cond-mat0009088/d.tex
1: \documentstyle[aps,psfig,epsf]{revtex}
2: 
3: \begin{document}
4: 
5: \def\overlay#1#2{\setbox0=\hbox{#1}\setbox1=\hbox to \wd0{\hss #2\hss}#1
6: -2\wd0\copy1}
7: 
8: \twocolumn[\hsize\textwidth\columnwidth\hsize\csname@twocolumnfalse\endcsname
9: \title  {Critical currents in Josephson junctions  with macroscopic
10: defects}
11: \author{N. Stefanakis, N. Flytzanis}
12: \address{ Department of Physics, University of Crete,
13:  P.O. Box 2208, GR-71003, Heraklion, Crete, Greece}
14: \date{\today}
15: \maketitle
16: \begin{abstract}
17: \begin{center}
18: \parbox{14cm}
19: {
20: 
21: The critical currents  in Josephson junctions of
22: conventional superconductors with macroscopic defects are
23: calculated for different defect critical current densities as a
24: function of the magnetic field. We also study the evolution of the
25: different modes with the defect position, at zero external field. We
26: study the stability of the solutions and derive simple arguments,
27: that could help the defect characterization. In most cases a reentrant
28: behavior is seen, where both a maximum and a minimum current exist.
29: }
30: \end{center}
31: \end{abstract}
32: \pacs{}
33: \vskip2pc]
34: 
35: \tighten
36: 
37: \section{Introduction} The interaction of localized magnetic flux
38: (fluxons) with defects (natural or artificial) or impurities in
39: superconductors or junctions has an important effect in the properties
40: of bulk superconductors or the behavior of Josephson junctions
41: correspondingly \cite{conf}. The flux trapping from defects which is
42: of major importance in Josephson junctions \cite{barone} can modify
43: the properties of polycrystalline materials with physical
44: dislocations, for example grain boundary junctions \cite{dimos}. In
45: this category one can also consider grain boundary junctions in
46: YBa$_2$Cu$_3$O$_7$ \cite{sarnelli} where the tunneling current is a
47: strongly varying function along the boundary. This strong
48: inhomogeneity makes them good candidates for SQUID type structures
49: \cite{gross1}. Phenomenologically the current-voltage characteristics
50: of grain boundary junctions are well described \cite{gross2} by the
51: resistively shunted junction model \cite{mccumber}.
52: The grain boundary lines often tend to curve, while the junction is
53: very inhomogeneous and contains nonsuperconducting impurities and
54: facets of different length scales\cite{miller,ayache}.
55: The linear increase of the critical current with length in grain boundary
56: junctions with high-$T_c$ superconductors, which is  a different
57: behavior from the saturation in the inline geometry of a perfect
58: junction, can be explained by the presence of impurities\cite{fehren}.
59: Therefore it is interesting to study flux trapping in impurities
60: especially when it can be controlled. Modern fabrication techniques
61: can with relative ease engineer any defect configuration in an
62: extremely controlled way.
63: 
64: In bulk materials are several types of defects that can
65: influence the critical current in high temperature superconductors
66: like YBa$_2$Cu$_3$O$_x$ materials. They include 3d
67: inclusions, 2d grain boundaries and twin boundaries, and point defects
68: like dopants substitutions, oxygen vacancies\cite{conf}. For example
69: the homogeneous precipitation of fine Y$_2$BaCuO$_5$ non-superconducting
70: particles in the melt processing of YBa$_2$Cu$_3$O$_x$ leads to high $J_c$
71: values due to the particle pinning centers\cite{oka}.
72:  Similar behavior is observed in NdBa$_2$Cu$_3$O$_x$ bulk crystals with
73: Nd$_4$Ba$_2$Cu$_2$O$_{10}$ particles\cite{takagi}. Of interest is also the
74:  case of the peak effect in twin-free $Y123$ with oxygen deficiency.
75:  In this case, one  sees a linear increase (peak effect) of the
76:  critical current at small magnetic fields, when growth is under
77:  oxygen reduction\cite{wolf}. For the fully oxidized crystal one
78:  expects a decrease. The peak effect is attributed to flux trapping.
79:  Information on the defect density and activation energies  can also
80:  be obtained from the I-V characteristics, as was the case for several
81:  types of defects which were also compared to $Au^+$ irradiated
82:  samples with artificial columnar defects\cite{camerlingo}. These
83:  columnar defects also act to trap flux lines in an YBCO film which is
84:  considered as a network of intergrain Josephson junctions modulated
85:  by the defects. In this case assuming a distribution of contact
86:  lengths one finds a plateau in the critical current density vs. the
87:  logarithm of the field\cite{mezzetti}.
88: 
89: The study of long size of impurities is going to give information
90:  beyond theories which concern small amplitude of inhomogeneities
91:  \cite{virocur}. Also it is possible for a direct comparison of the
92:  numerical results with experiments in long junctions obtained with
93:  electron beam lithography \cite{kroger}. This is a powerful technique
94:  which allows the preparation and control of arrays of pinning
95:  centers. Another method is the ionic irradiation which produces a
96:  particular kind of disordered arrays, consisting of nanosized
97:  columnar defects \cite{camerlingo,beek}.  The variation of the
98:  critical current density can also occur due to temperature gradients
99:  \cite{krasnov}.
100: 
101:    The activity in the area of high critical current
102:  densities in the presence of a magnetic field is hampered by defects
103:  due to the difficulty of having a high quality junction with a very
104:  thin intermediate layer. Thus significant activity has been devoted,
105:  since for example the energy resolution of the SQUID
106: \cite{ketchen} and the maximum operating frequency of the single
107: flux quantum logic circuit \cite{likharev}, to name a few
108: applications,  depend inversely and directly respectively on the plasma
109: frequency  $\omega_p$, with $\omega_p \sim J_c^{1/2}$. The fundamental
110: response frequency of Josephson devices, the Josephson frequency
111: $\omega_J$, also depends on the critical current density. On the other
112: hand, a drawback is that high critical current densities  lead to
113: large subgap leakage currents \cite{kleinsasser} and junction
114: characteristics degrade rapidly  with increasing $J_c$.
115: 
116: Variations in the critical current density also influence the $I-V$
117:  characteristics introducing steps under the influence of both a static
118:  bias current and the irradiation with microwaves \cite{reinisch}. In
119:  that case the variation is quite smooth (of $sech$ type ), so that
120:  the fluxon and its motion could be described by a small number of
121:  collective coordinates. Interesting behavior is also seen in both
122:  the static and dynamic properties for the case of a spatially
123:  modulated $J_c$ with the existence of "supersoliton" excitations
124:  \cite{oboznov,larsen} and the case of columnar         defects
125:  \cite{tinkham,balents} or disordered defects \cite{fehren}.
126: 
127:  The trapping of fluxons can be seen in the $I_{max}(H)$ curves where
128:  we also expect important hysteresis phenomena when scanning the
129:  external magnetic field. The hysteresis can be due to two reasons:
130:  (i) One is due to the non-monotonic relation between flux and
131:  external magnetic field \cite{caputo} arising from the induced
132:  internal currents, and (ii) from the trapping or detrapping of
133:  fluxons by defects.
134:  The effect of a defect on a fluxon and the
135:  strength of the depinning field depends strongly in the size of the
136:  defect, the type of defect and the position of the defect. Here we
137:  will consider case where the widths of the defects is of the order
138:  of the Josephson penetration depth. In this range we expect the
139:  strongest coupling between fluxons and defects. We will also consider
140: the case of a few defects in the low magnetic field region where
141: pinning and coercive effects are important.
142: 
143:  The organization of
144:  the paper is as follows. In Sec. 2 the sine-Gordon model for a
145:  Josephson junction is presented. In Sec. 3 we present the results
146:  of the critical current $I_{max}$ versus the magnetic field of a
147:  junction with an asymmetrically positioned  defect. The variation of
148:  the $I_{max}$ and the flux content  $N_f$ with the defect critical
149:  current density and the position are presented in sections 4 and 5
150: respectively. The effect of multiple pinning centers is examined in
151: sections 6 and 7. In Sec. 8 we examine a defect with a smooth
152: variation of the critical current density. In the last section we
153: summarize our results.
154: % \end{document}
155: 
156: \section{The junction geometry} The
157: electrodynamics of a long Josephson junction is characterized from the
158: phase difference $\phi(x)$ of the order parameter in the two
159: superconducting regions. The spatial variation of $\phi(x)$ induces a
160: local magnetic field given by the expression
161: \begin{equation}
162: {\cal H}(x) =  \frac{d { \phi}(x)}{dx}.~~~\label{phix},
163: \end{equation}
164: in units of $H_0=\frac{\Phi_0}{2\pi d \lambda_J}$, where $\Phi_0$ is
165: the quantum of flux, $d$ is the magnetic thickness and $\lambda_J$ is
166: the Josephson penetration depth. The magnetic thickness is given
167: by $d=2\lambda_L +t$ where $\lambda_L$ is the London penetration
168: depth in the two superconductors and $t$ is the oxide layer thickness.
169: The $\lambda_J$ is also taken as the unit of length. The current
170: transport across the junction is taken to be along the $z$ direction.
171: We describe a 1-D junction with width $w$ (normalized to $\lambda_J$)
172: in the $y$ direction, small compared to unity. The normalized length
173: in the $x-$direction is  $\ell$. The superconducting phase difference
174: $\phi(x)$ across the defected junction is then the solution of the
175: sine-Gordon equation \begin{equation} \frac{d^2 {
176: \phi}(x)}{dx^2} = \widetilde{J}_c(x)\sin[{\phi(x)}],~~~\label{eq01}
177: \end{equation}
178: with the inline boundary condition \begin{equation}
179: \frac{d { \phi}}{dx}\left|_{x=\pm\frac{\ell}{2}}\right. =\pm
180: \frac{{I}}{2}+H,~~~\label{eq02} \end{equation} where $I$ and $H$ are
181: the normalized bias current and external magnetic field. $\widetilde
182: J_c(x)$ is the local critical current density which is $\widetilde
183: J_c=1$ in the homogeneous part of the junction and $\widetilde
184: J_c=j_d$ in the defect. Thus the spatially varying critical current
185: density is normalized to its value in the undefected part of the
186: junction $J_0$ and the $\lambda_J$ used above is given by
187: \[\lambda_J=\sqrt{\frac{\Phi_0}{2\pi \mu_0 d  J_0}},\] where $\mu_0$
188: is the free space magnetic permeability. One can also define a
189: spatially dependent  Josephson penetration depth  by introducing
190: $\widetilde{J}_c(x)$ instead of $J_0$. This is a more useful quantity
191: in the case of weak distributed defects.
192: 
193: In the case of overlap boundary conditions Eqs. (\ref{eq01}) and
194: (\ref{eq02}) are modified as
195: \begin{equation} \frac{d^2 {
196: \phi}(x)}{dx^2} = \widetilde{J}_c(x)\sin[{\phi(x)}] -I,~~~\label{eq0a}
197: \end{equation}
198: and \begin{equation}
199: \frac{d { \phi}}{dx}\left|_{x=\pm\frac{\ell}{2}}\right. =
200: H.~~~\label{eq0b} \end{equation}
201: 
202: 
203: We can classify the
204: different solutions obtained from Eq. (\ref{eq01}) with their magnetic
205: flux content \begin{equation}  N_f = \frac{1}{2 \pi}
206: (\phi_R-\phi_L) ,~~~\label{phi} \end{equation}
207: in units of $\Phi_0$, where $\phi_{R(L)}$ is
208: the value of $\phi(x)$ at the right(left) edge of the junction.
209: Knowing the magnetic flux one can also obtain  the magnetization from
210: \begin{equation} M= \frac{2\pi}{\ell} N_f-H. ~~~~~\label{magn}
211: \end{equation}
212: For the perfect junction, a quantity of interest is the critical
213: magnetic field for flux penetration from the edges, denoted by
214: $H_{c1}$. For a long junction it is equal to 2 while for a short it
215: depends on the junction length. Due to the existence of the defect
216: this value can be modified since we have the possibility of trapping
217: at the defects. For a short junction we have penetration of the
218: external field in the junction length, so that the magnetization
219: approaches zero. For a long junction it is a non-monotonic function of
220: the external field $H$.
221: 
222: To check the stability
223: we consider small perturbations $u(x,t)=v(x)e^{st}$ on the static
224: solution $\phi(x)$, and linearize the time-dependent sine-Gordon
225: equation to obtain:\begin{equation}   \frac{d^2 v(x)}{dx^2}
226: -\widetilde J_c(x)\cos\phi(x) v(x)= \lambda v(x) ,~~~\label{eq10}
227: \end{equation} under the boundary conditions
228: $$\frac{dv(x)}{dx}|_{x=\pm \frac{\ell}{2}}=0,$$ where $\lambda=-s^2$.
229: It is seen that if the eigenvalue equation has a negative eigenvalue
230: the static solution $\phi(x)$ is unstable. There is considerable
231: eigenvalue crossing so that we must monitor several low eigenvalues.
232: This is especially true near the onset of instabilities.
233: 
234: \section{Asymmetric defect} In the following
235: we will consider the variation of the maximum critical current as a
236: function of the magnetic field for several defect structures. We
237: start with a long ($L>\lambda_J$)  junction of normalized length
238: $\ell=10$  with a defect
239: of length $d=2$ which is placed $D=1.4$ from the right edge. Thus the
240: defect is of the order of $\lambda_J$. We plot in Fig. 2a the
241: maximum critical current $I_{max}$ variation with the magnetic field.
242: The different curves correspond to phase distributions for which we
243: have a maximum current at a given value of the magnetic field $H$. The
244: overlapping curves called modes have different flux content as seen in
245: Fig. 2b where we plot the magnetic flux in units of $\Phi_0$ for zero
246: current versus the external field. The magnetic flux is only a weak
247: function of the external current.
248: 
249: For the perfect junction there is no
250: overlap  in the magnetic flux between the different modes. In fact
251: each mode has flux content between $n\Phi_0$ and $(n+1)\Phi_0$ and
252: therefore is labelled the ($n,n+1$) mode \cite{caputo}. Here in the
253: case of the defect the range (at zero current) of flux for each mode
254: can be quite different and the labelling is with a single index
255: $n=0,1,2, ...$ corresponding in several cases to the $(0,1)$, ($1,2$),
256: ($2,3$),... modes of the perfect junction. There are in several cases
257: several modes with similar flux. To distinguish them we add a letter
258: following the index $n$.
259: 
260: The maximum $I_{max}$ is obtained for mode $1$  and the increase
261: comes from the  trapping of flux by the defect. We have to note that
262: the $(d,e)$ part of this mode is a continuation of the $(a,b)$ part of
263: mode $0$. In both cases we have entrance of flux from the no defect
264: part of the junction and the instability in the critical current
265: occurs when $\phi(-\ell/2)=\pi$. Here and in the following  we will
266: take this to mean equal to $\pi$ modulo $2\pi$. For the maximum
267: current (at $H < 0$) the equation is $H-I/2=-2$. This can be
268: understood from the pendulum phase diagram, where the $\phi_x=-2$ is
269: the extremum slope, and thus the relation $I_{max}=4+2H$ holds. For
270: the $(b,c)$ part of mode $0$ the flux enters from the right where the
271: defect is. This reduces the critical current compared to the perfect
272: junction $0$ mode \cite{caputo,owen}. Note that $0$-mode has its
273: critical current $I_{max}$ peak slightly to the left of $H=0$ in the
274: $I_{max}$ vs $H$ diagram, and to the left of $N_f=0$ in an $I_{max}$
275: vs $N_f$ diagram (see Fig. 11a). Also in the absence of current,
276: reversing the  direction of $H$ only changes the sign of the slope
277: $d\phi/dx$, but the phase difference (in absolute value) at the two
278: ends will be the same. Thus the $0$ mode at $I=0$ extends for $-1.6\le
279: H \le 1.6$. This is not clearly seen due to curve overlapping in the
280: left side. Comparing the $I_{max}$ for the modes $1$ and $-1$ we see
281: that $I_{max}(1)>I_{max}(-1)$. In both cases a fluxon (or antifluxon)
282: is trapped in the defect. The major difference in the $I_{max}$ comes
283: mainly from the phase distribution which in the mode $1$ case leads to
284: a large positive net current in the undefected side, while in the $-1$
285: mode the net current in the undefected side is very small.
286: 
287: For the mode $1$, at $H\approx 0$ and
288: zero current the instability happens due to the competition of the
289: slope of the phase at the defect center and at the right edge, while
290: at the other end at $H=2$, the field at the defect center becomes
291: equal to the external field applied at the boundaries and there is no
292: such competition. In this case the instability sets in due to the
293: critical value of the phase at the undefected boundary (i.e.
294:  $\phi_x(-\ell/2)=2$). The situation is analogous for
295: the mode $-1$. For $H\approx 0$ the instability sets in due to the
296: depinning of the antifluxon while for $H=-2$ due to the critical
297: value of the phase at the free defect part of the junction.
298: For the mode $0$ we have no fluxon trapping at the defect,
299: even though the instability at the two extremes with $H=\pm1.6$ 
300: at zero current is
301: caused from the tendency to trap a fluxon or antifluxon
302: correspondingly at the defect.
303:  At higher
304: values of the magnetic field ($|H|>1.6$)  we have stability
305: for a range of non-vanishing current values as will be discussed
306: below.   Thus this value can be considered as the minimum value for
307: the introduction of fluxons in the junction. Let as remark that for
308: the perfect junction, or a junction with a centered defect, the
309: corresponding values for fluxon introduction would be equal to $2$.
310: Thus there is a decrease of the critical field as the defect moves
311: away from the center. For the $0$ mode a centered defect would have
312: no influence on the solution.
313: 
314: The results for the maximum current are in agreement with
315: the stability analysis. In Fig. 2c we present the lowest eigenvalue
316: $\lambda_1$ for the different modes in zero external current $I=0$ as
317: a function of the magnetic field $H$. The sudden change in slope 
318: for the modes $-1, 1$ is
319: because at that point a new eigenvalue becomes lower. The $\lambda_1$
320: is positive denoting stability and becomes zero at the critical value
321: of the magnetic field, where a mode terminates. The symmetry about
322: zero magnetic field is due to the symmetric boundary conditions for
323: $I=0$.  Change of the sign of $H$ changes the sign of the phase
324: distribution, but the $\cos\phi$ in (\ref{eq10}) remains unchanged.
325: This symmetry is being lost when a finite current is also applied.
326: Also there are solutions (not presented in the figure) for which the
327: stability analysis gives negative eigenvalues i.e instability. These
328: solutions may be stabilized when we insert multiple impurities.
329: 
330:   In Fig. 3a
331: we specifically draw only the $1$ mode, to be discussed in more
332: detail. Here we changed the procedure, in searching for the maximum
333: current. Up to now we followed the standard experimental procedure,
334: i.e. we scan the magnetic field and for each value of $H$ we increase
335: the current $I$, starting from $I=0$, until we reach the maximum
336: current. Here we consider the possibility that for $I>0$ there is
337: also a lower bound in the value of the current for some values of the
338: magnetic field. This requires a search where we vary both $H$ and
339: $I$ simultaneously. Thus we see that for $H<0$ there is a lower
340: bound given approximately by the line $H+I/2\approx H_{cl} $, where
341: $H_{cl}\approx 0$ is the critical value of $H$ at $I=0$, for which we
342:  have depinning of the trapped fluxon. Over this curve the slope
343:  $\phi_x$ at the right end (near defect) is kept constant and equal to
344:  $H_{cl}$ and above this line the fluxon remains pinned and it should
345:  be stable. This line ends at $H=-1$, since in that case the extremum
346:  value $\phi_x=-2$ is reached in the left end. Increasing now in that
347:  range of $H$ the bias current we find that also the $I_{max}$ curve
348:  extends further to the left. The equation for this line is
349:  approximately given by $H-I/2=-2$, with an extremum at
350:  $\phi_x(-\ell/2)=-2$. Thus the instability on this line arises from
351:  the left side (far from the defect). It extends up to $H=-1$ for a
352:  long junction and joins the other line $H+I/2=H_{cl}$.
353: 
354:  The above calculations were done for a
355:  long junction so that the fields at the two ends do not interfere.
356:  For shorter length however the two ends feel each other and in that
357:  case the two instabilities are not independent. This means that the
358:  tail of the defect free side field will compete with the slope of
359:  the trapped field. Then the two lines $H+I/2=H_{cl}$, and
360:  $H-I/2=-2$, end  before they meet (at $H\approx-1$) at a cutoff
361:  magnetic field. Also for short junctions we expect the straight
362:  lines to have some curvature.
363:  A similar discussion holds for the
364:  right end of the $1$ mode. Again there is a lower current
365:  (positive) bound given by $H-I/2=2$ due to instability at the left end
366:  ($\phi_x(-\frac{\ell}{2})=2$), and an upper bound given by
367:  $H+I/2=H_{cr}$, where $H_{cr}\approx 2.8$, due to fluxon depinning.
368:  On the same diagram, we show the lower bound for negative currents.
369:  Thus we see that there is strong asymmetry for positive and negative
370:  currents. Remark that for negative currents the mode $1$ is very
371:  similar to the mode (1,2) with no defect \cite{caputo}. This is
372:  because the right boundary is determined by an instability at the
373:  undefected side. The left boundary is again very close because
374:  $H_{cl}\approx 0$. So an interesting effect of the defect is that we
375:  have this strong asymmetry for positive and negative currents.
376: 
377:  We would get a similar picture if we considered
378:  the $-1$ mode. In fact we get the same curves (as for mode $1$) if we
379:  put $I\rightarrow -I$ and $H\rightarrow -H$. This is consistent with
380:  the $-1$ mode shown in Fig. 2a. The  discussion can also be
381:  extended to the other modes. In Fig. 3b we show the result of a
382:  similar scan for the $0$ mode, but for the sake of shortness we will
383:  not discuss the $-1,-2,2$ modes. In any case when the number of
384:  fluxons increases one must rely on numerical calculations rather than
385:  simple arguments.
386: 
387:  \section{Variation with the defect critical current}
388:  In the previous section we considered the case of a microresistance
389:  defect. With present day masking  techniques we
390:  can also consider any finite critical current (lower or higher) in
391:  the defect. This situation can also arise very often in junctions
392:  with high critical current densities, where small variations in the
393:  thickness can create strong critical current density variations. Thus
394:  for the previous asymmetric defect configurations we will study the
395:  effect of the defect critical current density in the magnetic
396:  interference pattern $I_{max}(H)$. We will concentrate on the $0$ and
397:  $1$ modes.
398: 
399:  {\it (i) mode $0$:}
400: 
401:  In Fig. 4  we see the $I_{max}(H)$ variation
402:  for the mode $0$ for decreasing values of the critical current
403:  density from $j_d=2$, to $j_d=0$. Let us discuss first the case
404:  for $j_d \le1$. For the perfect junction where $j_d=1$ we have a
405:  symmetric distribution about $H=0$. As we decrease $j_d$ the flux
406:  content of this mode (and the extremum $H$) is symmetrically reduced
407:  (see Fig. 2b). It is not apparent from the drawing, due to the
408:  superposition of several curves on the left side of the diagram, but
409:  as expected the range of the magnetic field is symmetric about $H=0$
410:  at zero current. The corresponding $I_{max}(H)$ curves, however, are
411:  not symmetric. The right hand side of the curves is displaced towards
412:  smaller critical fields with decreasing $j_d$.  This means that the
413:  critical field at $I=0$ to introduce a fluxon from the ends is
414:  decreased due to the existence of the defect which acts with an
415:  attractive force on the fluxon. The curves are linear and can be
416:  approximated by the equation $I(H)=4-2(H+\delta H_c)$, where $\delta
417:  H_c$ is the decrease in the critical field $H_{c0}$ and depends on
418:  $j_d$. A similar decrease happens for negative magnetic fields where
419:  the defect tries to pin an antifluxon. Even for higher currents the
420:  right side critical field is determined by the tendency of the defect
421:  to attract a fluxon.  The left hand side, however remains rigid (but
422:  is shifted along the line). This is due to the entrance of magnetic
423:  flux from that part of the junction where there is no defect. The
424:  instability in the critical current occurs when ($\phi(-\ell/2)=\pi$)
425:  for every value of $j_d$. From the pendulum phase diagram which is
426:  the classical analog of the Josephson junction, the extremum occurs
427:  at $\partial_x\phi (-\frac{\ell}{2})=-2$, or $H-I/2=-2$ which is the
428:  equation for this triangular side. At near zero current the critical
429:  field is influenced from the attractive action of the defect. At low
430:  currents and extreme negative magnetic field the $I_{max}$ curve
431:  shows a re-entrance behavior so that it is not stable at low and high
432:  currents, but only for a finite intermediate range of current values.
433:  This way we reconcile the different origins of the instability
434:  mechanisms $\phi(-l/2)=\pi$ at high current and the defect influence
435:  discussed for the right hand side of the mode.
436: 
437: For $j_d>1$ we see an increase in the $I_{max}$, while  the critical
438: magnetic field at $I=0$ remains almost constant at about
439: $H_{cr}\approx 1.9$. The instability at that point is due to the
440: trapping of flux in the region between the positive defect and the
441: right edge of the junction. The field for that is expected to be near
442: $H=2$ if the right undefected part is of length of the order of
443: Josephson length. Thus it is the same value for flux penetration from
444: the perfect junction edges. It will vary weakly with $j_d$.
445: 
446:  {\it (ii) mode $1$:}
447: 
448:  The mode $1$ in the perfect
449:  junction has a full fluxon for magnetic field $H=0.07$. The phase
450:  distribution is about $\theta=\pi$ where the energy has a minimum.
451:  At the end of this mode at $H=2.07$ where two fluxons have entered
452:  the junction the phase changes from $\phi(-\ell/2)=-\pi$ to
453:  $\phi(\ell/2)=3\pi$. When the defect is inserted this mode is
454:  significantly modified due to the flux trapping in the defect.
455: 
456:  In Fig. 5 we see the magnetic interference pattern for this mode for
457:  different values of the defect critical current density.
458:  For $0 < j_d < 0.7$ the $I_{max}$ vs $H$ curves are displaced
459:  downwards, and a fluxon is trapped in the defect. We notice that all
460:  the curves for $j_d<0.7$ have the same critical magnetic field
461:  $H=2$ for $I=0$. This is because at this end of the mode, at $I=0$
462:  the instability arises at the side with no defect where the phase
463:  reaches the critical value $\phi=\pi$ (modulo $2\pi$). Of course as
464:  discussed in the previous section we have a reentrant behavior
465:  above $H=2$. At the other end for small magnetic field the instability
466:  is due to depinning of the trapped fluxon.
467:   For $0.7 < j_d \le 1.0$ the defect can trap the flux only for
468:  $H<H_{cd}$, where the value of the $H_{cd}$ depends on the defect
469:  critical current $j_d$ and in Fig. 5 it is shown for $j_d=0.9$.
470:  Notice that for this value of $j_d$ the fluxon is very weakly
471:  trapped, and the untrapping process happens slowly over a range of
472:  magnetic field values. For $H > H_{cd}$ the fluxon has moved away from
473:  the defect, and for this weak defect the junction does not feel it.
474:  The critical current goes abruptly close to the curve for the
475:  perfect junction. We conclude that the behavior of the junction for
476:  values of $j_d$ close to $j_d=1$ is determined by the ability of the
477:  defect to trap one fluxon. This can be seen also from the change in
478:  the lowest eigenvalue variation with the external field $H$, at
479:  values of the critical density $j_d > 0.7$, in Fig. 6.
480: 
481: For $j_d>1$ (thin lines in Fig. 5) it has a similar form as for
482: $j_d=1$, i.e. there is no fluxon trapping. Again, as in the $0$ mode,
483: the $H_{cr}$  at $I=0$ stays around $2.0$ and is again due to the
484: trapping of flux in the right edge.
485: 
486: 
487:  In Fig. 7 we present the evolution of $I_{max}$  with
488:  the defect critical current density $j_d$ for a magnetic field
489:  $H=1.5$. For this value of the magnetic field there are no solutions
490:  with trapped fluxons for $j_d > 0.83$. The lowest eigenvalue at
491:  $I=0$ becomes zero at this point. For $j_d > 0.83$ and $H > 1.5$ there
492:  are solutions which are not trapped. For these solutions the maximum
493:  current coincides with the one of the perfect
494:  junction and there is a discontinuity in the curves. Notice the point
495:  at $j_d=1.0$. In the same figure we also show the magnetic flux at
496:  $I=0$ and at $I_{max}$ which is almost constant as a function 
497:  of $j_d$ as expected, with
498:  small difference between the two different current curves.
499: 
500:  \section{Variation with the defect position} In Fig. 8a
501:  we see the evolution of the critical current at zero magnetic field
502:  as we move the defect from the right edge of the junction $D=0$ to
503:  the left edge where $D=8$. The position is measured from the edge of
504:  the junction to the nearest edge of the defect. We examine the
505:  several modes separately:
506: 
507:  {\it (i) mode $0$}
508: 
509:  For this mode and for
510:  $I=0$, we are able to find solutions for all the defect positions.
511:  As we can see in Fig. 8b the corresponding magnetic flux at $I_{max}$
512:  is slowly changing and equal to zero when the defect is in the
513:  junction center.
514:  But when the defect is placed close to the ends the magnetic flux at
515:  the maximum current deviates from zero. The
516:  critical current for this mode is symmetric for defect
517:  positions about the junction center, and has its maximum value when
518:  the defect is at the center. This is because at that position it
519:  does not influence the solution at the edges which is very close to
520:  the undefected case, while near the center the phase is almost
521:  zero. But when the defect comes close to the junction ends the
522:  defect cuts into the area by which the current flows, and the
523:  critical current is reduced.
524: 
525:  For even smaller distances $D=0.2$, and
526:  $D=0$ there is a jump to  solutions which correspond to a
527:  current, which is much higher than that of the $0$ mode for
528:  nearby $D$ values. This is because the defect cuts negative current
529:  regions  and for this position we have an increase of the critical
530:  current. In fact these solutions (see $++$ symbols in Fig. 8a) are
531:  very close to the solutions of a perfect junction within the
532:  undefected area, except that now the defect at the edge can give
533:  contribution to the flux but no contribution to the current. Thus the
534:  flux is much higher than that of the $0$ mode and it approaches that
535:  of mode $1$.  Nevertheless these points should be considered as a
536:  separate mode. In fact they are part of a branch (crosses). In these
537:  distances there are no other modes for $H=0$. Similar results were
538:  obtained by Chow $et$ $al.$ \cite{chow} where they attributed this
539:  enhancement in the $I_{max}$ for small distances to a self field
540:  which was generated by the current, penetrating into the defect and
541:  resisted any further penetration of field. To overcome this
542:  resistance it was necessary to apply a higher current. But they do
543:  not distinguish between modes with different flux content, and their
544:  evolution with the defect position.
545: 
546:  {\it (ii) modes $1$, $-1$}
547: 
548:  For these modes we do not have solutions for all the defect positions
549:  at $I=0$ and $H=0$, but only in the range $1.4 < D < 6.6$ as seen in
550:  Fig. 8c where the lowest eigenvalue is plotted as a function of the
551:  defect position for the different modes. The curves for the 1 and -1
552:  modes coincide, while the 0 mode shows a change of slope
553:  corresponding to the last two points ( $D=0 $ and 0.2 discussed above)
554:  which belong to another curve.  Mode $-1$ has a trapped antifluxon in
555:  the defect. When the defect is to the left ($4.0 < D <6.5$), then the
556:  instability in the current of mode $-1$ at $H=0$ is created at the
557:  right-end of the junction when the phase reaches the critical value
558:  $\phi(l/2)=\pi$.  This instability occurs for currents which are less
559:  than those necessary to unpin the antifluxon. Notice that increasing
560:  the current there is no competition with the slope of the antifluxon
561:  trapped in the left end. Thus at this point  (for $ 4.0 < D < 6.5 $)
562:  the maximum current is very close to the undefected junction mode
563:  $0$, except that in this case $N_f\approx -1$ is close to an
564:  antifluxon.  At the other end ($D < 4.0$) the instability for mode
565:  $-1$ is caused by the depinning action of the applied current, which
566:  takes now much smaller values (close to zero) because of competition
567:  with the pinned fluxon.  The phase distribution at the defect free
568:  end is that expected for $H=0$ and $I$ close to zero. The mode $1$
569:  with a fluxon trapped has a symmetrically reflected (about the
570:  center)  form in $I_{max}$ vs $D$ and the instability for $D>4.0$
571:  occurs at the left end of the junction, which is the opposite case of
572:  mode $-1$. The eigenvalue becomes zero at the positions $D=1.4$ and
573:  $D=6.6$. The $\lambda_1(D)$ curve coincides for the modes $1$ and
574:  $-1$ due to the fact that the phase distributions for the same $D$
575:  for these modes are symmetric about $x=L/2$, and the $\cos\phi(x)$
576:  that enters the eigenvalue equation is the same.
577: 
578:  In the rest we
579:  examine the variation of the critical value at which the instability
580:  sets in, as we scan the magnetic field in the positive (negative)
581:  direction $H_{cr} (H_{cl})$ for zero current, for the different
582:  modes, as a function of defect position. This instability can be
583:  attributed to the pinning, or the depinning field or to the
584:  critical value of $\frac{d\phi}{dx}$ at the defect free  edge,
585: depending on the particular mode that we are considering.
586: Explicitly for the mode $0$
587: the instability in the $H_{cl} (H_{cr})$ is due to the pinning of a
588: fluxon (antifluxon), respectively. In this mode the defect has no
589: influence for positions close to the center as seen in Fig. 9a and 9b.
590: However as we move the defect close to the edges the pinning field
591: $H_{cl} (H_{cr})$ is reduced in absolute value because it is easier to
592: trap  a fluxon (antifluxon).
593: For the mode $1$ the $H_{cr}$ is constant for
594: all defect positions. This is due to the fact that at $I=0$ it is
595: the phase distribution at the undefected  edge of the junction that
596: determines the instability. Notice that due to the reentrant character
597: the critical  magnetic field takes higher values at larger
598: bias currents which vary with defect position. The $H_{cl}$ curve
599: depends on the phase distribution near the defect and therefore is
600: strongly defect position dependent.
601: For the mode $-1$ the
602: picture is reversed compared with the $1$ mode. In this case the
603: $H_{cl}$ is constant while the $H_{cr}$ varies with position. Note
604: that in this mode the depinning of an antifluxon is the reason that
605: causes the instability at $H_{cr}$.
606: 
607: \section{Two symmetric pinning centers}
608: As noted  defects (with $j_d<1$) or inhomogeneities
609: in the junction can play the role of pinning centers for a fluxon. In
610: this section we discuss more precisely the effect of multiple pinning
611: centers on the magnetic interference patterns $I_{max}(H)$ and the
612: flux distribution. The pinning effect of the Josephson junction has
613: also been analyzed in \cite{yamashita1,yamashita2}, by using a simple
614: mechanical analog. The analogies of the mixed state of type II
615: superconductors and vortex state of the Josephson junction has been
616: discussed in these references. In Fig 10a we present, as an example,
617: the critical current $I_{max}$ versus the magnetic field for a
618: junction which contains two defects of length $d=2$ placed
619: symmetrically at a distance $D=2$ from the junction's edges. We
620:  examine the following modes grouped according to flux content:
621: 
622: {\it (i) modes $0$, $0a$}
623: 
624:  These modes have magnetic flux antisymmetrical around zero field, as
625:  seen form Fig. 10b where the magnetic flux is plotted versus the
626:  magnetic field. At $I=0$ and magnetic field $H=-0.7$, the $0a$ mode
627:  contains one fluxon trapped in the left defect, while an antifluxon
628:  exists at the other part of the junction. 
629:  As $H$ increases towards $0.7$ the picture changes
630:  slowly, so that the antifluxon is pinned in the right defect. The
631:  stability analysis shows that this mode is unstable. We remark that
632:  there are also other unstable modes near zero flux, which we will
633:  not present here. For example there is another unstable mode with
634:  the same flux as $0a$ but a much higher critical current (the same as
635:  the $0$ mode). Mode $0$ has phase distributions which are similar to
636:  the corresponding mode of the homogeneous junction since it has no
637:  trapped flux in each defect.
638: 
639:  {\it (ii) modes $1l$, $1r$}
640: 
641:  These modes have
642:  magnetic flux close to unity, and are both stable. For the mode $1r$
643:  one fluxon has been trapped to the right defect, and in the mode
644:  $1l$, the vortex is trapped in the left defect. Due to the symmetry
645:  this mode has the same magnetization as the mode $1r$, but the
646:  critical current is reduced. The phase distribution for the modes
647:  $1r$, and $1l$, at zero current are related by
648:  $\phi_{1l}(x)=2\pi-\phi_{1r}(-x)$. The maximum field $H=1.9$ (at $I=0$)
649:  for both modes is determined by an instability at the defect free
650:  side.  At the other extreme
651:  there is a competition at the fluxon side between the applied field
652:  and the field created by the pinned fluxon. Thus the critical field
653:  at $H=-0.62$ can be considered as a coercive field and below this
654:  value the fluxon gets unpinned. The two modes have characteristically
655:  different currents and this depends on the current through the
656:  fluxon free defect, since the pinned fluxon itself gives no major
657:  contribution. Thus the maximum current is much larger for the $1r$
658:  mode. The opposite would be true if we look for negative currents.
659: There are also the symmetrically situated modes that correspond to
660: an antifluxon in the left or right defect, which are not shown in
661:  Fig. 10a. The respective flux is antisymmetric with $H$ around
662:  $H=0$.
663: 
664:  In Fig. 10a we also show the mode $2$ with flux around $2$
665:  fluxons. Several unstable modes are not shown, for the sake of
666:  clarity. Their analysis however, can show the connection between
667:  different modes, while a defect in the correct place with proper
668:  characteristics can stabilize these solutions.
669:  We conclude that
670:  depending on the positions where the vortex is trapped we may have
671:  modes with the same magnetic flux content, but different critical
672:  currents. Also due to soliton localization on the defects, we may
673:  have stable states with magnetic flux close to unity, for zero
674:  magnetic field. These states together with the one existing in the
675:  homogeneous junction form a collection of stable states in a large
676:  $H$ interval. We must comment here that states with unit flux, for
677:  zero magnetic field ($H=0$) exist in the homogeneous junction, as a
678:  continuation of the stable ($1$) mode to negative magnetic fields,
679:  but as we found in a previous work \cite{caputo}, are unstable. So
680:  we may argue here that the presence of defects stabilizes these
681:  states.
682: 
683:  In comparing the results for one (Fig. 2a) and two defects
684:  (Fig. 10a) we see some similarities and differences. In the case of
685:  two defects new modes appear but also the region of stability of the
686:  equivalent modes is different. This is more clearly seen in Fig. 11
687:  where we plot the $I_{max}$ vs $N_f$ for both cases.  This
688:  presentation is useful since the $N_f$ is a nonlinear function of
689:  $H$. This plot (Fig. 11a) is a combination of Figs. 2a and 2b.  We
690:  should point out that the maximum peak in the current in both case
691:  comes due to the trapping of a fluxon in the defect at the right
692:  side. The maximum of $0$-mode is very close in both cases and this
693:  happens because this mode does not involve fluxon trapping. The $1r$
694:  mode for the two defect case is very close to the $1$ mode of the
695:  single defect, since in both cases there is a fluxon trapped in the
696:  same side. In the two defect case we see an enlargement of the region
697:  of stability so that the modes overlap. The thin continuation lines
698: in modes $0$ and $1$ for the single defect are in the reentrant region
699: of flux as discussed in section 2.
700: 
701:  \section{Symmetric distribution of pinning centers}
702:  In this section  we study as an example the case where a junction of
703:  length $\ell=14.2$ contains three defects of length $d=2$, and the
704:  distance between them is $2$. The length was augmented, so that we
705:  keep the same width of the defects when we increase the number of the
706:  defects, since we saw that the width of the order $d =2$, gives the
707:  possibility of fluxon trapping and increased maximum current when the
708:  defect is situated asymmetrically. We will study the phase
709:  distribution at $I=0$ and try to extract information about the
710:  critical field values and magnetization. We find the following modes
711:  grouped according to flux content:
712: 
713:  {\it i) modes $0$, $0l$, $0r$, $0c$}
714: 
715:  In Fig. 12a we
716:  present the critical current versus the magnetic field for the modes
717:  with magnetic flux around  zero (see Fig. 12d). This is indicated by
718:  the $0$ symbol. There are four modes belonging in this category,
719:  which are stable. The solutions for the mode $0$ are similar to the
720:  homogeneous junction mode $0$, with no flux trapping in the defects.
721:  The only difference is that the instability in the critical field
722:  occurs when the phase at one edge, reaches a value, which is smaller
723:  (due to pinning) than the corresponding value for the undefected
724:  junction, which is $\phi(-\ell/2)< \pi$. The same was true for the
725:  two defect case. Mode $0c$ has the maximum critical current
726:  $I_{max}=5.08$ for $H=0$. One antifluxon is trapped to the leftmost
727:  defect, one fluxon to the rightmost, and the phase in the center
728:  defect is constant. The trapping at the edge defects leads to a
729:  positive current distribution between them, for this particular
730:  length, and enlarges the maximum current. The same type of mode was
731:  not found for the two defect case (with a shorter junction length),
732:  and we conclude that the extra defect along with the increased
733:  junction length stabilizes this solution. For the mode $0l$ one
734:  fluxon is trapped in the left defect where the phase changes about
735:  the value $\phi=\pi$. The antifluxon is distributed at the other two
736:  defects, where the phase is about the values $3\pi/2$ (or $\pi/2$),
737:  and we have a cancellation of the positive and negative current
738:  density in this region. Similar for the mode $0r$ the fluxon is
739:  trapped to the right defect, and the current is distributed with
740:  opposite sign to the other two defects.  These modes are similar to
741:  the $0a$ mode for the two defect case.
742: 
743:  {\it ii) modes $1l$, $1c$, $1r$}
744: 
745:  In Fig. 12b we see the maximum
746:  current versus the magnetic field for the modes with magnetic flux
747:  around $N_f=1$ (see Fig. 12e). There are three modes with flux close
748:  to $N_f=1$ each of which corresponds to the trapping of one fluxon in
749:  one defect. In the mode $1c$ the fluxon is trapped in the center
750:  defect. In the mode $1l$ ($1r$) it has been trapped in left (right)
751:  defect. Due to the symmetry the lowest eigenvalue, and the magnetic
752:  flux coincides for these two modes, but as we showed in the previous
753:  section, their critical currents are different, depending on the
754:  tunneling current distribution in the region with no trapping. The
755: $1r$ mode corresponds to a higher critical current.
756: 
757:  {\it iii) modes $2$, $2a$, $2b$}
758: 
759:  In Fig. 12c we see the maximum
760:  current versus the magnetic field for the modes with magnetic flux
761:  around  $N_f=2$ (see Fig. 12f). Only the mode $2$ corresponds to
762:  stable solutions. There we have two fluxons trapped in the side
763:  defects. In mode $2a$ one fluxon is trapped in the right defect,
764:  while in mode $2b$ this trapping occurs in the center defect. We
765:  conclude that distributed pinning centers are more effective in
766:  trapping the vortex, and lead to an increased critical current. Some
767:  conclusions  will continue to be valid for larger number of defects
768:  where we keep the defect width and separation fixed. In that case we
769:  also expect the results to change significantly when there is either
770:  a periodic array of defects, where we expect higher fluxon modes to
771:  give the highest current peak \cite{tinkham}.
772: 
773:  \section{Defect with a smooth variation of current density}
774:  Up to now we considered
775:  defects with abrupt changes in the local critical current density 
776:  and the question
777:  arises whether the abruptness of $J_c$ variation is crucial in the
778:  significant change in $I_{max}$ for the $n=1$ mode. We will see that
779:  similar effects exist for smoother variation, where again the fluxon
780:  pinning is an important feature. For this reason we chose a single
781:  defect at the junction center with a smoothly varying critical current
782:  density
783:  given by \begin{equation}\widetilde J_c (x)=\tanh ^2\left[
784:  \frac{2}{\mu}(x-x_0) \right],~~~\label{delta}\end{equation}
785:  where the defect is centered at $x_0$, and the width is
786:  determined by  $\mu$.
787:  In Fig. 13 we show the results for the case $x_0=7.6$ and $\mu=2$,
788:  which can be compared with the results of the asymmetric defect in
789:  Fig. 2a. For the modes shown the curves are very similar and thus we
790:  see that the main results survive since the defect strengths are
791:  similar. Of course  there is a quantitative difference. But most of
792:  the stability criteria described earlier are still valid.
793: 
794:  In Fig. 14 we consider the effect of the form of current input and
795:  compare the case of inline with overlap for a smooth defect situated
796:  at the center of the junction i.e. $x_0=0$ with $\mu=0.5$. In Fig.
797:  14a we present the $I_{max}$ for inline boundary conditions, and we
798:  show only the $-1,0,1$ modes. The $0$ mode is not influenced at all
799:  from the defect since all the phase variation is at the boundaries.
800:  There is a strong similarity with $I_{max}$ for  the $1$, and $-1$
801:  modes. The reason is that in these cases there is a trapped fluxon
802:  or antifluxon at the center and at zero current and magnetic field
803:  the phase variation dies out at the boundaries. Thus when increasing
804:  the current at $H=0$ towards $I_{max}$ we have the same situation at
805:  the boundaries as for the $0$ mode and  the instability happens at
806:  close $I_{max}$ values.  Of course due to the pinning, the fluxon
807:  content is very different from  the $0$ mode. The $-1$ and $1$ modes
808:  have an enhanced $I_{max}$ and the  small difference in $I_{max}$
809:  from the $0$ mode is attributed to the small influence of the trapped
810:  fluxon to the boundaries. Let us remark that a similar situation was
811:  seen in Fig.  8a for the square well defect, when the defect position
812:  is at the center for $H=0$. By comparing with Fig. 9 the $H_{cl}$
813:  and $H_{cr}$ values we see close agreement with the case of $j_d=0$
814:  in the defect.  These results could change for a smaller length
815:  junction or if we move the defect towards the edges (as seen in Fig.
816:  8a).
817: 
818:  For the same defect we also investigated the effect of
819:  the overlap current input, where the current is distributed along
820:  the whole junction. In Fig. 14(b) we present the maximum current per
821:  unit junction length versus the magnetic field, and it should
822:  be compared to the inline  case in Fig. 14(a). We see a significant change
823:  for the $-1$ and $1$ modes. Of course at $I=0$ both current
824:  inputs give the same solutions, but $I_{max}$ is much smaller for the
825:  overlap boundary conditions. This is from the fact that due to the
826:  applied current the fluxon is pushed against the pinning barrier
827:  until it is overcome at the critical current. In the absence of
828:  applied current the phase at the defect center is $\phi(0)=\pi$,
829:  while the application of the current pushes the fluxon to the edge of
830:  the defect which is taken to be near the point where the curvature of
831:  the defect critical current distribution changes sign. So we can
832:  consider in this case this maximum current as a measure of the
833:  pinning force.
834: 
835:  In Fig. 15a we plot the magnetic flux $N_f$ at zero
836:  current versus the magnetic field $H$ for the  inline  case.
837:  The lowest eigenvalues for the different modes versus the magnetic
838:  field are seen in Fig. 15b. For a homogeneous junction the $0$ mode
839:  is the only stable state available at $H=0$.  However in the problem
840:  we consider here, the mode $1$($-1$) exists and it is stable for
841:  $H=0$ and corresponds to the localization of the soliton
842:  (antisoliton) in the inhomogeneity. For these modes we have pinned
843:  flux at $H=0$, with $\phi(0)=\pi$, and $\frac{d\phi}{dx}=2$. In Fig.
844:  16 we show the evolution of $\phi$ and $\frac{d\phi}{dx}$, for the
845:  mode $1$, as we change the magnetic field at $I=0$. Near $H=-1.9$ the
846:  fluxon content is near zero and for $H<-1.9$ an instability sets in
847:  due to the depinning of the fluxon. This is because the slope at the
848:  pinned fluxon competes with the opposite slope tried to be imposed by
849:  the external negative magnetic field at the boundaries. At the other
850:  end the flux is equal to $2$, and the instability sets in when $\phi$
851:  at the boundaries approaches $\pi$ (or odd multiplies). The range of
852:  $H$ values for the $1$ mode, when the defect is at the center is
853:  significantly broadened and gives a corresponding range for the flux
854:  of two fluxons. Usually each mode has about one extra fluxon and in
855:  particular for the perfect junction it contains only one extra
856:  fluxon. This is because the defect is at the center and far enough
857:  from the edges where the magnetic field is applied and therefore
858:  even for negative fields there is no significant competition, with
859:  the field at the defect center. This is especially true when the
860:  distance of the defect from the edges is grater than
861:  $2\lambda_J$. When however the defect is near the edge the
862:  instability sets in before
863: we cross to negative magnetic fields.
864: 
865: The maximum current $I_{max}$ for the mode $0$ is greater than in the
866: modes $1$,$-1$, but is reduced compared with the $I_{max}$ for the
867: mode $0$ in the homogeneous junction, in zero field. In
868: \cite{filippov} they approximated this reduction in an analytical
869: calculation using a delta function for the defect potential, and they
870: found $\Delta I_{max}=-\mu/2L\approx0.02$. In \cite{filippov} they
871: arrived at the analytical result, by minimizing the fluxon free
872: energy, for the maximum overlap current versus the magnetic field
873: $H$, for these modes, which is a good approximation of the numerical
874: solution we consider here in the limit $L \gg 1$.
875: 
876: 
877: 
878: \section{Conclusions}  In several applications it is
879: desirable to work in an extremum of the current for a region of the
880: magnetic field. This can be achieved by the appropriate distribution
881: of defects so that the negative lobes of the current distribution in
882: the junction due to the fluxons are trapped in the defect with no
883: contribution to the current. Of course if the defect is isolated
884: (far from other defects or the edges) we expect zero contribution to
885: the current. Due to the effect of the applied current and magnetic
886: field at the boundaries in certain cases we can obtain  positive
887: current lobes outside the defect. In several cases in section 3 this
888: was the reason for the  increased current. Because the control of
889: magnetic field is very easy compared to other system parameters (like
890: temperature, disorder, etc.) the measurement of the effect of magnetic
891: field on junction behavior, provides a convenient probe for the
892: junction. The calculation of the $I_{max}$ can characterize the
893: quality of the junction or verify the assumed distribution of defects
894: when they are artificially produced. The spatial variation of the
895: critical current density on low $T_c$-layered junctions, and high
896: $T_c$ grain boundary junctions can be directly imaged with a spatial
897: resolution of $1\mu m$ using low temperature scanning electron
898: microscopy $(LTSEM)$ \cite{huebener,gerdemann}. Information on smaller
899: scale inhomogeneities has to rely on the magnetic field dependence of
900: the maximum tunneling current $I_{max}$.
901: 
902: The purpose of this paper is the consideration of large defects in
903: order to study the interaction between fluxons and defects and give
904: estimates of the coercive field for pinning or depinning of a
905: fluxon from a defect. The region of consideration puts us far from
906: the region of perturbation calculations and is amenable to direct
907: experimental verification since it is easy to design a junction
908: with the above characteristics. The defects influence strongly the low
909: fluxon modes. At high magnetic fields larger than the depinning field
910: of a single fluxon we expect only minor effect and fluxon trapping. Of
911: course for a large number of defects interesting behavior can be
912: obtained.\cite{oboznov,larsen} The interaction between fluxons in the
913: few defect case also assists to overcome coercive fields and untrap
914: fluxons. The results of two trapped fluxons in the two defect case
915: show that the fluxons are strongly coupled and one cannot consider an
916: exponential interaction type potential between the fluxons. Also the
917: critical current in a long junction, cannot be calculated as the
918: Fourier transform of the spatial distribution of the critical current
919: density $J_c(x)$, at least for weak magnetic fields.  For strong
920: magnetic fields, where we have the field penetrating uniformly the
921: junction, as is the case for short junctions, we recover the
922: diffraction like pattern.
923: 
924: In summary we saw that the bounds of the
925: different modes determined by the stability analysis depend on two
926: factors: (i) the instability at the boundaries away from the defect
927: when $\phi_x$ reaches its extremal values equal to $\pm 2$, and (ii)
928: the instability due to the pinning or depinning of a fluxon by the
929: defect. If the junction is near one end then we saw that both criteria
930: play a role in determining the instability, independently in different
931: areas. In general, however, there will be coupling between defects and
932: the edges (surface  defects) especially in the case of multiple
933: defects.  Defects also introduce hysteresis phenomena which are weaker
934: in the case of smooth defects. We also saw that due to fluxon trapping
935: in general we see a reentrant behavior, i.e. there are regions of
936: magnetic field for which there is both an upper and a lower bound on
937: the maximum current.
938: We also find  that
939: due to the pinning of magnetic flux from the defect there exist
940: additional stable states in a large interval of the magnetic
941: field. The abrupt change in the critical current density is not
942: crucial for the trapping. Similar results are expected from smooth
943: defects, with quantitative differences. The above results can be
944: checked experimentally since it is easy to design a junction with a
945: particular defect structure, using masking techniques.
946: In fact a few parameters or characteristics could give at least
947: partial information on defect properties. In particular the measurement
948: of $H_{cr} $ or $H_{cl}$ can give some information of the defects near
949: the edges. Also one can imagine the situation where we scan locally
950: with an electron beam affecting thus the local critical current and
951: observe the variation of the $I_{max} $ as we increase the heating.
952: Once a fluxon is trapped we can decrease the heating (or increase
953: $j_d$) and observe the variation of $I_{max}$. Thus one can have
954: pieces of information to put together in guessing the defect structure
955: that might fit the whole $I_{max}$ pattern. The extension to many
956:   defects requires considerable numerical work. It is hoped, however,
957:   that some of the stability criteria will still be useful.
958: 
959: {\bf Acknowledgements} One of us (N. S.) would like to acknowledge the
960: ESF/FERLIN programme for partial support. Part of this work was done
961: under grant PENED 2028 of the Greek Secretariat for Science
962: and Research.
963: 
964: 
965: \bibliographystyle{prsty}\begin{thebibliography}{99}
966: \bibitem{conf} See several papers in the Conference Proceedings on
967:  {\it Vortex Matter in Superconductors at Extreme Scales and
968:  Conditions}, Eds. V. V. Molshchakov, P. H. Kes, E. H. Brandt in
969:  Physica {\bf C 332}, Nos. 1-4, May (2000).
970: 
971: \bibitem{barone} A. Barone, and G. Paterno, {\it Physics and
972: Applications of the Josephson Effect} (Willey, New York, 1982).
973: \bibitem{dimos} D. Dimos, P. Chaudhari, and J. Mannhart, Phys.
974: Rev. B {\bf 41}, 4038 (1995).
975: \bibitem{sarnelli} E. Sarnelli, P. Chaudhari, and J. Lacey,
976: Appl. Phys. Lett. {\bf 62}, 777 (1993).
977: \bibitem{gross1} R. Gross, P. Chaudhari, M. Kawasaki, M.B.
978: Ketchen, and A. Gupta, Phys. Rev. Lett. {\bf 57}, 727 (1990).
979: \bibitem{gross2} R. Gross, P. Chaudhari, D. Dimos, A. Gupta,
980: and G. Kozen, Phys. Rev. Lett. {\bf 64}, 228 (1990).
981: \bibitem{mccumber} D.E. McCumber, J. Appl. Phys. {\bf 39}, 3113
982:  (1968).
983: \bibitem{miller} D. J. Miller, H. Talvacchio, D. B. Buchholz, R. P. H.
984: Chang, Appl.Phys. Lett. {\bf 66}, 2561 (1995).
985: \bibitem{ayache} J. Ayache, A. Torel, J. Lesueur, U. Dahmen, J.
986: Appl. Phys. {\bf 84} 4921 (1998).
987:  \bibitem{fehren} R. Fehrenbacher, V. B. Geshkenbein and G. Blatter,
988:  Phys. Rev. {\bf B45 },5450 (1992).
989: \bibitem{oka} T. Oka, Y. Itoh, Y. Yanagi, H. Tanaka, S. Takashima, Y.
990: Yamada and U. Mizutani, Physica {\bf C 200}, 55 (1992).
991: \bibitem{takagi} A. Takagi, T. Yamazaki, T. Oka, Y. Yanagi, Y. Itoh,
992:  M. Yoshikawa, Y. Yamada and U. Mizutani, Physica {\bf C 250}, 222
993:  (1995).
994:  \bibitem{wolf} Th. Wolf, A-C. Bornarel, H. Kupfer, R. Meier-Hirmer,
995:   and B. Obst, Phys. Rev. {\bf B 56}, 6308 (1997).
996: 
997:  \bibitem{camerlingo} C. Camerlingo, C. Nappi, M. Russo, G. Testa, E.
998:  Mezetti, R. Gerbaldo, G. Ghigo and L. Gozzelino, Physica {\bf C 332},
999:  93 (2000).
1000:  \bibitem{mezzetti} E. Mezzetti, A. Chiodoni, R. Gerbaldo, G. Ghigo,
1001:  L. Gozzelino, B. Minetti, C. Camerlingo and A. Monaco, Physica {\bf
1002:  332}, 115 (2000).
1003: 
1004:  \bibitem{virocur}
1005:   M. Virokur, and A.E. Koshelev, JETP {\bf 70}, 547 (1990).
1006:  \bibitem{kroger} K. Kroger, L.N. Smith, and D.W. Jllie,
1007:  Appl. Phys. Lett. {\bf 39}, 280 (1981).
1008:   \bibitem{beek} C. J. van der Beek, M. Konczykowski, R. J. Drost, P.
1009:   H. Kes, A. V. Samoilov, N. Chikumoto, S. Bouffard, M. V. Feigel'man,
1010:   Physica {\bf C 332}, 178 (2000).
1011: 
1012:    \bibitem{krasnov} V. M. Krasnov, V. A. Oboznov and N. F. Pedersen,
1013:  Phys. Rev. {\bf B55}, 14486 (1997).
1014:  \bibitem{ketchen} M. B. Ketchen, IEEE trans. Magnetics {\bf 27}, 2916
1015:  (1991).
1016:  \bibitem{likharev} K. K. Likharev and V. K. Semenov, IEEE Trans.
1017:  Appl. Supercond. {\bf 1}, 3 (1991).
1018:  \bibitem{kleinsasser} A. W. Kleinsasser, W. H. Mallison, R. E. Miller
1019:  and G. B. Arnold, IEEE Trans. Appl. Supercond. {\bf 5}, 2735 (1995).
1020:  \bibitem{reinisch} G. Reinisch, J. C. Fernandez, N. Flytzanis, M.
1021:  Taki and S. Pnevmatikos, Phys. Rev. B {\bf 38}, 11284 (1988).
1022:  \bibitem{oboznov} V. A. Oboznov and A. V. Ustinov, Phys. Lett. {\bf
1023:  A139}, 481 (1989).
1024:  \bibitem{larsen} B. H. Larsen, J. Mygind and A. V. Ustinov, Phys.
1025:  Letters {\bf A193}, 359 (1994).
1026:  \bibitem{tinkham} M. A. Itzler and M. Tinkham, Phys. Rev. {\bf B51},
1027:  435 (1995).
1028:  \bibitem{balents} L. Balents and S. H. Simon, Phys. Rev. {\bf B51},
1029:  6515 (1995).
1030:  \bibitem{caputo} J.-G. Caputo, N. Flytzanis, Y. Gaididei, N. Stefanakis,
1031:  and E. Vavalis, Supercond. Sci. Technol. {\bf 13}, 423 (2000).
1032:  \bibitem{owen} C.S. Owen, D.J. Scalapino, Phys. Rev. {\bf 164}, 538 (1967).
1033:  \bibitem{chow} T.C. Chow, H. Chou, H.G. Lai, C.C. Liu, Y.S. Gou,
1034:  Physica C {\bf 245}, 143 (1995).
1035:  \bibitem{yamashita1} T. Yamashita, L. Rinderer, K. Nakajima, and Y.
1036:  Onodera J. Low Temp. Phys. {\bf 17}, 191 (1974).
1037:  \bibitem{yamashita2} T. Yamashita, and L. Rinderer, J. Low Temp. Phys.
1038:  {\bf 21}, 153 (1975).
1039:  \bibitem{filippov} A.T. Filippov, Yu.S. Gal'pern, T.L. Boyadjiev, and
1040:  I.V. Puzynin, Phys. Lett. A {\bf 120}, 47 (1987).
1041:  {\bf 21}, 153 (1975).
1042:  \bibitem{huebener} R. PP. Huebener, Adv. Electron. Elect. Phys. {\bf
1043:  70}, 1 (1988).
1044:  \bibitem{gerdemann} R. Gerdemann, K. D. Husemann, R. Gross, L. Alff,
1045:  A. Beck, B. Elia, W. Reuter and M. Siegel, J. Appl. Phys. {\bf 76},
1046:  8005 (1994).
1047:  \end{thebibliography}
1048: % \end{document}
1049: \newpage
1050: 
1051:  \begin{figure}
1052:  \centerline{\psfig{figure=fig1.eps,width=8.5cm,angle=0}} 
1053:  \caption{The geometry of the junction. The dark shaded
1054:  region marks the defect in the intermediate
1055:  layer. $\ell$ is the junction length and $D$
1056:  the separation between the left edges of the
1057:  defect and the junction. }\label{fig1}\end{figure}
1058: 
1059:  \begin{figure}
1060:  \centerline{\psfig{figure=figIH.eps,width=8.5cm,angle=0}}
1061:  \centerline{\psfig{figure=figfI=0H.eps,width=8.5cm,angle=0}}
1062:  \centerline{\psfig{figure=figeigH.eps,width=8.5cm,angle=0}}
1063:  \caption{(a) Critical current $I_{max}$ and (b)
1064:  magnetic flux at zero current, versus the magnetic field $H$, for the
1065:  different modes.  $\ell=10$ and $D=1.4$.
1066:  (c) The evolution of the
1067:  lowest eigenvalue $\lambda_1$ with the external field for the
1068:  different modes. At the extremes of each mode $\lambda_1$
1069:  vanishes.}\label{fig2}\end{figure}
1070: 
1071:  \begin{figure}
1072:  \centerline{\psfig{figure=figreentr1.eps,width=8.5cm,angle=0}}
1073:  \centerline{\psfig{figure=figreentr0.eps,width=8.5cm,angle=0}}
1074:  \caption{(a) Critical
1075:  values of the bias current as a function of the magnetic field. The
1076:  solid line is drawn for the mode 1 obtained with the usual procedure
1077:  starting form zero current and increasing the current to the critical
1078:  value, while the dashed line represents the values obtained with the
1079:  reentrant procedure described in the text. (b) The same information as
1080:  in (a), but for the mode 0. }\label{fig3}\end{figure}
1081:  
1082:  \begin{figure}
1083:  \centerline{\psfig{figure=figjdmode0.eps,width=8.5cm,angle=0}} 
1084:  \caption{Critical current as a function of the magnetic
1085:  field, for the mode $0$, for different values of the defect critical
1086:  current density $j_d$.}\label{fig4}\end{figure}
1087: 
1088:  \begin{figure}
1089:  \centerline{\psfig{figure=figjdmode1.eps,width=8.5cm,angle=0}} 
1090:  \caption{Same as in Fig. 4, but for the mode $1$.}
1091:  \label{fig5}\end{figure}
1092: 
1093:  \begin{figure}
1094:  \centerline{\psfig{figure=figeigmode1.eps,width=8.5cm,angle=0}}
1095:  \caption{The lowest eigenvalue versus the magnetic
1096:  field for the mode $1$ for different values of the defect critical
1097:  current density $j_d$.}\label{fig6}\end{figure}
1098: 
1099:  \begin{figure}
1100:  \centerline{\psfig{figure=figh=1.5.eps,width=8.5cm,angle=0}}
1101:  \caption{Critical current versus the defect critical
1102:  current density $j_d$, for magnetic field equal to $H=1.5$, for the
1103:  mode $1$. In the same graph the magnetic flux at zero and maximum
1104:  current, and the lowest eigenvalue at $I=0$ are plotted as a function
1105:  of the $j_d$.}\label{fig7} \end{figure}
1106: 
1107:  \begin{figure}
1108:  \centerline{\psfig{figure=figIpos.eps,width=8.5cm,angle=0}}
1109:  \centerline{\psfig{figure=figfImaxpos.eps,width=8.5cm,angle=0}}
1110:  \centerline{\psfig{figure=figeigpos.eps,width=8.5cm,angle=0}}
1111:  \caption{The  variation of (a) the maximum
1112:  current $I_{max}$   and (b) the magnetic
1113:  flux $N_f$ at the maximum current, versus the defect position $D$,
1114:  measured from the right edge of the junction, for the modes $0$, $1$,
1115:  $-1$. The crosses and stars lines are continuations of the two
1116:  points at the two ends of the graph.  (c) the corresponding lowest
1117:  eigenvalue at zero current versus the defect position $D$, for the
1118:  modes $0$, $1$. The  $-1$ mode eigenvalue is the same as for the
1119:  $1$ mode.}\label{fig8} \end{figure}
1120: 
1121:  \begin{figure}
1122:  \centerline{\psfig{figure=figHcrpos.eps,width=8.5cm,angle=0}}
1123:  \centerline{\psfig{figure=figHclpos.eps,width=8.5cm,angle=0}}
1124:  \caption{(a) The
1125:  critical value of instability as we scan the magnetic field to the
1126:  right$H_{cr}$ as a function of the defect position $D$, for the
1127:  modes $0$, $1$, $-1$. (b) The same as (a) but to the left for the
1128:  $H_{cl}$.}\label{fig9}\end{figure}
1129:  
1130:  \begin{figure}
1131:  \centerline{\psfig{figure=figdyoIH.eps,width=8.5cm,angle=0}}
1132:  \centerline{\psfig{figure=figdyofI=0.eps,width=8.5cm,angle=0}}
1133:  \caption{(a) Critical
1134:  current $I_{max}$ and (b) magnetic flux $N_f$, versus the magnetic
1135:  field $H$, for the different modes, for a junction of length
1136:  $\ell=10$, which contains two symmetric pinning centers of length
1137:  $d=2$.}\label{fig10} \end{figure}
1138: 
1139:  \begin{figure}
1140:  \centerline{\psfig{figure=figIflux.eps,width=8.5cm,angle=0}}
1141:  \centerline{\psfig{figure=figdyoIflux.eps,width=8.5cm,angle=0}}
1142:  \caption{Critical
1143:  current $I_c$, versus the magnetic flux $N_f$, at the maximum current
1144:  for the different modes, for a junction of length $\ell=10$, (a) for
1145:  the asymmetric defect case, and (b) for the two symmetric pinning
1146:  centers of length $d=2$.}\label{fig11}\end{figure}
1147:  
1148:  \begin{figure}
1149:  \centerline{\psfig{figure=figtria0IH.eps,width=8.5cm,angle=0}}
1150:  \centerline{\psfig{figure=figtria1IH.eps,width=8.5cm,angle=0}}
1151:  \centerline{\psfig{figure=figtria2IH.eps,width=8.5cm,angle=0}}
1152:  \centerline{\psfig{figure=figtria0fI=0H.eps,width=8.5cm,angle=0}}
1153:  \centerline{\psfig{figure=figtria1fI=0H.eps,width=8.5cm,angle=0}}
1154:  \centerline{\psfig{figure=figtria2fI=0H.eps,width=8.5cm,angle=0}}
1155:  \caption{Critical current $I_{max}$ versus the magnetic
1156:  field $H$, for the different modes (a) $0$, $0l$, $0r$, $0c$, (b)
1157:  $1r$, $1l$, $1c$, and (c) $2$, $2a$, $2b$. for a junction of length
1158:  $\ell=14.2$, which contains three symmetric pinning centers of length
1159:  $d=2$. The corresponding magnetic flux is presented in  (d), (e),
1160:  (f).}\label{fig12}\end{figure}
1161:  
1162:  \begin{figure}
1163:  \centerline{\psfig{figure=figm=2.eps,width=8.5cm,angle=0}} 
1164:  \caption{Inline critical
1165:  current $I_{max}$ versus the magnetic field $H$, for the
1166:  modes $0$, $1$ and $2$,  for the junction with an
1167:  asymmetric defect but smooth variation of the critical current
1168:  density.$x_0=7.6$ and $\mu=2$.}\label{fig13}\end{figure}
1169:  
1170:  \begin{figure}
1171:  \centerline{\psfig{figure=figIHsmov.eps,width=8.5cm,angle=0}}
1172:  \centerline{\psfig{figure=figIHsm.eps,width=8.5cm,angle=0}}
1173:  \caption{Critical current $I_{max}$ versus the magnetic
1174:  field $H$, for the different modes, for (a) inline current and (b)
1175:  overlap current, for the junction with a centered defect, and smooth
1176:  variation of the critical current density.}\label{fig14}\end{figure}
1177:  
1178:  \begin{figure}
1179:  \centerline{\psfig{figure=figfI=0Hsm.eps,width=8.5cm,angle=0}}
1180:  \centerline{\psfig{figure=figeigHsm.eps,width=8.5cm,angle=0}}
1181:  \caption{(a) Magnetic  flux at zero current
1182:  as a function of the external field, for the  same
1183:  type of inhomogeneity as in Fig.  14. (b)
1184:  The  corresponding evolution of the lowest eigenvalue
1185:  $\lambda_1$  for the different modes. At the
1186:  end of each mode the $\lambda_1$ vanishes.}\label{fig15}\end{figure}
1187:  
1188:  \begin{figure}
1189:  \centerline{\psfig{figure=phase.eps,height=17cm,width=8.5cm,angle=0}}
1190:  \caption{The  evolution of $\phi(x)$ and
1191:  $\frac{d\phi(x)}{dx}$, for the mode $1$, as we  change the magnetic
1192:  field at $I=0$, for the same type inhomogeneity  as in Fig. 14.
1193:  Numbers are $H$ values. }\label{fig16}\end{figure} 
1194:  
1195:  \end{document}
1196: 
1197: