS line is associated with surface and geometric barriers and the fits to theoretical forms for these lines are shown. Refer to Zeldov et al (1995b). Rodriguez et al (1993) reported a gradual smoothing out of the Josephson characteristic below 10 K. From these experiments as well as AC susceptibility and DC magnetometry, de la Cruz et al (1994a), suggested that the three-dimensional solid exists up to high magnetic fields at the lowest temperatures, modifying the original Cubitt et al picture. In fact more recent neutron and muon data also support this claim Aegerter (1996), Bernhard et al (1995). Zeldov et al (1995b) produced a detailed phase diagram based on local magnetization measurements using Hall bar arrays. The high-temperature melting line was discussed in section 4.4.1. At temperatures below the suggested critical point, bulk pinning starts to be important and the non-equilibrium phase diagram is extremely sensitive to the dimensionality of the pinning and the superconducting anisotropy. 4.5.4. Irradiated YBCO 123 and BSCCO 2212 crystals. Irradiation enhances the screening current density, alters the position of the fishtail or arrowhead peak and has been seen to enhance or suppress the position of Hirr line in the H –T plane. YBCO crystals show unique lock in signatures to twin planes (Oussena et al 1996, Zhukov et al 1996), to CuO2 (intrinsic) planes and to columnar defects. As a function of field orientation the Hirr line in both YBCO crystals (Krusin-Elbaum et al 1994a) and BSCCO crystals (Zech et al 1995) display characteristics which resemble the predicted Bose glass cusp at high temperatures. Klein et al (1993a) discussed ‘flux flop’ effects associated with locking onto columns at low fields and small angles away from the c-axis, have been also been reported. Hardy et al (1996), studied the accommodation of vortices to tilted line defects with various electronic anisotropies from crystals of 2212, 2223, 1223 and 123 composition and also present a brief review of the subject. For both YBCO and BSCCO at low temperatures isotropic pinning enhancement is observed. Directional effect are observed at higher temperatures. The isotropic regime is ascribed to vortices zig-zagging between the ab planes and the Magnetic behaviour of superconductors 1661 Figure 52. The persistent current density J and normalized relaxation S as a function of temperature, for two different irradiated YBCO crystals where α is the angle of the columns with respect to the c-axis and Bφ is the matching field. After Civale et al (1996). columns, keeping their mean-field direction along the applied field. The model invoked by Hardy can explain the data but assumes that the vortices are line-like and have line tension at all temperatures for all crystals studied. This is then in conflict with the concept that BSCCO is two-dimensional-like at low temperatures. There are few published systematic studies of the dynamics of vortices in columnar-defected YBCO or BSCCO crystals. Irradiated YBCO 123. There are many papers on the effect of irradiation on flux dynamics. Initially the influence of point-defect irradiation was studied in YBCO for example by Civale et al (1990) using 3 MeV protons, and it was found that although the critical current was enhanced, the irreversibility line and creep rates (and therefore pinning potential) were almost unaffected. Thompson et al (1991a, 1993) took the proton irradiation YBCO studies further, using the Maley analysis to extract a functional form for the Ueff (J ) function which 1662 L F Cohen and H J Jensen agreed with collective pinning theory. These experiments were only carried out at one fixed field of 1 T. They concluded that the quasi-exponential temperature dependence of the current density results from flux creep and is not inconsistent with collective pinning theory. Sun et al (1992), examined both the temperature and field dependence of the activation energy in proton irradiated YBCO. Ueff (H ) was found to vary as H −α and α depended on both temperature and current. Schindler (1991) found that fast neutrons with energy greater than 0.1 MeV increased the critical current and decreased the creep rate, implying a change of the pinning potential in YBCO crystals. Konczykowski et al (1991) found that irradiating with 5.3 GeV lead ions increased the Hirr line dramatically and the critical current and also decreased the creep rate. Civale et al (1991a, b) found similar current and Hirr line enhancement from discontinuous tracks of amorphous material produced by 580 MeV Sn ions at 30◦ to the c-axis. They called these tracks columnar defects. The effectiveness of the tracks were explored as a function of angle of applied field, irradiation dosage and temperature. It was found that the tracks were most effective when the field was aligned parallel to the tracks. Above 88 K, Hirr was independent of dose and similar to the unirradiated crystal. Flux dynamics of columnar-defected twinned YBCO crystals was first studied by Konczykowski et al (1991, 1993). Long-time relaxation was found to be non-logarithmic exhibiting an increase in effective barrier for flux creep with decreasing current in agreement with vortex loop nucleation as proposed by Nelson and Vinokur (1992). The experiments were only made at very low applied fields of the order of 50 mT. Long-time relaxation measurements have since been made by Civale et al (1996) at fields less than the matching field and by Thompson et al (1997) at fields both less than and greater than the matching field. Both groups report that for fields less than the matching field the normalized creep rate S(T ), shows an anomolous rise at intermediate temperatures associated with a drop in current density J as shown in figure 52. The peak in the creep rate occurs at the same temperatures that the Malozemoff plateau in S(T ) was observed (see unirradiated YBCO section) and the value of S at the peak is of the order of six times that of the plateau (Thompson et al 1997). Civale et al considered that at low temperatures and low fields, the vortices are individually pinned by the columns, and vortex–vortex interactions are negligible. As the vortex density increases the elastic interactions increase and when the elastic energy is comparable with the pinning energy of individual tracks, collective effects take over. Bcr is the field at which this occurs and it is temperature dependent. At low temperatures Bcr ' Bφ . According to Bose glass theory, initial stages of relaxation should take place, via half-loop excitations. As relaxation progresses, the size of the vortex loops become of the order of the columnar track spacing, so that segments of the same vortex can sit on neighbouring columns. Further relaxation should be dominated by double-kink excitation. At the intermediate temperatures where the peak in S(T ) is observed the relaxation is anomalous in the sense that it varies nonmonotonically with time suggestive of two competing processes. Civale et al suggested that these processes are associated with double kink excitations of individually pinned vortices at short times and collective behaviour at longer times. Thompson et al offer a very similar interpretation, also consistent with the Nelson and Vinokur (1992, 1993) Bose glass theory. At B > Bφ , the peak in S(T ) is not observed. Beauchamp et al (1995) explored Bose glass/quantum creep behaviour at millikelvin temperatures in YBCO crystals irradiated with 605 MeV Xe ions. They found that the relaxation rate can be divided into three regimes of behaviour depending on ratio of vortex density to columnar defect density. Quantum creep occurs in the dilute limit, vanishing magnetic relaxation is observed at B = Bφ in the so-called Mott insulator phase, and for B > Bφ they observe a temperature-dependent vortex motion. Larkin and Vinokur (1995) Magnetic behaviour of superconductors 1663 extended the original Bose glass theory to consider the dilute and dense vortex limits. Gray et al (1996, 1997) showed that because of vortex–vortex interactions, the columns can effect pinning at fields many times more than the matching field at low temperatures. In the Bose glass theory for parallel columns, once a segment of vortex reaches an adjacent column, (by thermal activation or quantum-mechanical process), the remaining part of the vortex can follow at no additional cost. Hwa et al (1993) proposed that pinning would be improved even further if the columnar tracks were splayed or tilted with respect to each other. In the splayed glass phase, during the vortex hop, an ever increasing segment of line is forced into an energetically unfavourable region. Krusin-Elbaum et al (1994a, 1996) and Schuster et al (1995a, b) (who also imaged the flux penetration into crystals using magneto-optics), confirmed that there is a dramatic enhancement of J when the vortices are splayed. Devereaux et al (1995) raised the issue that the misalignment of the magnetic field and the columns may weaken the localization of the vortices and reduce the Hirr line. 4.5.5. Irradiated BSCCO 2212 crystals. Thompson et al (1992) first pointed out that the angular selectivity seen in YBCO at high temperatures is absent in BSCCO 2212 crystals irradiated with 580 MeV Sn ions at 20 K, although the current density and the irreversibility line are enhanced over the unirradiated crystals. Klein et al (1993b, 1994) later showed that in fact uniaxial enhancement is observed in irradiated BSCCO crystals, but only above 40 K. The loss of angular selectivity is either related to the fact that the system is more two-dimensional-like at low temperatures or that random point defect rather than columnar pinning dominates or a combination of both. Leghissa et al (1993) demonstrated loss of translational order in columnar-defected BSCCO using high-resolution Bitter patterns. The Hirr line was studied by Krusin-Elbaum et al (1994b) for BSCCO crystals irradiated with 1 GeV Au ions along the c-axis. A well defined crossover field Bcr ∼ 1/2Bφ was established. Below Bcr the Hirr ∝ (1 − T /Tc )α where α is dose dependent. Above Bcr the Hirr (T ) line is linear. The paper also discusses the influence of columnar defects on the melting scenario. Moshchalkov et al (1994) found close agreement with the Bose glass theory predictions and the temperature dependence of the critical current density extracted from the magnetization measurements. Unfortunately because these measurements were made at remanence, the influence of self-field effects is unclear. Konczykowski et al (1995) found giant, strongly non-logarithmic magnetic relaxation in irradiated BSCCO 2212. The authors converted their flux creep data into I –V curves and extracted values for the exponent µ in the interpolation formulae. (The interpolation formulae can be used because the functional form of dependences characterizing the Bose glass phase are identical to that for the vortex glass phase and predict power law Ueff (J ) form.) The values of µ were found to agree with the Nelson and Vinokur predictions at 60 K and at fields much less than the matching field. Although there is quite a bit of scatter in the data, the predicted µ = 13 for variable range hopping was observed. Steel et al (1996) pointed out the fact that columnar defects influence electrical properties of Tl 2212 thin films up to fields at least 40 times that of the matching field, demonstrating the importance of vortex–vortex interactions and also suggesting that the matching field has no sharp significance. This is not inconsistent with Bose glass theory. 5. Summary of the questions at the brink of resolution The complexity of the theoretical description of the vortex state has increased significantly with the contributions from statistical mechanics produced after the discovery of the HTS 1664 L F Cohen and H J Jensen (Blatter et al 1994). The balance of the competition between the three energy scales: the vortex–vortex interaction, the vortex–pinning interaction, and the thermal energy can lead to many very different types of behaviour. Many of these theoretical developments are concerned with the equilibrium phases and the nature of the transition between these phases in various model systems. As such these theoretical developments might not be of direct relevance to experiments. One of the problems encountered when dealing with the flux system in real superconductors is the irreversibility, i.e. non-equilibrium features, encountered whenever pinning is relevant. A result of the massive investigation into the properties of the flux system in HTS is that today we have a fairly precise idea about the cardinal questions still to be completely resolved. It is useful to distinguish between situations where the pinning energies are negligible compared with the vortex–vortex interaction energy and the thermal energy and the situation where the pinning energy is competing with these two energy scales. In the case where pinning can be neglected we believe that the following list of issues are among the most important yet to be settled and in fact are sufficiently well posed to allow a resolution in the near future. (1) From sharp drops in magnetization, entropy and resistance, clean untwinned YBCO and BSCCO crystals appear to show evidence of a line of first-order transition in the H –T phase diagram. However, theoretical concerns have raised the issue whether finitesize effects associated with a crossover from two- to three-dimensional behaviour could be producing the semblance of a transition. (2) In clean crystals is there always coincidence of decoupling and melting? How does pinning influence this coincidence? (3) Are the regions of the H –T plane fundamentally similar for YBCO and BSCCO but occurring at different fields and temperatures refelcting the different anisotropy. Based on the entropy change δS(T ) extracted from the magnetization jump in YBCO and BSCCO are there fundamental differences? (4) Is there any evidence for a line liquid? If the temperature is low enough, the pinning originating from static disorder in the superconducting material always becomes relevant. One of the lessons of recent research is that the specific nature of the defects that cause the pinning is important. Point pins, columnar pins, pinning by planes all induce very different behaviour. The following list of questions are what we believe is the most well defined and important issues to clear up when pinning cannot be neglected. (5) In isotropic point disordered crystals transport critical scaling analysis appears to show evidence for a vortex glass transition at the irreversibility line. In magnetization measurements, the field dependences of Uc and Jc cannot easily be reconsiled with glassy or collective pinning behaviour close to the irreversibility line measured at much lower electric fields. This inconsistency may be related to the fact that over most of the H –T plane, the magnetization measurement where J Jc , sets up plastic rather than elastic behaviour. Alternatively, is it plausible that the transport measurements where J is much closer to Jc is simply not sampling the transition effectively and cannot determine the nature of the solid? (6) In what way do columnar defects change the nature of the coupling between the planes in the more anisotropic materials? How does the influence they have on the reversible properties impact their influence on irreversible pinning behaviour? There is an inconsistency in the way that columnar defects influence reversible properties within 1–2 K of Tc , but do not act as effective pinning sites within 25–20 K of Tc . (7) In the presence of correlated disorder is there sufficient evidence to prove the Magnetic behaviour of superconductors 1665 existence of Bose glass behaviour? (8) What is the thermodynamic equilibrium phase of the vortex system at low temperatures in the presence of point disorder? (This question has not been addressed by the experiments discussed in this review.) The strive to understand the magnetic properties of the HTS has inspired an amazingly vigourous theoretical as well as experimental line of research. Although many questions are still open this research has been particularly fruitful in causing many new developments in the statistical mechanics and lead to a number of beautiful experiments. Not only has the research influenced basic science, in this way it has also laid the needed foundation for the phenomenological understanding needed to turn the HTS into technologically useful materials. This is a field of reseach with the potential of many new important developments in future years. Acknowledgments The authors would like to thank Yuri Bugislavsky, Richard Doyle, Gary Perkins and Sasha Zhukov for their critical reading of the text and insightful discussions. Support from the EPSRC (LC grant no GR/K60916, and HJJ grant no GR/J36952) and from the Royal Society are gratefully acknowledged. References Abrikosov A A 1957 Sov. Phys.–JETP 5 1174 Abulafia Y et al 1996 Phys. Rev. Lett. 77 1596 Abulafia Y, Shaulov A, Wolfus Y, Prozorov R, Burlachkov L, Majer D, Zeldov E, Vinokur V M and Yeshurun Y 1997 J. Low-Temp. Phys. (MOS96) 107 455–65 Abulafia Y, Shaulov A, Wolfus Y, Prozorov R, Burlachkov L, Yeshurun Y, Majer D, Zeldov E and Vinokur V M 1995 Phys. Rev. Lett. 75 2404 Aegerter C M, Lee S L, Keller H, Forgan E M and Lloyd S H 1996 Phys. Rev. B 54 15 661–4 Almasan C C, de Andrade M C, Dalichaouch Y, Neumeier J J, Seaman C L, Maple M B, Guertin R G, Kurie M V and Garland J C 1992 Phys. Rev. Lett. 69 3812 Ando Y, Komiya S, Kotaka Y and Kishio K 1995 Phys. Rev. B 52 3765–8 Angadi M A, Shen Z X and Caplin A D 1991 Physica 185–9C 2195–6 Arribere A, Pastoriza H, Goffman M F, de la Cruz F, Mitzi D B and Kapitulnik A 1993 Phys. Rev. B 48 7248 Bardeen J and Stephen M J 1965 Phys. Rev. A 140 1197 Bauhofer W, Biberacher W, Gegenheimer B, Joss W, Kremer R K, Mattausch H, Muller A and Simon A 1989 Phys. Rev. Lett. 63 2520 Beasley M R, Labusch R and Webb W W 1969 Phys. Rev. 181 682 Beauchamp K M, Rosenbaum T F, Welp U, Crabtree G W and Vinokur V M 1995 Phys. Rev. Lett. 21 3942–5 Berghuis P, Herzog R, Somekh R E, Evetts J E, Doyle R A, Baudenbacjer F and Campbell A M 1996 Physica 256C 13–32 Berlinsky A J, Fetter A L, Franz M, Kallin C and Soinien P I 1995 Phys. Rev. Lett. 75 2200 Bernhard C, Wenger C, Niedermayer C, Pooke D M, Tallon J L and Kotaka Y 1995 Phys. Rev. B 52 R7050 Bhattacharya S and Higgins M J 1993 Phys. Rev. Lett. 70 2617 Bhattacharya S, Higgins M J and Ramakrishnan T V 1994 Phys. Rev. Lett. 73 1699 Blatter G, Feigle’man M V, Geshkenbein V B, Larkin A L and Vinokur V M 1994 Rev. Mod. Phys. 66 1125–388 Blatter G and Geshkenbein V 1996b Phys. Rev. Lett. 77 4958–61 Blatter G, Geshkenbein V, Larkin A and Nordborg H 1996a Phys. Rev. B 54 72–5 Blatter G and Ivlev B I 1994 Phys. Rev. B 50 10 272 Blatter G, Ivlev B and Nordborg H 1993 Phys. Rev. B 48 10 448 Brandstätter G, Yang X, Sauerzpf F M and Weber H W 1995 Inst. Ohys. Conf Ser. No 148 ed D Dew-Hughes (Bristol: IOP) Brandt E H 1989 Phys. Rev. Lett. 63 1106 ——1995 Rep. Prog. Phys. 58 1465–594 1666 L F Cohen and H J Jensen Brawner D A, Ong N P and Wang Z Z 1993a Phys. Rev. B 47 1156–9 Brawner D A, Schilling A, Ott H R, Huag R J, Ploog K and von Klitzing K 1993b Phys. Rev. Lett. 71 785 Bulaevskii L N, Ledvij M and Kogan V G 1992 Phys. Rev. Lett. 68 3773 Bulaevskii L N, Vinokur V M and Maley M P 1996 Phys. Rev. Lett. 77 936 Busch R et al 1992 Phys. Rev. Lett. 69 522 Cai X Y, Gurievich A, Larbalastier D C, Kelley R J and Onellion M, Berger H and Margaritondo G 1994 Phys. Rev. B 50 16 774–7 Cambell A M and Evetts J 1972 J. Adv. Phys. 72 199 Caneiro G 1995 Phys. Rev. Lett. 75 521 Caplin A D, Cohen L F, Perkins G P and Zhukov A A 1994a Super. Sci. Tech. 7 412–22 Caplin A D, Perkins G P and Cohen L F 1995 Super. Sci. Tech. 8 366–7 Charalambous M 1991 Phys. Rev. Lett. 71 436 Charalambous M, Chaussy J, Lajay P and Vinokur V 1993 Phys. Rev. Lett. 71 436 Chen T and Teitel S 1995 Phys. Rev. Lett. 74 2792 Chikumoto N 1994 M 2 –HTSC III, Grenoble Chikomoto N, Konczykowski M, Motohira N and Malozemoff A P 1992 Phys. Rev. Lett. 69 1260 Cho J H, Maley M P, Fleshler S, Lacerda A and Bulaevskii L N 1994 Phys. Rev. B 50 6493 Civale L, McElfresh M W, Marwick A D, Holzberg F H, Feild C, Thompson J R and Christen D K 1991a Phys. Rev. B 43 13 732 Civale L, Matwick A D, McElfresh M W, Worthington T K, Malozemoff A P, Holtberg F H, Thompson J R and Kirk M A 1990 Phys. Rev. Lett. 65 1164 Civale L, Matwick A D, Worthington T K, Kirk M A, Thompson J R, Krusin-Elbaum L, Sun Y, Clem J R and Holtzberg F H 1991b Phys. Rev. Lett. 67 648 Civale L, Pasquini G, Levy P, Nieva G, Casa D and Lanza H 1996 Physica 263C 389–95 Cohen L F, Lacey D, Perkins G P, Caplin A D, Xu M, Dou S X, Klestov S A and Voronkova V L 1994a Proc. 2nd Int. Symp. of HTS Tunneling Phenomena (Donetsk, Ukraine) Cohen L F, Laverty J R, Perkins G P, Caplin A D and Assmus W 1993 Cryogenics 33 352–6 Cohen L F, Perkin G, Caplin A D, Zhukov A A, Klestov S A, Voronkova V L, Küpfer H, Wolf T and Abell S 1994c Proc. 7th Int. Workshop on Critical Currents in Superconductors ed H W Weber (Alpbach: World Scientific) p 271 Cohen L F, Totty J T, Perkins G K, Doyle R A and Ladowaki K 1997 Supercond. Sci. Technol. 10 195 Cohen L F, Zhukov A A, Perkins G K, Jensen H J, Klestov S A, Voronkova V L, Abell S, Küpfer H, Wolf T and Caplin A D 1994b Physica 230C 1–8 Cooper J R, Loram J W, Johnson J D, Monod P, Enriquez H, Hodby J and Changkang C 1997 3D XY scaling of the irreversibility line of YBO crystals Phys. Rev. Lett. 79 1730–3 Coppersmith S N, Inui M and Littlewood P B 1990 Phys. Rev. Lett. 64 2585 Coppersmith S N and Millis A J 1991 Phys. Rev. B 44 7799 Cubitt R et al 1993 Nature 365 407 Daemen L L, Bulaevskii L N, Maley M P and Coulter J Y 1993 Phys. Rev. Lett. 70 1167 ——1993 Phys. Rev. B 47 11 291 Daumling M, Seuntjens J M and Larbalestier D C 1990 Nature 346 332 Deak J, Hou L, Metcalf P and McElfresh M 1995 Phys. Rev. B 51 705 de Gennes P G 1966 Superconductivity of Metals and Alloys (New York: Benjamin) de la Cruz F, Lopez D and Nieva G (1994b) Phil. Mag. B 70 773 de la Cruz F, Pastoriza H, Lopez D, Geoffman M F, Aribere A and Nieva G 1994a Physica 235V 83–6 Dekker C, Eidelloth W and Koch R H 1992 Phys. Rev. Lett. 68 3347–50 Delin K A, Orlando T P, McNiff E J, Foner S, Vandover R B, Schneemeyer L F and Waszczak J 1992 Phys. Rev. B 46 11 092–101 Devereaux T P, Scalettar R T, Zimanyi G T, Moon K and Loh Y 1995 Phys. Rev. Lett. 75 4768 Doyle R A, Johnson J D, Hussey N E, Campbell A M, Balakrishnan G, Paul D McK and Cohen L F 1995a Phys. Rev. B 52 9368 Doyle R A, Liney D, Seow W S, Campbell A M and Kadowaki K 1995b Phys. Rev. Lett. 75 4520 Doyle R A, Seow W S, Yan Y, Campbell A M, Mochiku T, Kadowaki K and Wirth G 1996 Phys. Rev. Lett. 77 1155–8 Eltsev Y, Holm W and Rapp Ö 1994 Phys. Rev. B 49 333 Eltsev Y and Rapp Ö 1995 Phys. Rev. B 51 9419 ——1995 Phys. Rev. Lett. C 75 2446 Erb A, Genoud J Y, Marti F, Daumling M, Walker E and Flukiger R 1996 J. Low Temp. Phys. 105 1023–8 Magnetic behaviour of superconductors 1667 Faleski M C, Marchetti M C and Middleton A A 1996 Phys. Rev. B 54 12 427 Farrell D E 1994 Physical Properties of High Temperature Superconductors I–V ed D M Ginsberg (Singapore: World Scientific) Farrell D E et al 1996 Phys. Rev. B 53 11 807–16 Farrell D E, Kwok W K, Welp U, Fendrich J and Crabtree G W 1995 Phys. Rev. B 51 9148 Fastampa R, Giura M, Marcon R, Sarti S and Silva E 1993 Supercond. Sci. Technol. 6 53 Farrell D E, Rice J P and Ginsberg D M 1991 Phys. Rev. Lett. 67 1165 Feigel’man M V, Geshkenbein V B, Ioffe L B and Larkin A I 1993 Phys. Rev. B 48 16 641 Feigel’man M V, Geshkenbein V B, Larkin A I and Vinokur V M 1989 Phys. Rev. Lett. 63 2303 Feigel’man M V and Vinokur V M 1990 Phys. Rev. B 41 8986 Fendrich J A, Kwok W K, Giapintzakis J, van der Beek C J, Vinokur V M, Fleshler S, Welp U, Viswananathan H K and Crabtree G W 1995 Phys. Rev. Lett. 74 1210 Fischer K H 1993 Physica 210C 1–2 179–87 ——1994 Physica 235C 4 2691–2 Fischer P, Busch R, Neumuller H W, Ries G and Braux H F 1992 Supercond. Sci. Technol. 5 4403 Fisher M P A 1989 Phys. Rev. Lett. 62 1415 Fisher D S, Fisher M P A and Huse D A 1985 Phys. Rev. Lett. 55 2924 ——1991 Phys. Rev. B 43 130 Flesher S, Kwok W K, Welp U, Vinokur V M, Smith M K, Downey J and Crabtree G W 1993 Phys. Rev. B 47 14 448 Flippen R B, Askew T R, Fendrich J A and van der Beek C J 1995 Phys. Rev. B 52 R9882–5 Fuchs D T, Zeldov E, Majer D, Doyle R A, Tamegai T, Ooi S and Konczykowski M 1996 Phys. Rev. B 54 R796–9 Gammel P L, Bishop D J, Rice J P and Ginsberg D M 1992 Phys. Rev. Lett. 68 3343 Gammel P L, Huse D A, Kleinman R N, Batlogg B, Ogelsby C S, Bucher E, Bishop D J, Mason T E and Martensen K 1994 Phys. Rev. Lett. 72 278–81 Gammel P L, Schneemeyer L F and Bishop D J 1991 Phys. Rev. Lett. 66 953 Gao L, Xue Y Y, Hor P H and Chu C W 1991 Physica 177C 438–44 ——1992 Phys. Rev. B 46 14 325 Gao Z X, Osquiquil E, Maenhoudt M, Wuyts B, Libbrecht S and Bruynseraede Y 1993 Phys. Rev. Lett. 71 3210 Giamarchi T and Le Doussal P 1996 Phys. Rev. Lett. 76 3408 Giever I 1965 Phys. Rev. Lett. 15 825 Glazman L I and Koshlev A E 1991 Phys. Rev. B 43 2835 Gordeev S N, Jahn W, Zhukov A A, Kupfer H and Bush A A 1994 Proc. of 7th Int. Workshop on Critical Currents in Superconductors ed H Weber (Singapore: World Scientific) p 165 Gray K E 1994 Applied Supercond. 2 295–304 Gray K E, Hettinger J D, Miller D J, Washburn B R, Moreau C, Lee C, Glagola B G and Eddy M M 1996 Phys. Rev. B 54 3622–7 Gray K E and Kim D H 1993 Phys. Rev. Lett. 70 1693 Gray K E, Steel D G, Hettinger J D, Miller D J, Washburn B R, Ware M, Parkman J T, Yoder M E, Moreau C and Eddy M M 1997 IEEE Trans. Appl. Supercon. 7 1987–92 Griessen R 1990 Phys. Rev. Lett. 64 1674–8 ——1991 Physica 175C 315–23 Grigorieva I V, Steeds J W, Balakrishnan G and Paul D M 1995 Phys. Rev. B 51 3765 Grigorieva I V, Steeds J W and Sasaki K 1993 Phys. Rev. B 48 16 865 Gurevich A and Küpfer A 1993 Phys. Rev. B 48 6477–85 Gurevich A, Küpfer H, Runtsch B, Meier-Hirmer R, Lee D and Salama K 1991 Phys. Rev. B 44 12 090–3 Gyorgy E M, van Dover R B, Jackson K A, Schneemeyer L F and Waszczak J V 1989 Appl. Phys. Lett. 55 283 Hagen C W and Griessen R 1989 Phys. Rev. Lett. 62 2857–60 Hanaguri T, Tsuboi T, Maeda A, Nishizaki T, Kobayashi N, Kotaka Y, Shimoyama J and Kishio K 1996 Physica 256C 111–18 Hao Z and Clem J R 1991 Phys. Rev. B 43 7622 Hardy V, Wahl A, Hert S, Ruyter A, Provost J, Groult D and Simon Ch 1996 Phys. Rev. B 54 656–64 Hebard A F and Fiory A T 1982 Physica 109–110B 1637 Houghton A, Pelcovits R A and Sudbo A 1989 Phys. Rev. B 40 6763 Howson M, Overend N, Lawrie L D and Salamon M B 1995 Phys. Rev. B 51 11 984 Hubbard M A, Salamon M B and Veal B W 1996 Physica 259C 309 Hubener R P 1979 Magnetic Flux Structures in Superconductors (Springer Series in Solid State Sciences 6) (Berlin: Springer) 1668 L F Cohen and H J Jensen Huse D A and Majumdar S N 1993 Phys. Rev. Lett. 71 2473 Hussey N E, Carrington A, Cooper J R and Sinclair D C 1994 Phys. Rev. B 50 13 073–6 Hussey N E, Cooper J R, Doyle R A, Lin C T, Liang W Y, Sinclair D C, Balakrishnan G, Paul D M and Revcolevschi A 1996 Phys. Rev. B 53 6752–8 Hwa T, LeDoussal P, Nelson D R and Vinokur V M 1993 Phys. Rev. Lett. 71 3545 Ikeda R 1995 J. Phys. Soc. Japan 64 1683 Ikeda R, Ohmi T and Tsuneto T 1991 J. Phys. Soc. Japan 60 1051 Iye Y, Nakamra S and Tamegai T 1989 Physica 159C 433 Iye Y, Oguro I, Tamegai T, Datars W R, Motohira N and Kitazawa K 1992 Physica 199C 154–60 Jagla E A and Balseiro C A 1997 Phys. Rev. B 55 3192–9 Jahn W, Gordeev S N, Zhukov A A, Küpfer H and Wolf T 1995 Inst. Phys. Conf. Series 148 267–70 Jensen H J 1995 J. Phys. A: Math. Gen. 28 1861 Jensen H J, Brass A and Berlinsky A J 1988a Phys. Rev. Lett. 60 1676 Jensen H J, Brass A, Brechet Y and Berlinsky A J 1988b Phys. Rev. B 38 9235 Jensen H J and Minhagen P 1991 Phys. Rev. Lett. 66 1630 Jiang W, Yeh N-C, Reed D S, Kriplani U Beam D A, Konczykowski M, Tombrello T A and Holtzberg F 1994 Phys. Rev. Lett. 72 550 Jiang W, Yeh N-C, Reed D S, Kriplani U and Holtzberg F 1995 Phys. Rev. Lett. 74 1438 Jirsa M, Pust L, Schnack H G and Griessen R 1993 Physica 207C 85–96 Jirsa M, van Dalen A J J, Koblischka M R, Ravi Kumar G and Griessen R 1994 Critical Currents in Superconductors ed H W Weber (Singapore: World Scientific) p 221 Josephson B D 1965 Adv. Phys. 14 419 Kadowaki K 1996 Physica 263C 164 Kadowaki K, Songliu Y and Kitazawa K 1994 Supercond. Sci. Technol. 7 519–40 Keener C D, Trawick M L, Ammirata S M, Hebboul S E and Garland J C 1997 Phys. Rev. B 55 R708–11 Kes P H, Aarts J, Vinokur V M and van der Beek C J 1990 Phys. Rev. Lett. 64 1063 Kes P H, Vanderbeek C J, Maley M P, McHenry M E, Huse D A, Menken M J V and Menovsky A A 1991 Phys. Rev. Lett. 67 2383–6 Khaykovich B, Zeldov E, Majer D, Li W, Kes P H and Konczykowski M 1996 Phys. Rev. Lett. 76 2555 Kishio K, Shimoyama J, Kotaka Y and Yamafuji K 1994 Proc. 7th Int. Workshop on Critical Currents in Superconductors ed H W Weber (Singapore: World Scientific) p 339 Kiss T and Yamafuji K 1996 Physica 258C 197–212 Klein L, Yacoby E R, Tsameret A, Yesurun Y and Kishio K 1994 J. Appl. Phys. 75 6322–7 Klein L, Yacoby E R, Wolfus Y, Yeshurun Y, Burlachkov L, Shapiro B Y, Konczykowski M and Holtzberg F 1993b J. Appl. Phys. 73 5862–4 Klein L, Yacoby E R, Yesurun Y, Konczykowski M and Kishio K 1993b Phys. Rev. B 48 3523 Kleiner R and Müller P 1994a Phys. Rev. B 49 1327–41 Kleiner R, Müller P, Kohlstedt H, Pedersen N F and Sakai S 1994b Phys. Rev. B 50 3942–52 Klemm R A and Clem J R 1980 Phys. Rev. B 21 1968 Klemm R A, Luther A and Beasley M R 1975 Phys. Rev. B 12 877 Kobayashi N, Hirano K, Nishizaki T, Iwasaki H, Sasaki T, Awaji S, Watanabe K, Asaoka H and Takei H 1995a Physica 251C 255–62 Kobayashi N, Nishizaki T, Onodera Y, Asaoka H and Takei H 1995 Proc. 1995b Taiwan Int. Conf. of Superconductivity (Chinese Journal of Physics) Kobayashi T, Nakayama Y, Kishio K, Kimura T, Kitazawa K and Yamafuji K 1993 Appl. Phys. Lett 62 1830 Koblischka M R and Wijngaarden R J 1995 Supercond. Sci. Technol. 8 199–213 Koch R H, Foglietti V, Gallagher W J, Koren G, Gupta A and Fisher M P A 1989 Phys. Rev. Lett. 63 1511 Konczykowski M, Chikomoto N, Vinokur V M and Feigelman M V 1995 Phys. Rev. B 51 3957 Konczykowski M, Rullier-Albenque F, Yacoby E R, Shaulov A, Yeshurun Y and Lejay P 1991 Phys. Rev. B 44 7167–70 Konczykowski M, Vinokur V M, Rullier-Albenque F, Yeshurun Y and Holtzberg F 1993 Phys. Rev. B 47 5531–5 Korshunov S E and Larkin A I 1992 Phys. Rev. B 46 6395 Koshelev A E 1996a Phys. Rev. Lett. 76 1340–3 ——1996b Proc. 9th Int. Symp. on Superconductivity (ISS’96), Sapporo, Hokkaido, Japan to be published Koshelev A E, Glaseman L I and Larkin A I 1996 Phys. Rev. B 53 2786–91 Koshelev A E and Vinokur V M 1994 Phys. Rev. Lett. 73 3580 Kramer E J 1973 J. Appl. Phys. 44 1360 Krasnov V M, Larkin V A and Ryazanov V V 1991 Physica 174C 440–6 Magnetic behaviour of superconductors 1669 Kresin V Z and Wolf S A 1992 Phys. Rev. B 46 6458 Krusin-Elbaum L, Civale L, Blatter G, Marwick A D, Holtzberg F and Feild C 1994a Phys. Rev. Lett. 72 1914 Krusin-Elbaum L, Civale L, Vinokur V M and Holtzberg F 1992 Phys. Rev. Lett. 69 2280 Krusin-Elbaum L, Marwick A D, Wheeler R, Field C, Vinokur V M, Leaf G K and Palumbo M 1996 Phys. Rev. Lett. 76 2563–6 Krusin-Elbaum L, Thompson J R, Wheeler R, Marwick A D, Li C, Patel S and Shaw D T 1994b Appl. Phys. Lett. 64 3331–3 Küpfer H, Zhukov A A, Will A, Jahn W, Meier-Hirmer R, Wolf Th, Voronkova V L, Kläser M, Saito K 1996 Phys. Rev. B 54 54 644–55 Kwok W K, Fendrich J A, Fleshler S, Welp U, Downey J and Crabtree G W 1994b Phys. Rev. Lett. 72 1092–5 Kwok W K, Fendrich J A, Fleshler S, Welp U, Downey J, Crabtree G W and Giapintzakis J 1994 Physica 197B 579–87 Kwok W K, Fendrich J A, van der Beek C J and Crabtree G W 1994 Phys. Rev. Lett. 73 2614 Kwok W K, Fendrich J A, Welp U, Fleshler S, Downey J and Crabtree G W 1994a Phys. Rev. Lett. 72 1088–91 Kwok W K, Fleshler S, Welp U, Vinokur V M, Downey J, Crabtree G W and Miller M M 1992 Phys. Rev. Lett. 69 3370 Lacey D, Cohen L F, Perkins G and Caplin A D 1994 Physica 235–40C 2595 Landau L D and Lifshitz E M 1969 Statistical Physics 2nd edn (London: Pergamon) Larkin A I, Marchetti M C and Vinokur V M 1995 Phys. Rev. Lett. 75 2992 Larkin A I and Ovchinnikov Yu N 1979 J. Low Temp. Phys. 34 409 ——1986 Nonequilibrium Superconductivity ed D N Langenberg and A I Larkin (New York: Elsevier) Larkin A I and Vinokur V M 1995 Phys. Rev. Lett. 75 4666 Lawrence W E and Doniach S 1971 Proc. 12th Int. Conf. Low Temp. Phys. LT12 ed E Kanda (Kyoto: Academic Press of Japan) p 361 Le Blanc M A R and Little W A 1960 Proc. Seventh Int. Conf. Low Temp. Phys. (Toronto: University of Toronto Press) p 198 Lee S L et al 1993 Phys. Rev. Lett. 71 3862 ——1997 Phys. Rev. B 55 5666–9 Leghissa M, Gurevich L A, Kraus M, Saemannischenko G and Vinnikov L Y 1993 Phys. Rev. B 48 1341–4 Liang R, Bonn D A and Hardy W N 1996 Phys. Rev. Lett. 76 835 Li Q et al 1993 Phys. Rev. B 48 9877 Li Q, Fukumoto Y, Zhu Y M, Suenaga M, Kaneko T, Saa K and Simmon Ch 1996 Phys. Rev. B 54 1–4 Li Y H and Teitel S 1994 Phys. Rev. B 49 4136 Lopez D, Nieva G and de la Cruz F 1994a Phys. Rev. B 50 7219 Lopez D, Nieva G, de la Cruz F, Jensen H J and O’Kane D 1994b Phys. Rev. B 50 9684 Lopez D, Righi E F, Nieva G and de la Cruz F 1996 Phys. Rev. Lett. 76 4034–7 Mackenzie A P, Julian S R, Lonzarich G G, Carrington A, Hughes S D, Liu R S and Sinclair D C 1994 J. Supercon. 7 271 Majer D, Zeldov E and Konczykowski M 1995 Phys. Rev. Lett. 75 1166 Maki K 1964 Physica 1 127 Malozemoff A P 1991 Physica 185–9C 264–9 Malozemoff A P and Fisher M P A 1990 Phys. Rev. B 42 6784–6 Maley M P, Willis J O, Lessure H and McHenry M E 1990 Phys. Rev. B 42 2639 Martinez J C, Brongersma S, Koshelev A, Ivlev B, Kes P H, Gressen R, de Groot D G, Tarnavski Z and Menovky A A 1992 Phys. Rev. Lett. 69 2276 Mason T 1991 Private communication Mikheev L V and Kolomeisky E B 1991 Phys. Rev. B 43 10 431–5 Minnhagen P 1987 Rev. Mod. Phys. 59 1001 Miu L, Wagner P, Hadish A, Hillmer F, Adrian H, Wiesner J and Wirth G 1995 Phys. Rev. B 51 3953 Moon K, Scalettar R T and Zimànyi 1996 Phys. Rev. Lett. 77 2778 Moore M A 1997 Phys. Rev. B 55 14 136–9 Moore M A and Newman T J 1995 Phys. Rev. Lett. 75 533 Moshchalkov V V, Metlushko V V, Güntherodt G, Goncharov L N, Didyk A Y and Bruynseraede Y 1994 Phys. Rev. B 50 639 Nagel S R 1993 Phase Transitions and Relaxation in Systems with Competing Energy Scales ed T Riste and D Sherrington (Dordrecht: Kluwer) Nakamura N, Gu G D and Koshizuka N 1993 Phys. Rev. Lett. 71 915 Nelson D R 1988 Phys. Rev. Lett. 60 1973 1670 L F Cohen and H J Jensen Nelson D R and Vinokur V M 1992 Phys. Rev. Lett. 68 2398 ——1993 Phys. Rev. B 48 13 060 Nguyen A K, SudbøA and Hetzel R E 1996 Phys. Rev. Lett. 77 1592 Nideröst, Suter A, Visani P, Mota A C and Blatter G 1996 Phys. Rev. B 53 9286 Nordborg H and Blatter G 1996 Czech. J. Phys. 46 1817–8 Osofsky M S, Cohn J L, Skelton E F, Miller M M, Soulen Jr R J, Wolf S A and Vanderah T A 1992 Phys. Rev. B 45 4916 Osofsky M S et al 1993 Phys. Rev. Lett. 71 2315–8 ——1994 J. Supercon. 7 279 Ossandon J G, Thompson J G, Christen D K, Sales B C, Kerchner H R, Thomson J O, Sun Y R, Lay K W and Tkaczyk J E 1992 Phys. Rev. B 45 12 534–47 Oussena M, de Groot P A J, Deligiannis K, Volkozub A V, Gagnon R and Taillefer L 1996a Phys. Rev. Lett. 76 2259–62 ——1996b Phys. Rev. Lett. 77 792 Oussena M, de Groot P A J, Marshall A and Abell J S 1994 Phys. Rev. B 49 1484–7 Oussena M, de Groot P A J, Porter S J, Gagnon R and Taillefer L 1995 Phys. Rev. B 51 1389 Overend N, Howson M A, Lawrie I D, Abell S, Hirst P J, Chankang C, Chowdhury S, Hodby J W, Inderhees S E and Salamon M B 1996 Phys. Rev. B 54 9499 Palstra T T M, Batlogg B, Schneemeyer L F and Waszczak J V 1988 Phys. Rev. Lett. 61 1162 Palstra T T M, Batlogg B, Vandover R B, Schneemeyer L F and Waszczak J V 1989 Appl. Phys. Lett. 54 763 Pastoriza H, de la Cruz F, Mitzi D B and Kapitulnik A 1992 Phys. Rev. B 46 9278 Pastoriza H, Goffman M F, Arribere A and de la Cruz F 1994a Phys. Rev. Lett. 72 2951 Pastoriza H and Kes P 1995 Phys. Rev. Lett. 75 3525 Pastoriza H, Safar H, Righi E F and de la Cruz F 1994b Physica 194–6B 2237 Perkins P K and Caplin A D 1996 Phys. Rev. B 54 12 551–6 Perkins P K, Cohen L F, Zhukov A A, Caplin A D 1995 Phys. Rev. B 51 8513 Persico V, Cataudella V, Fontana F and Minnhagen P 1996 Physica 260C 41–51 Pippard A B 1969 Phil. Mag. 19 217 Polak M, Windte V, Schauer W, Reiner J, Gurevich A and Wuhl H 1991 Physica 174C 14 Pozek M, Ukraincyk I, Ratvin B and Dulcic A 1991 Europhys. Lett. 16 683 Pradhan A K, Roy S B, Chaddah P, Kanjilaal D, Chen C and Wanklyn D M 1996 Supercond. Sci. Technol. 9 1 Prozorov R, Tsameret A, Yeshurun Y, Koren G, Konczykowski M and Bouffard S 1994 Physica 234C 311–7 Pust L 1990 Supercond. Sci. Technol. 3 598 Pust L, Jirsa M, Griessen R and Schnack H G 1993 Appl. Supercon. 1 835 Pust L, Kadlecova J, Jirsa M and Durcok S 1990 J. Low Temp. Phys 78 P179 Rae A I M 1996 Czech J. Phys. 46 1807–8 Ravikumar G et al 1997 Physica 276C 9 Reed D S, Yeh N C, Konczykowski M, Samoilov A V, Holtzberg F 1995 Phys. Rev. B 51 16 448 Rodriguez E, Geoffman M F, Arribere A and de la Cruz F 1993 Phys. Rev. Lett. 71 3375 Roulin M, Junod A and Walker E 1996 Science 273 1210–12 Safar S et al 1990 Progress in High Temperature Superconductivity vol 25, ed R Nicolvsky (Singapore: World Scientific) p 140 Safar H, Gammel P L, Bishop D J, Mitzi D B and Kapitulnik A 1992b Phys. Rev. Lett. 68 2672–5 Safar H, Gammel P L, Huse D A, Bishop D J, Lee W C, Giapintzakis J and Ginsberg D M 1993 Phys. Rev. Lett. 24 3800 Safar H, Gammel P L, Huse D A, Bishop D J, Rice J P and Ginsberg M 1992c Phys. Rev. Lett. 69 824 Safar H, Gammel P L, Huse D A, Majumdar S N, Schneemeyer L F and Bishop D J 1994 Phys. Rev. Lett. 72 1272 Safar H, Rodriguez E, de la Cruz F, Gammel P L, Schneemeyer L F and Bishop D J 1992a Phys. Rev. B 46 14 238 Schilling A, Fisher R A, Phillips N E, Welp U, Dasgupta D, Kwok W K and Crabtree G W 1996 Nature 382 791–3 Schilling A and Jeandupeux O 1995 Phys. Rev. B 52 9714–23 Schilling A, Jin J, Guo J D and Ott H R 1993 Phys. Rev. Lett. 71 1899 Schindler W 1991 J. Appl. Phys. 70 1877–9 Schmidt B, Konczykowski M, Morozov N and Zeldov E 1996 Int. Workshop on Vortex Dynamics in HTS (Shoresh, Israel) Schmitt P, Kummeth P, Schults L and Saemann-Ischenko G 1991 Phys. Rev. Lett. 67 270 Schnack H G 1995 PhD Thesis 1995 Schnack H G, Griessen R, Lensink J G and Hai-hu W 1993 Phys. Rev. B 48 13 178 Magnetic behaviour of superconductors 1671 Schnack H G, Griessen R, Lensink J G, van der Beek C J and Kes P H 1992 Physica 197C 337–61 Schopohl N and Maki K 1995 Phys. Rev. B 52 490 Schuster Th, Koblischka M R, Moser N, Ludescher B and Kronmiller H 1991 Cryogenics 30 811 Schuster Th, Kuln H and Brandt E H 1995b Phys. Rev. B 51 697 Schuster Th, Kuln K, Indenbom M, Leghissa M, Kraus M and Konczykowski M 1995a Phys. Rev. B 51 16 358 Sengupta S, Shi D, Wang Z, Smith M E and McGinn P J 1993 Phys. Rev. B 47 5165–9 Seow W S, Campbell A M, Doyle R A, Kadowaki K, Wirth G and Koshelev A 1996 Phys. Rev. B 53 14 611–20 Silva E, Fastampa R, Giura M, Marcon R, Murtas F and Sarti S 1995 Appl. Magn. Res. 8 11–23 Smith J L et al 1994 J. Supercond. 7 269 Spencer S and Jensen H J 1997 Phys. Rev. B 55 8473 Steel D G and Greybeal J M 1992 Phys. Rev. B 45 12 643 Steel D G, Hettinger J D, Parkman J T, Yoder M E, Gray K E, Glagola B G and Eddy M M 1996 Physica 265C 159–62 Supple F, Campbell A M and Cooper J R 1995 Physica 242C 233–45 Sun J Z, Lairson B, Ecom C B, Bravman J and Geballe T 1990 Science 247 307 Sun Y R, Thompson J R, Chen Y J, Cristen D K and Goyal A 1993 Phys. Rev. B 47 14 481–8 Sun Y R, Thompson J R, Cristen D K, Holtzberg F, Marwick A D and Ossandon J G 1992 Physica 194C 403–10 Tachiki M and Takashasi S 1989 Solid State Commun. 72 1083 Tallon J L 1994 Proc. 7th International Workshop on Critical Currents in Superconductors ed H Weber (Singapore: World Scientific) p 52 Tang C, Ling X S, Bhattacharya S B and Chaikin P M 1996 Europhys. Lett. 35 597–602 Tes̆nović Z 1995 Phys. Rev. B 51 16 204 Tes̆anović Z and Xiang L 1991 Phys. Rev. Lett. 67 2729 Tes̆anović Z, Xing L, Bulaevskii L, Li Q and Suenaga M 1992 Phys. Rev. Lett. 69 3563 Thomas J, Perkins G K, Lacey D E, Cohen L F and Caplin A D 1996 Proc. 8th Int. Workshop on Critical Currents (Kitakyushu, Japan) ed T Matsishita and K Yamafuji (Singapore: World Scientific) Thompson J R, Krusin-Elbaum L, Civale L, Blatter G and Feild C 1997 Phys. Rev. Lett. 78 3181–4 Thompson J R, Sun Y R, Civale L, Malozemoff A P, McElfresh M W, Marwick A D and Holtzberg F 1993 Phys. Rev. B 47 14 440–7 Thompson J R, Sun Y R and Holtzberg F 1991a Phys. Rev. B 44 458 Thompson J R, Sun Y R, Kerchner H R, Chisten D K, Sales B C, Chakoumalos B C, Marwick A D, Civale L and Thompson J O 1992 Appl. Phys. Lett. 60 2306 Thompson J R, Sun Y R, Malozemoff A P, Christen D K, Kerchner H R, Ossandon J G, Marwick A D and Holtzberg F 1991b Appl. Phys. Lett. 59 2612–14 Thorel P, Hahn R, Simon Y and Cribier D 1973 J. Physique 34 447 Thuneberg E V 1984 J. Low Temp. Phys. 57 415 Tinkham M 1963 Phys. Rev. 129 2413 ——1995 Introduction to Superconductivity (New York: McGraw-Hill) Totty J T, Cohen L F, Perkins G K, Hossain A K M, Jemsen H J, Doyle R A, Kadowaki K and Wirth G 1996 Proc. 8th IWCC Japan (Singapore: World Scientific) p 389 Tsuneto T 1988 J. Phys. Soc. Japan 57 3499 Umezawa A, Crabtree G W, Liu J Z, Moran T J, Malik S K, Nunez L H, Kwok W L and Sowers C Ha 1988 Phys. Rev. B 38 2843–6 van Dalen A J J 1995 PhD Thesis Free University of Amsterdam van Dalen A J J, Griessen R and Koblischka M R 1996 Physica 257C 271–83 van der Beek C J, Kes P H, Maley M P, Menken M J V and Menovsky A A 1992 Physica 195C 307 van der Beek C J, Konczykowski M, Vinokur V M, Li T W, Kes P H and Crabtree G W 1995 Phys. Rev. Lett. 74 1214 van der Beek C J, Konczykowski M, Li T W, Kes P H and Benoit W 1996 Phys. Rev. B 54 R792 Vargas L J and Larbalestier D C 1992 Appl. Phys. Lett. 60 1741 Vinokur V M, Kes P H and Koshelev A E 1990 Physica 168C 29 ——1995 Physica 248C 179 Wan Y M, Hebboul S E and Garland J C 1994 Phys. Rev. B 72 3867 Wan Y M, Hebboul S E, Harris D C and Garland J C 1993 Phys. Rev. Lett. 71 157 Watauchi S, Ikuta H, Shimoyama J, Kishio K 1996 Physica 259C 373–8 Welp U, Fendrich J A, Kwok W K, Crabtree G W and Veal B 1996 Phys. Rev. Lett. 76 4809–12 Welp U, Kwok W K, Crabtree G W, Vandervoort K G and J Z Liu 1989 Phys. Rev. Lett. 62 1908 Werthamer N R, Helfand E and Hohenberg P C 1966 Phys. Rev. 147 295 1672 L F Cohen and H J Jensen Werner M, Sauerzopf F M, Webwe H W, Veal B D, Licci F, Winzer K and Koblischka M K 1994 Physica 235–40C 2833 Woo K C, Gray K E, Kampwirth R T, Kang J H, Stein S J, East R and McKay D M 1989 Phys. Rev. Lett. 63 1877 Worthington T K, Fisher M P A, Huse D A, Toner J, Marwick A D, Zabel T, Feild C A and Holtzberg F 1992 Phys. Rev. B 46 11 854–61 Wördenweber R and Kes P 1986 Phys. Rev. B 34 494 Wördenweber R, Kes P and Tsuei C C 1986 Phys. Rev. B 33 3172 Wu D H and Sridar S 1990 Phys. Rev. Lett. 65 2074 Wu H, Ong N P and Li Y Q 1993 Phys. Rev. Lett. 71 2642 Yang G, Abell J S and Gough C E 1994 Proc. 7th Int. Workshop on Critical Currents in Superconductors ed H Weber (Singapore: World Scientific) p 264 Yang G, Shang P, Sutton S D, Jones I P, Abell J S and Gough C E 1993a Phys. Rev. B 48 4054 Yang G, Sutton S, Shang P, Gough C E and J S Abell 1993b IEEE Trans. Appl. Supercond. 3 1663 Yaron U et al 1996 Nature 381 p 253 Yaron U, Gammel P L, Huse D A, Kleiman R N, Ogleby C S, Bucher E, Batlogg B, Bishop D J, Mortensen K and Clausen K N 1995 Nature 376 753–5 Yeshurun Y, Bontemps N, Burlachkov L and Kapitulnik A 1994a Phys. Rev. B 49 1548 Yeshurun Y and Malozemoff A P 1988 Phys. Rev. Lett. 60 2202 Yeshurun Y, Malozemoff A P and Shaulov A 1996 Rev. Mod. Phys. 68 911–49 Yeshurun Y, Yacoby E R, Klein L, Burlachkov L, Prozorov R, Bontemps N, Wuhl H and Vinokur V 1994b Proc. 7th IWCC, Alpbach ed H W Weber (Singapore: World Scientific) pp 237–9 Zech D, Lee S L, Heller H, Blatter G, Janossy B, Kes P H, Li T W and Menovsky A A 1995 Phys. Rev. B 52 6913 Zeldov E, Amer N M, Koren G, Gupta A, McElfresh M W, Gambino R J 1990 Appl. Phys. Lett. 56 680 Zeldov E, Larkin A I, Geshkenbein V B, Majer D, Konczykowski M, Vinokur V M and Shtrikman H 1994 Phys. Rev. Lett. 73 1428 Zeldov E, Majer D, Konczykowski M, Geshkenbein V B, Vinokur V M and Shtrikman H 1995a Nature 375 373 Zeldov E, Majer D, Konczykowski M, Larkin A I, Vinokur V M, Geshkenbein V B, Chikumoto N and Shtrikman H 1995b Europhys. Lett. 30 367 Zhukov A A 1992 Solid State Commun. 82 983–5 Zhukov A A, Küpfer H, Klestov S A, Voronkova V L and Yanovsky V K 1993 J. Alloy Compounds 195 479–81 Zhukov A A, Küpfer H, Perkins G, Cohen L F, Caplin A D, Klestov S A, Claus H, Voronkova V I, Wolf T and Wül 1995 Phys. Rev. B 51 12 704 Zhukov A A, Perkins G, Cohen L F, Kuper H, Totty J T, Caplin A D, Klestov S A, Claus H, Voronkova V I and Wuhl H 1994 Physica 235–40C 2837 Zhukov A A, Perkins G K, Thomas J V, Caplin A D, Kupfer H and Wolf T 1996 Proc. 8th IWCC Japan (Singapore: World Scientific) p 317 Zhukov A A, Perkins G, Thomas J V, Caplin A D, Kupfer H and Wolf T 1997 Phys. Rev. B 56 3481