JOURNAL OF APPLIED PHYSICS VOLUME 92, NUMBER 10 15 NOVEMBER 2002 Magnetic flux bifurcation and frequency doubling in rotated superconductors A. Badı́aa) Departamento de Fı́sica de la Materia Condensada-ICMA, CPS University of Zaragoza, Marı́a de Luna 1, E-50.018 Zaragoza, Spain C. López Departamento de Matemática Aplicada, CPS University of Zaragoza, Marı́a de Luna 1, E-50.018 Zaragoza, Spain 共Received 14 June 2002; accepted 1 September 2002兲 Starting from our variational statement of the general critical state in type II superconductors, we develop anisotropic current flow simulations in various conditions. The theory is applied to the slab geometry under rotating applied field, parallel to the surface of the sample. By comparison to the isotropic case, we show that anisotropy strongly influences the underlying physical phenomena. A magnetic-flux bifurcation point arises for the isotropic hypothesis. The issue of this point defines a boundary between two groups of vortex lines, one of which rigidly settles within the sample and another one, which frictionally rotates relative to the sample. A much more complex scenario arises in the anisotropic case, for which dynamic fronts must be defined. As a consequence, we predict the appearance of nonlinear phenomena such as the magnetization frequency doubling. The anharmonic contribution may be tuned by the applied field modulus. © 2002 American Institute of Physics. 关DOI: 10.1063/1.1516867兴 I. INTRODUCTION well below H c2 and allows to use the linear relation B⫽ 0 H. Although the main facts have been explained, some observations are still lacking a theoretical understanding. In particular, as it was indicated in Ref. 2, the appearance of higher frequency oscillations in the magnetic moment components, as the applied field modulus is increased, cannot be predicted by the available models. In this article, we will show that, as it was suggested by the authors of Ref. 2 such phenomena can be ascribed to anisotropy in the current flow. In particular, the investigation of the evolutionary field penetration profiles reveals a class of physical phenomena, such as the appearance of counter-rotating flux tubes within the sample. These mechanisms are behind the mentioned anharmonic effects. By the application of our variational statement of the critical state8 one can describe anisotropic specimens by a single continuous parameter ␥, which takes the limiting value of unity for the isotropic case. This will allow to elucidate the physical processes which directly relate to anisotropy. In brief, our method is based on the minimization of a cost functional which weighs the magnetic field changes within the sample. This is done under the constraint J苸⌬ with ⌬ a bounded set. In this work, we have explored two cases: 共i兲 ⌬ is an elliptic region with a given eccentricity and 共ii兲 ⌬ is a circle with variable radius, according to the magnetic field orientation with respect to some given axis. It will be shown that both models produce several distinctive features. However, the relevant underlying phenomena behind the experimental observations are jointly described and, thus, independent of the specific model in use. Rotating field experiments have been used by several groups in order to provide information on crossflow of flux lines in superconductors.1– 6 Typically, an irreversible type II superconducting disk is subjected to an applied magnetic field HS parallel to the flat faces. The disk is slowly rotated about the axis perpendicular to the surface and the components of magnetic flux parallel and perpendicular to HS monitored. The analysis of the experimental data has led to important concepts on the dynamics of nonparallel flux line lattices. In particular, Clem and Pérez-González7 have shown that most of the observed features can be well described by a general critical-state model which incorporates both the fluxline cutting and pinning concepts. The underlying idea is that the electrical current density vector J within the sample cannot exceed two critical values. When the component of J along the direction of the magnetic field exceeds J c 储 flux-line cutting between adjacent tilted lines occurs. When the component of J perpendicular to the field exceeds J c⬜ flux depinning occurs. As long as both phenomena take place with high associated resistivities, one can neglect relaxation processes between the successive metastable equilibrium states. In practical terms, the critical state approach makes sense if slow variations of the applied field are imposed. Then, the induced electric fields are small enough and one can neglect resistive losses. The second fundamental assumption on which the theory relies is that the reversible magnetization contribution to the sample’s response can also be neglected. This relates to strong pinning specimens well above H c1 and a兲 Electronic mail: anabadia@posta.unizar.es 0021-8979/2002/92(10)/6110/9/$19.00 6110 © 2002 American Institute of Physics Downloaded 30 Sep 2004 to 155.210.31.124. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp J. Appl. Phys., Vol. 92, No. 10, 15 November 2002 A. Badı́a and C. López In Sec. II we present the mathematical statement of the anisotropic critical state models. This encloses the timediscretized description of the successive field penetration profiles in an arbitrary magnetization process. The theory is then applied to the rotating field experiment 共Sec. III兲. The results are compared to the already established predictions of the isotropic model. In Sec. IV we summarize the main findings and discuss possible extensions of the theory. II. ANISOTROPIC CRITICAL STATE Consider a type II superconducting infinite slab with applied magnetic field parallel to the surfaces. We take the X axis perpendicular to the faces and the origin of coordinates at the midplane. Thus, the induced fields in the superconductor H, J, and E are also parallel to the YZ plane and only dependent on the coordinate x and the time. Anisotropy will be modeled as a property related to the Y and Z axes defined earlier. In this work, we attempt a phenomenological approach with highest simplicity, but enough physical content, so as to predict the experimental facts. Inspired by the isotopic model, which was shown to be the minimal theory for a number of observations in rotating and crossed field experiments,8,9 we propose two different approximations. First, we will consider that the critical current density obeys a tensor relation of the kind 兩 J y 兩 ⭐J c , 兩 J z 兩 ⭐ ␥ J c , with ␥ some anisotropy parameter. As a straightforward generalization of the isotropic law 兩 J兩 ⭐J c we introduce the rule J 2y ⫹ J z2 ␥ ⭐J 2c . 2 兩 J兩 ⭐J c 共 ␣ 兲 , the field penetration profiles corresponding to parallel flux line lattices either along Y or Z axes may be determined by Maxwell laws together with Eqs. 共3a兲 or 共3b兲. For H directed along an arbitrary direction, one needs to make some further statement on the restrictions upon J. In this section, we investigate the rule that J lies within the elliptic region given by Eq. 共1兲. This includes the isotropic model as the particular case for which ␥ ⫽1. Additionally, the one-dimensional problems of current flow along principal axes are also included as particular cases. The mathematical statement will be built on previous work,8 where we showed that the critical state can be understood as a variational problem with a restriction of the kind J苸⌬ for the current density. Here, ⌬ is defined by Eq. 共1兲. Now, Ampère’s law, together with the critical current restriction may be written as dH y ⫽␥uy dx 共4a兲 dH z ⫽u z , dx 共4b兲 with u a two-dimensional vector within the unit disk D. Notice, in passing, that Eqs. 共4a兲 and 共4b兲 arise when x is measured in units of the slab half-thickness a and H in units of J c a. Next, we require the minimization of the functional C关 Hn⫹1 共 x 兲兴 ⫽ 21 共1兲 It is apparent that Eq. 共1兲 may be expressed as 兩 J兩 苸⌬, ⌬ being an elliptic region with semiaxes J c and ␥ J c . In consequence, this will be named elliptic model hereafter. The second approximation will introduce the tensorial characteristic as a dependence of the critical current density on the magnetic field orientation. Namely, we will use the relation 共2兲 with ␣ the angle between the magnetic field and a given axis within the sample. This hypothesis may also be written as 兩 J兩 苸⌬ with ⌬ a circle of variable radius as a function of ␣. Thus, we will call it pseudoisotropic model. Let us assume that the superconductor possesses nonisotropic limiting values for the current density along the Y and Z axes. Just for convenience, these will be considered as the principal directions of the sample, i.e., 共 0,0,H z 兲 → 兩 J y 兩 ⭐J c 共3a兲 共 0,H y ,0兲 → 兩 J z 兩 ⭐ ␥ J c , 共3b兲 with J c the critical current density, which is taken to be constant along this article. The sample’s response to an external drive may be straightforwardly obtained by conventional critical state principles as far as oblique current flow is not induced. Thus, 冕 1 0 兩 Hn⫹1 ⫺Hn 兩 2 dx, 共5兲 which measures the magnetic field changes along successive time steps, satisfying Eqs. 共4a兲 and 共4b兲. As usual, Hn (x) stands for the magnetic field profile at the time layer n ␦ t. Recall that minimization is attained upon defining the associated Hamiltonian H⫽p y ␥ u y ⫹ p z u z ⫺ 21 共 Hn⫹1 ⫺Hn 兲 2 , 共6兲 and imposing the algebraic condition * ,p* ,u* 兲 ⫽maxH共 Hn⫹1 * ,p* ,u兲 . H共 Hn⫹1 共7兲 u苸D From the previous equation one can solve for the control variable u: u* ⫽ A. Elliptic model 6111 共 ␥ p ⬘y ,p z⬘ 兲 冑␥ 2 p *y 2 ⫹ p z* 2 . 共8兲 Eventually, the Hamiltonian equations, which hold the solution of the problem, read * dH n⫹1,y dx * dH n⫹1,z dx dp* y dx ⫽ ⫽ ␥ 2 p *y 冑␥ 2 p *y 2 ⫹ p z* 2 p z* 冑␥ 2 p *y 2 ⫹ p z* 2 * ⫺H n,y ⫽H n⫹1,y 共9a兲 共9b兲 共9c兲 Downloaded 30 Sep 2004 to 155.210.31.124. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp 6112 J. Appl. Phys., Vol. 92, No. 10, 15 November 2002 dp z* dx * ⫺H n,z . ⫽H n⫹1,z A. Badı́a and C. López 共9d兲 As usual, this set of equations requires an appropriate number of boundary conditions to be supplied. Continuity of H across the surface will produce two conditions at x⫽1. The remaining two will be determined according to the partial or full penetration regime for the field changes 共see Ref. 8 for details兲. Thus, the iterative solution of the system 关Eqs. 共9兲兴 for successive time steps will produce the field and current penetration profiles for a changing external excitation. A specific application of the previous formalism will be given in Sec. III, where an excitation rotating field with constant modulus H S at the sample’s surface is considered for various values of H S . B. Pseudoisotropic model Later we develop the pseudoisotropic hypothesis 兩 J兩 ⭐J c f 共 ␣ 兲 , 共10兲 in which the current density modulus is constrained by a certain function of the magnetic field tilt with respect to some axis within the sample. For definiteness, let us take it with respect to the Z axis and consider f 共 ␣ 兲 ⫽ ␥ sin2 ␣ ⫹cos2 ␣ . 共11兲 Notice that the earlier expression leads to the correct limit for ␥ ⫽1 and produces the required particular models for current flow along principal axes ( ␣ ⫽0, /2). This allows a direct comparison to the previous model. Just on following the same steps as before, and in terms of the same dimensionless units, the critical current restriction may be written as dH y ␥ H 2y ⫹H z2 ⫽ uy dx H2 共12a兲 dH z ␥ H 2y ⫹H z2 ⫽ uz , dx H2 共12b兲 with u belonging to the unit disk D. The minimization of the cost functional C关 Hn⫹1 (x) 兴 is now attained by maximizing the Hamiltonian H⫽ 2 2 ␥ H n⫹1,y ⫹H n⫹1,z 2 H n⫹1 1 p"u⫺ 共 Hn⫹1 ⫺Hn 兲 2 . 2 共13兲 This leads to the condition uÄp* /p * and one gets the Hamiltonian equations * dH n⫹1,y dx * dH n⫹1,z dx dp * y dx ⫽ ⫽ * 2 ⫹H n⫹1,z *2 p* y ␥ H n⫹1,y p* 共14a兲 *2 H n⫹1 * 2 ⫹H n⫹1,z *2 p z* ␥ H n⫹1,y p* 共14b兲 *2 H n⫹1 * ⫺H n,y ⫺2 p * ⫽H n⫹1,y * H n⫹1,z *2 共 ␥ ⫺1 兲 H n⫹1,y *4 H n⫹1 共14c兲 d p z* dx * ⫺H n,z ⫺2 p * ⫽H n⫹1,z * H n⫹1,y *2 共 1⫺ ␥ 兲 H n⫹1,z *4 H n⫹1 . 共14d兲 Again, this first order system must be solved by supplying appropriate boundary conditions. In the next section, this will be done for the rotating field experiment. III. ROTATIONAL PROPERTIES Here we apply the mathematical models developed earlier to analyze the magnetic response of a superconducting slab, rotating in a uniform magnetic field perpendicular to the axis of rotation. Just for convenience we use a reference frame which rotates with the sample and matches the anisotropy directions. Thus, we have an applied magnetic field at the surface HS (t)⫽H S (0,sin ␣S ,cos ␣S) with ␣ S ⫽ 0 t. 0 stands for a constant angular velocity, which is assumed to be small enough for relaxation effects to be neglected. The field penetration profiles H(x,t) will be derived for such a process. Eventually, the magnetic response of the sample will be obtained by using averaged values, i.e.: M⬅ 具 H(x) 典 ⫺HS . Although the rotational properties for isotropic samples have been well studied before, we will make a brief description of the main facts. This will allow to reveal the physical phenomena with are either hindered or promoted by anisotropy. A. Isotropic sample rotation The most outstanding experimental fact, which one can well describe within the isotropic model is the so-called magentic flux bifurcation. Such a phenomenon is clearly observed in field cooled experiments for which the sample holds a nonmagnetic initial profile 共0,0,H S ). This will be the situation considered hereafter. As rotation takes place, the process consists of the separation of the vortex lines in two groups, one which rigidly pins to the sample (V R ) and another one, which frictionally rotates relative to the sample (V F ). The inset in Fig. 1 illustrates this behavior. We have depicted the characteristic stationary state V shape profile for the field modulus, as well as several rotation angle profiles subsequent to the transient period. It is apparent that the vortices within the region V R are rigidly fixed to the sample 共constant modulus and angle兲, whilst those in region V F frictionally rotate relative to it. Although Fig. 1 has been obtained by application of our isotropic model ( ␥ ⫽1), we want to stress that such behavior can be predicted by other theoretical approaches.2,4,7 As a final observation we recall that the volume occupied by the region V R is progressively reduced as H S increases and vanishes for a characteristic field H S* , which penetrates to the center with maximum gradient. The experimental counterpart of the previously described process corresponds to the appearance of a stationary state with harmonic oscillations in the magnetization components. It is customary to record the magnetic moment com- Downloaded 30 Sep 2004 to 155.210.31.124. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp J. Appl. Phys., Vol. 92, No. 10, 15 November 2002 A. Badı́a and C. López 6113 ponents along the longitudinal and transverse directions to the applied field (M L ,M T ). The vortices within V R contribute as the projections of a constant modulus rotating vector and the vortices within V F as the projections of a vector with static components. Thus, stationary dc ⫽M L,T M L,T 共 H S 兲 ⫹M 0 共 H S 兲 sin共 0 t⫹ L,T 兲 . 共15兲 As indicated earlier, the amplitude factor M 0 (H S ) goes to zero when the region V R disappears. This leads to constant values M L ,M T , coming from the V F vortices which, in this case, occupy the whole sample. This transition defines the disappearance of the flux bifurcation point x 0 . We have also depicted this behavior in Fig. 1. B. Anisotropic elliptic model FIG. 1. Calculated longitudinal magnetization curves of an isotropic slab as a function of the rotation angle ␣ S relative to the applied field at the surface. Open symbols show the oscillations for low values of the applied field (H S ), and filled symbols correspond to high fields. The inset displays the field modulus H and angle rotation ␣ profiles within the sample for the stationary regime in the case of low fields. H has reached a fixed V shape. The angle variation is blocked below the decoupling point x 0 . Steady state solutions leading to harmonic magnetization components as given by Eq. 共15兲 will be no longer valid for anisotropic samples. It will be shown that, in this case, the magnetic moment of the sample must be described as a combination of the Fourier components 0 and 2 0 . Later, we discuss the distinctive features of rotation experiments analyzed within the elliptic model framework. For definiteness, we display the solution of Eqs. 共9兲 in the case ␥ ⫽0.5. The FIG. 2. Rotation angle ␣ and field modulus H penetration profiles as calculated according to the elliptic model for a field cooled rotation experiment. The horizontal axis displays the normalized coordinate within the slab 共midplane x⫽0 and surface x⫽1). Corresponding modulus and angle curves have been plotted in the same column and using a common line style. Decoupling points for which field cancellation occurs and counter-rotation starts have been labeled (x 0,1 and x 0,2). x A labels the boundary with rigid vortices within the sample. The rotation angle is expressed in radians and the magnetic field modulus in dimensionless units. The insets display the rotation stages for which penetration profiles were plotted. Downloaded 30 Sep 2004 to 155.210.31.124. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp 6114 J. Appl. Phys., Vol. 92, No. 10, 15 November 2002 A. Badı́a and C. López boundary value problem for the nonlinear ordinary differential equations of our system has been solved by a finitedifference approximation. 1. Low-field rotation First, we analyze the rotation experiment for low values of the applied field (H S ⫽0.1 and H S ⫽0.25 in our dimensionless units兲. As it will become clear later, this means that the midplane is not reached by the perturbation. Again, we choose the modulus and angle representation, which is most significant for rotation experiments. Figure 2 shows the results in the case H S ⫽0.25 for several intermediate stages when the sample undergoes two turnabouts (0 ⭐ ␣ S ⭐4 ). As the sample turns from 0 to 2, flux consumption produces the well known behavior observed for isotropic cases. The rotation angle penetration regime evolves towards a step shape. Simultaneously, the field modulus becomes V-shaped, and a decoupling point 关 H(x 0,1⫽0 兴 is defined. However, when rotation proceeds, the profile H(x) keeps no longer stationary. On the contrary, a complex structure with two minima arises. Notice, in addition that flux lines are no longer blocked for x⬍x 0,1 . The region of rigid vortices is shifted to 0⬍x⬍x A , with x A as the boundary. Now, the negative angle values for x A ⬍x⬍x 0,1 correspond to a counter rotation of the flux tubes, subsequent to the decoupling step. A second turnabout 共2 – 4兲 completes the full picture of a periodic fashion in the physical phenomena described earlier. Note that the system evolves again towards a V shape for the modulus and a step-like angle profile, defining a second blocking point x 0,2 . Again, this is accompanied by counter rotating flux lines for x A ⬍x⬍x 0,2 . We notice that the flux consumption previous to decoupling is asymmetric and this is the reason why one gets a shifted cancellation point and a branch point for the rotation angle. 2. High-field rotation We consider now the case that H S is large enough for magnetic field changes to reach the midplane as the sample rotates. The characteristic behavior has been displayed in Fig. 3. Basically, ␣ (x) shows a quasilinear penetration profile as rotation proceeds. On the other hand, subsequent to a transient period, the field modulus evolves in between well defined boundary profiles. This evolution takes place in a nearly harmonic stationary regime at the frequency 2 0 . 3. Magnetization curves Finally, we analyze the experimental fingerprint of the field penetration profiles described above. In Fig. 4 we have plotted the longitudinal and transverse components of the averaged magnetization for three values of the applied field modulus. Notice that, for the lowest field value, M L,T basically display a stationary harmonic behavior at the sample’s rotation frequency 0 . As H S increases, one can readily observe a frequency mixing phenomenon. This corresponds to the competition between the fundamental frequency contribution from the rigid vortices region, and the double fre- FIG. 3. Rotation angle ␣ and field modulus H penetration profiles as calculated according to the elliptic model for a field cooled rotation experiment. In this case, the high value of the magnetic field (H S ⫽1) produces a full penetration regime. Symbols have been used as a guide of the evolutionary successive profiles. quency arising from the frictionally rotated vortices. Eventually, for high enough fields, the rigid core disappears and M becomes a rotating vector at frequency 2 0 . C. Pseudoisotropic model In this part we present the rotational characteristics of the anisotropic slab as calculated by the pseudoisotropic model equations 共Sec. II B兲. For comparison purposes, the set of Eqs. 共14兲 has been solved for ␥ ⫽0.5 and the same boundary conditions as in Sec. III B. Again, we provide numerical solutions obtained by a finite-difference algorithm. 1. Low-field rotation Figure 5 depicts the modulus and angle penetration profiles for the rotation experiment at H S ⫽0.25. Several stages have been selected as the sample turns from 0 to 4. On purpose, we have chosen the curves for which decoupling occurs and some nearby steps. We emphasize that, again, a decoupling point (x 0,1) arises in the first turn 共0–2兲, followed by flux transport and counter-rotation. Also, a second turnabout 共2 – 4兲 leads to a V shape for the modulus profile, as well as a bifurcation point for the rotation angle. As one could expect, the values of x A ,x 0,1 , and x 0,2 are close but not coincident with the corresponding quantities in the elliptic model simulations. Downloaded 30 Sep 2004 to 155.210.31.124. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp J. Appl. Phys., Vol. 92, No. 10, 15 November 2002 A. Badı́a and C. López 6115 FIG. 4. Longitudinal and transverse components of the averaged magnetization as a function of the rotation angle ␣ S for a field cooled slab at different values of the applied magnetic field. Calculations have been made for the current density elliptic model. 2. High-field rotation The calculated field penetration profiles in this case display similar properties to the elliptic model predictions, but some differences appear. Basically, the field modulus develops a stationary state behavior, in which also evolves in between an upper and lower boundary. However, these boundaries are not so rigid and some intermediate profiles can develop a small hump beyond the boundary 共see Fig. 6兲. Notice also the vertical scale for the field modulus, which indicates the prediction of a smaller field variation in the present model 共compare to Fig. 3兲. 3. Magnetization curves The magnetization tracings M L,T ( ␣ S ) for three values of the applied magnetic field are shown in Fig. 7. As predicted in the previous model 共see Fig. 4兲, subsequent to a transient period, we get nearly harmonic oscillations for low fields, a frequency mixing phenomenon for intermediate values, and a double frequency oscillation for high fields. Having fixed all the parameters of the theory, the two models just differ in some features of the double frequency response. Notice the phase shift and smaller amplitude for the pseudoisotropic curves. IV. CONCLUSIONS In this work we have explored the application of our variational statement of the critical state8,9 to anisotropic systems. Theoretical predictions have been made, which confirm the long suspected relation of nonlinear effects in rotation experiments to anisotropy in the critical current density. Mathematically, the theory is posed as a minimization problem under a constraint of the kind J苸⌬ for the current density. Two different phenomenological selections of ⌬ have been studied: 共i兲 ⌬ is an elliptic region and 共ii兲 ⌬ is a circle with variable radius, depending on the local magnetic field orientation relative to some axis within the sample. Though quite difference in their statement, both models provide quite similar predictions, in good agreement with experiment. We obtain harmonic oscillations in the magnetic moment when the sample is rotated at low values of the Downloaded 30 Sep 2004 to 155.210.31.124. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp 6116 J. Appl. Phys., Vol. 92, No. 10, 15 November 2002 A. Badı́a and C. López FIG. 5. Same as Fig. 2, but calculated according to the pseudoisotropic model. applied field. These oscillations take place at the sample’s rotation frequency. Moderate field values, however, involve frequency mixing phenomena, while high field oscillations occur at a nearly second harmonic regime. By comparison of the magnetic field penetration profiles obtained within our anisotropic models, to the predictions of the isotropic model 共⌬ is a circle兲, we can reveal the physical phenomena on which the magnetic properties rely. For field cooled isotropic systems, a stationary rotational state is established in which two groups of vortices exist: a rigid core well within the rotating sample and a region of flux tubes near the surface which frictionally rotate relative to the sample, but remain at rest with respect to the applied field. Both regions are connected by a zero field point 共decoupling point x 0 ). The rigid core contributes as a harmonically oscillating magnetic moment and the rotating flux as a constant signal in the laboratory reference. On increasing the applied magnetic field H S , the boundary x 0 is shifted towards the center of the sample, and the oscillations are gradually reduced until their amplitude becomes null. When anisotropy in the critical current density is allowed, one gets a quite different scenario. The rigid core may still exist, but multiple decoupling points must be defined. Specifically, a modulus cancellation and angle branching point appears for each turn of the sample. However, as rotation proceeds, flux transport occurs and the zero field point disappears until the next turn. On the other hand, we obtain 2 steps in the rotation angle which are related to flux counter-rotation. FIG. 6. Same as Fig. 3, but calculated according to the pseudoisotropic model. Downloaded 30 Sep 2004 to 155.210.31.124. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp J. Appl. Phys., Vol. 92, No. 10, 15 November 2002 A. Badı́a and C. López 6117 . 共16b兲 FIG. 7. Same as Fig. 4, but calculated according to the pseudoisotropic model. As a result of the earlier described events for anisotropic systems, the frictionally rotating vortices do not contribute as a constant signal and one gets a superimposed doublefrequency rotation. Eventually, the second harmonic signal at high fields relates the filed modulus bouncing between well defined boundaries, instead of the frozen profile for the isotropic case. From the mathematical point of view, as well as for physical interpretation purposes, our variational principle is a convenient, but nonunique formulation of the critical state in superconductors. For instance, the use of Maxwell equations, together with appropriate constitutive J(E) laws is a standard practice. Of particular relevance to the present work is to mention that an alternative model has been posed in terms of nonlinear diffusion of electromagnetic fields.10 In that reference 共see Chap. 4兲 the author uses Faraday’s law in the form ⵜ 2 E⫽ 0 t J(E), for constitutive relations, which in the critical state and within our selection or coordinate axes read J y ⫽k 共 1⫹ ⑀ 兲 Ey 冑共 1⫹ ⑀ 兲 E 2y ⫹ 共 1⫺ ⑀ 兲 E z2 , 共16a兲 J z ⫽k 共 1⫺ ⑀ 兲 Ez 冑共 1⫹ ⑀ 兲 E 2y ⫹ 共 1⫺ ⑀ 兲 E z2 Here k and ⑀ are parameters accounting for the superconductor properties. These equations are valid both for isotropic ( ⑀ ⫽0) and anisotropic 共elliptic兲 cases. Later, we show that this model and ours are equivalent in such situations. The key point is that the transient electric field, which arises when the penetration profile evolves from the equilibrium state Hn (x) to Hn⫹1 (x) may be basically identified with the momentum p(x) in our theory. p(x) was introduced as a Lagrange multiplier, but here we give a definite physical interpretation of this variable. Notice that Eqs. 共9c兲 and 共9d兲 are nothing bu the time-discretized form of Faraday’s law in convenient dimensionless units for E 共defined by 0 J c a 2 / ␦ t): dp* y dx * ⫺H n,y ⬅ ⫽H n⫹1,y dE z ⇒E z ⫽ p * y , dx 共17a兲 Downloaded 30 Sep 2004 to 155.210.31.124. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp 6118 J. Appl. Phys., Vol. 92, No. 10, 15 November 2002 dp z* dx * ⫺H n,z ⬅⫺ ⫽H n⫹1,z dE y ⇒E y ⫽⫺p z* . dx A. Badı́a and C. López 共17b兲 Of course, these equalities hold safe arbitrary constants. However, is we recall that p must vanish either for free boundary conditions 关partial penetration, p(x * )⫽0] or when the midplane is reached by field changes 关full penetration, p(0)⫽0], these constants must be zero. The physical counterpart of the former relations are the continuity conditions for the electric field 关leading to E(x * )⫽0 at free boundaries兴 and the antisymmetry with respect to the midpoint 关 E(0) ⫽0 兴 . On substituting Eqs. 共17a兲 and 共17b兲 in the Hamiltonian relations and recalling Ampère’s law, we get the corresponding constitutive equations in our formalism J y ⫽J c J z ⫽J c 冑 Ey , 2 E y ⫹ ␥ 2 E z2 ␥ 2E z 冑E 2y ⫹ ␥ 2 E z2 . 共18a兲 共18b兲 The equivalence of this system and Eqs. 共16a兲 and 共16b兲 is apparent upon using the relations 1⫹ ⑀ ⫽ J 2c k 2 1⫺ ⑀ ⫽ ␥ 2 k2 共19兲 where g ⫾ are constants related to the sample’s demagnetizing factor,  stands for the rotation axis misalignment relative to the sample’s surface vector, and ␣ S is once more used for the rotation angle. Finally, we will comment on some issues in prospect. This work explores the consequences of using anisotropic constraints on the current density vector (J苸⌬) within the critical state of hard superconductors. A reasonable understanding of available experiments has been provided. However, our work is developed at a phenomenological level and a rule for the selection of ⌬ would be of interest. Two research lines are suggested: 共i兲 from a more fundamental point of view, we propose the consideration of the flux line lattice and pinning sites interactions in anisotropic systems with flux cutting phenomena, and 共ii兲 it would be also desirable to design experiments which allow a well-posed inverse problem, i.e., such that one can unambiguously reconstruct ⌬ on the basis of experimental data. ACKNOWLEDGMENTS The authors are indebted to Professor J. R. Clem for helpful discussions and suggestions. Financial support from Spanish CICYT 共Project Nos. MAT99-1028 and BFM-20001066/C0301兲 is acknowledged. , J 2c M L ⫽⫺H S 关 g ⫹ ⫹g ⫺ 共 sin2  cos 2 ␣ S ⫺cos2  兲兴 . An important remark from the practical point of view is that double-frequency oscillations could also be obtained for isotropic samples, as a result of the rotation axis misalignment. However, this can be considered as a small correction unless for very noticeable deviations. Basically, the correction depends on the square of the error angle. In order to see how this factor arises one can start with the relations of Ref. 11 where the longitudinal and transverse magnetic moment of a flat sample under oblique field have been obtained for the Meissner state. For instance, after some algebra, the longitudinal magnetization component may be written as R. Boyer and M. A. R. LeBlanc, Solid State Commun. 24, 261 共1977兲. R. Boyer, G. Fillion, and M. A. R. LeBlanc, J. Appl. Phys. 51, 1692 共1980兲. 3 J. R. Cave and M. A. R. LeBlanc, J. Appl. Phys. 53, 1631 共1982兲. 4 L. Liu, J. S. Kouvel, and T. O. Brun, Phys. Rev. B 43, 7859 共1991兲. 5 H. P. Goeckner and J. S. Kouvel, Phys. Rev. B 50, 3435 共1994兲. 6 M. K. Hasan, S. J. Park, and J. S. Kouvel, Physica C 254, 323 共1995兲. 7 J. R. Clem and A. Pérez-González, Phys. Rev. B 30, 5041 共1984兲. 8 A. Badı́a and C. López, Phys. Rev. Lett. 87, 127004 共2001兲. 9 A. Badı́a and C. López, Phys. Rev. B 65, 104514 共2002兲. 10 I. D. Mayergoyz, Nonlinear Diffusion of Electromagnetic Fields (with applications to eddy currents and superconductivity) 共Academic, New York, 1998兲. 11 S. Candia and L. Civale, Supercond. Sci. Technol. 12, 192 共1999兲. 1 2 Downloaded 30 Sep 2004 to 155.210.31.124. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp