1: \documentclass[prb,twocolumn,amssymb,amsmath,aps,superscriptaddress,showpacs]{revtex4}
2: \usepackage{graphicx}
3: \usepackage{hyperref}
4: \usepackage{amsfonts}
5: \usepackage{epsfig}
6: \def\bea{\begin{eqnarray}}
7: \def\eea{\end{eqnarray}}
8: \def\be{\begin{equation}}
9: \def\ee{\end{equation}}
10: \def\nn{\nonumber\\}
11: \def\l{\lambda}
12: \def\r#1{(\ref{#1})}
13: \def\k{\vec{k}}
14: \def\r{\vec{r}}
15: \def\cT{{\cal T}}
16: \begin{document}
17: \title{Vortex Penetration into a Type II Superconductor due to a Mesoscopic
18: External Current}
19: \author{Eran Sela and Ian Affleck \date{\today}}
20: \affiliation{Department of Physics and Astronomy, University of
21: British Columbia, Vancouver, B.C., Canada, V6T 1Z1}
22:
23: \begin{abstract}
24: Applying the London theory we study curved vortices produced by an
25: external current near and parallel to the surface of a type II
26: superconductor. By minimizing the energy functional we find the
27: contour describing the hard core of the flux line, and predict the
28: threshold current for entrance of the first vortex. We assume that
29: the vortex entrance is allowed due to surface defects, despite the
30: Bean-Livingston barrier. Compared to the usual situation with a
31: homogeneous magnetic field, the main effect of the present geometry
32: is that larger magnetic fields can be applied locally before
33: vortices enter the superconducting sample. It is argued that this
34: effect can be further enhanced in anisotropic superconductors.
35: \end{abstract}
36: \pacs{74.25.Ha 74.25.Op 74.25.Qt}
37:
38: \maketitle
39: \section{Introduction}
40: \label{se:int} Surface barrier effects in type II superconductors
41: have been predicted by Bean and Livingston~\cite{Bean64} and de
42: Gennes.\cite{DeGennes66} The entry of flux lines into a planar type
43: II superconductor situated in an external magnetic field $H_{ext}$
44: parallel to its surface is opposed by a strong surface barrier when
45: $H_{ext} = H_{c1}$, the first critical field. Therefore the entry of
46: flux lines could occur at a field value $H_{ext} = H_{S} \sim H_{c2}
47: \gg H_{c1}$, where $ H_{c2}$ is the second critical field. These
48: surface barrier effects have been observed experimentally in the
49: $60$`s on lead thallium alloys~\cite{Joseph64} and on niobium
50: metal,\cite{Blois64} and make it difficult to measure directly
51: thermodynamic properties of the superconductor.
52:
53: Typically surface barriers are reduced due to surface disorder,
54: which creates large local magnetic fields and allows for nucleation
55: of vortices. Suppression of surface barriers for flux penetration
56: was observed on YBaCuO~\cite{Yeshurun91} and in BiSrCaCuO
57: whiskers~\cite{Gregory01} due to heavy ion irradiation. In
58: ellipsoid-shaped YBaCuO it has been argued that due to roughness of
59: submicrometer order the surface barrier does not push the
60: penetration field $H_S$ above $H_{c1}$ but only lowers the rate of
61: vortex entry.\cite{Liang94}
62:
63: Another source for the delay of the entry of flux lines into
64: superconductors is the ``geometrical
65: barrier",\cite{Zeldov94,Brandt99} which is particularly important in
66: thin films of constant thickness (i.e., rectangular cross section).
67: This effect is absent only when the superconductor is of exactly
68: ellipsoidal shape or is tapered like a wedge with a sharp edge where
69: flux penetration is facilitated. The resulting absence of hysteresis
70: in wedge-shaped samples was nicely shown by Morozov et
71: al.~\cite{Morozov97}
72:
73: \begin{figure}[h]
74: \begin{center}
75: \includegraphics*[width=70mm]{figmodel.eps}
76: \caption{Curved flux line near the surface of a superconductor
77: enabled by an external current $I$. \label{fg:1} }
78: \end{center}
79: \end{figure}
80: In this paper we study another source for delay of entrance of flux
81: lines, due to inhomogeneity of the external magnetic field. In
82: particular we consider magnetic field produced by an external
83: current $I$ flowing parallel to the surface of a type II
84: superconductor, see Fig. 1. The magnetic field produced by the
85: external current enters the sample as curved vortices at
86: sufficiently large current. We find that the entrance of the first
87: line occurs when the induced magnetic field at the surface at the
88: position closest to the wire already exceeds the bulk critical field
89: $H_{c1}$. This delay in entrance of the curved vortices occurs due
90: to geometrical reasons: The entry and outlet points are associated
91: with an energy cost $\sim \frac{\phi_0^2}{\mu_0 \lambda}$, where
92: $\lambda$ is the penetration depth, $\phi_0 $ is the flux quantum,
93: and $\mu_0$ is the free permeability. Note that $\phi_0^2/\mu_0 k_B
94: = 0.2464 ~10^6 K - \mu m$, implying that in typical superconductors
95: this is a large energy scale. In addition the spatially averaged
96: magnetic field experienced by the vortex is lower than the the
97: maximal one occurring closest to the wire. Considering those effects
98: in an actual calculation we find how large a magnetic field can be
99: applied locally without introducing vortices into the sample.
100:
101: This implies that application of magnetic field by an external
102: current near the SC can be convenient for experiments demanding
103: sizable magnetic fields in the vortex-free state. As such an
104: experiment we mention the London-Hall effect.\cite{London50} Whereas
105: this effect was observed in regular
106: superconductors,\cite{Bok68,Brown68,Morris71} it is now interesting
107: to measure it in high temperature superconductors. Typically
108: $H_{c1}$ is quite low in these materials and therefore vortices
109: penetrate the sample at very low homogeneous magnetic fields; hence
110: our geometry can be useful. However other surface effects seem to be
111: an additional obstacle for the observation of the London effect in
112: high temperature superconductors.\cite{Lipavsky04}
113:
114: A parameter which we leave out of consideration in this work is
115: anisotropy of the superconductor, which is particularly important in
116: high temperature layered superconductors. In the case of strong
117: anisotropy additional complications enter the problem even in the
118: case of uniform magnetic field, where the direction of the vortices
119: deviates from the direction of the external magnetic
120: field.\cite{Blatter93} For certain (elliptical) treatment of the
121: short distance cutoff the vortices can have two different
122: directions, corresponding to two degenerate minima in the free
123: energy.\cite{Sudbo93}
124:
125: We argue that strong anisotropy is expected to have important
126: effects in our geometry, increasing further the maximal local
127: magnetic field allowed before curved flux lines penetrate the
128: sample. Consider the case where the $\hat{c}$ axis of a uniaxially
129: anisotropic superconductor corresponds to the direction $\hat{x}$ in
130: our geometry, $\hat{c}
131: \parallel \hat{x}$. In this case the surface of
132: the superconductor, parallel to the external wire, corresponds to an
133: ab-plane. In the limit of strong anisotropy $\lambda_{ab}\ll
134: \lambda_c$ the bulk critical field parallel to the surface $H_{c1}
135: \cong \frac{\phi_0}{4 \pi \mu_0 \lambda_{ab} \lambda_c} \log
136: \frac{\lambda_c}{\xi}$ becomes very small. On the other hand, the
137: entry and outlet points of the flux line are associated with a large
138: energy cost $\sim \frac{\phi_0^2}{\mu_0 \lambda_{ab}}$ independent
139: of $\lambda_c$. Therefore we expect the maximal surface magnetic
140: field before the entry of the first vortex to increase relative to
141: $H_{c1}$ as a function of $\lambda_c/\lambda_{ab}$. We leave a
142: detailed treatment of anisotropy in this geometry for a future work.
143:
144: The paper is organized as follows. In Sec.~\ref{se:Formulation} we
145: formulate the problem and obtain expressions for the magnetic field
146: and free energy within London theory. In Sec.~\ref{se:Barrier} we
147: present and discuss the numerical results for the minimization of
148: the free energy as a function of vortex contour. In
149: Sec.~\ref{se:Formulation} and Sec.~\ref{se:BarrierA} we consider the
150: simpler but unrealistic case of a wire with zero width (i.e. $\ll
151: \lambda$). In Sec.~\ref{se:BarrierB} we generalize to wires with
152: finite width. Sec.~\ref{se:Conclusions} contains conclusions. Some
153: details about the derivation of the expression of the free energy
154: are relegated to the appendix.
155:
156: \section{Formulation}
157: \label{se:Formulation} Suppose that a type-II superconductor (SC)
158: occupies the region $x>0$ and magnetic field is induced by an
159: external current $I$ flowing along a wire of zero cross-section at
160: $(x,z)=(-d,0)$, see Fig.~(\ref{fg:1}). Our main object under
161: consideration is a curved flux line lying in the plane $y=0$. Let
162: $\gamma$ denote the closed contour in Fig.~(\ref{fg:axialline})
163: consisting of the axial line of the flux line $\Gamma$ and a line
164: $\Gamma_1$ symmetric to $\Gamma$ with respect to the plane $x=0$,
165: corresponding to an image vortex. Upon further increasing the
166: current a lattice of curved vortices is expected to form along the
167: wire. However here we shall concentrate on small currents and a
168: single flux line.
169: \begin{figure}[h]
170: \begin{center}
171: \includegraphics*[width=45mm]{axialline.eps}
172: \caption{$\Gamma$ is the axial line of the vortex line. The closed
173: contour $\gamma$ is $\Gamma+\Gamma_1$. \label{fg:axialline} }
174: \end{center}
175: \end{figure}
176:
177: In the type II limit, where the coherence length $\xi$ is much
178: shorter than the penetration depth $\lambda$, the total free energy
179: at zero temperature is given by~\cite{DeGennes66}
180: \begin{eqnarray}
181: \label{eq:F} F[\Gamma]&=&\frac{\mu_0}{2} \int_{r>\xi} d^3 r
182: \bigl[\vec{H}^2+\theta(x) \lambda^2 (\vec{\nabla} \times \vec{H})^2
183: \bigr] \nonumber \\
184: &-&\mu_0 \int d^3 r \vec{A} \cdot \vec{j}_{\rm{ext}}.
185: \end{eqnarray}
186: Here $\vec{j}_{\rm{ext}}=-I \delta(x+d) \delta(z) \hat{y}$, $I$ is
187: the applied current through the wire, $\vec{A}$ is the vector
188: potential $\vec{H} = \vec{\nabla} \times \vec{A}$ and $\theta(x)$ is
189: the unit step function. The integral $\int_{r>\xi}$ is carried out
190: in all space outside of the vortex ``hard core" $\Gamma$. We assume
191: that the radius of curvature of $\Gamma$ is larger than $\xi$ at any
192: point in $\Gamma$. Note that at $x=0$ there is an apparent kink in
193: $\gamma$, however this should be though of as a kink only for length
194: scales large compared to $\xi$.
195:
196: The corresponding equations for the magnetic field $\vec{H}$ are the
197: Maxwell equation, $\vec{\nabla} \times \vec{H}=\vec{j}_{\rm{ext}}$
198: for $x<0$, and the London equation, $(1-\lambda^2
199: \vec{\nabla}^2)\vec{H}(\vec{r}) = \frac{\phi_0}{\mu_0} \int_{\Gamma}
200: d\vec{r}' \delta^3(\vec{r}-\vec{r}') $ for $x>0$. For all $x$ we
201: also have $\vec{\nabla} \cdot \vec{H}=0$. In addition we impose
202: appropriate boundary conditions at $x=0$: the magnetic field is
203: continuous, and no supercurrent flows perpendicular to the surface:
204: $\vec{j}_x = (\vec{\nabla} \times \vec{H})_x=0$. To construct a
205: solution we use the functions
206: \begin{eqnarray}
207: \label{eq:functions}
208: \vec{H}^{hom}_{A(\vec{k}_2),B(\vec{k}_2)}(\vec{r})&=&\int \frac{d^2
209: k_2}{(2 \pi)^2}e^{i \vec{k}_2 \cdot \vec{r}}\times \nonumber \\
210: &\times& \left \{
211: \begin{array}{ll}
212: A(\vec{k}_2)[-k_2^2 \hat{x}+i \vec{k}_2 \tau(k_2)]e^{-\tau(k_2) x} & x> 0 \\
213: B(\vec{k}_2)[k_2 \tau(k_2) \hat{x}+i \vec{k}_2 \tau(k_2)]e^{k_2 x}
214: & x< 0
215: \end{array} \right. , \nonumber \\
216: \vec{H}_{\gamma}(\vec{r})&=&\frac{\phi_0}{\mu_0}\int_{\gamma}
217: d\vec{r}' \int \frac{d^3k}{(2 \pi)^3} e^{i \vec{k}\cdot
218: (\vec{r}-\vec{r}')}
219: \frac{1}{1+\lambda^2 k^2}, \nonumber \\
220: \vec{H}_{I',d'}(\vec{r})& =&\frac{I'}{2 \pi} \frac{(-z,0,x +
221: d')}{(x+ d')^2+z^2}.
222: \end{eqnarray}
223: Here $\vec{k}_2=k_y \hat{y}+k_z \hat{z}$, $k_2 =
224: \sqrt{k_y^2+k_z^2}$, and $\tau(k)=\sqrt{k^2+\lambda^{-2}}$. For any
225: $A(\vec{k}_2),B(\vec{k}_2)$, the function $\vec{H}^{hom}$ satisfies
226: the homogeneous equations $\vec{\nabla} \times \vec{H}^{hom}=0$ for
227: $x<0$, and $(1-\lambda^2 \vec{\nabla}^2)\vec{H}^{hom} =0 $ for
228: $x>0$. The function $\vec{H}_{\gamma}$ satisfies the London equation
229: $(1-\lambda^2 \vec{\nabla}^2)\vec{H}_{\gamma}(\vec{r}) =
230: \frac{\phi_0}{\mu_0} \int_{\gamma} d\vec{r}'
231: \delta^3(\vec{r}-\vec{r}') $ in all space. The function
232: $\vec{H}_{I',d'}$ satisfies Maxwell equation $\vec{\nabla} \times
233: \vec{H}_{I',d'}(\vec{r})=\vec{j}'_{\rm{ext}}$ for
234: $\vec{j}'_{\rm{ext}}=I' \delta(x+d') \delta(z) \hat{y}$ for all
235: space.
236:
237: Defining the surface 2-dimensional Fourier transform
238: $\vec{H}_\mu^{surf}(\vec{k}_2) = \int dy dz e^{-i \vec{k}_2 \cdot
239: \vec{r}} \vec{H}_\mu (0,y,z)$, for $\mu = \gamma, \{I',d' \}$, one
240: finds
241: \begin{eqnarray}
242: \vec{H}_\gamma^{surf}(\vec{k}_2) = \frac{\phi_0}{2\mu_0 \lambda^2}
243: \int_\gamma d\vec{r}' e^{-i \vec{k}_2 \cdot \vec{r}'}
244: \frac{e^{-\tau(k_2) |r_x'|}}{\tau(k_2)}, \nonumber \\
245: \vec{H}_{I,d}^{surf}(\vec{k}_2) =\delta(k_y)\pi I e^{-|k_z d|}(i~
246: {\rm{sgn}}~k_z,0,{\rm{sgn}}~d).
247: \end{eqnarray}
248: The solution of the equations satisfying the desired boundary
249: conditions is obtained by adding together the functions in
250: Eq.~(\ref{eq:functions}), and solving for $A(\vec{k}_2)$ and
251: $B(\vec{k}_2)$ to give continuity. It is convenient to include an
252: image current at $x=-d$. The total magnetic field is
253: \begin{eqnarray}
254: \label{eq:H}
255: \vec{H}&=&\vec{H}_0+\vec{H}_v+\vec{H}_s ,\nonumber \\
256: \vec{H}_0
257: &=&\theta(-x)(\vec{H}_{I,d}+\vec{H}_{-I,-d})+\vec{H}_{s0},~~~\vec{H}_{s0}=\vec{H}^{hom}_{A_0,B_0},\nonumber \\
258: \vec{H}_v &=&\theta(x)\vec{H}_{\gamma},~~~~ \vec{H}_s
259: =\vec{H}^{hom}_{A_1,B_1},
260: \end{eqnarray}
261: where
262: \begin{eqnarray}
263: A_0(\vec{k}_2)&=&\frac{2[\vec{H}_{I,d}^{surf}(\vec{k}_2)]_z}{i
264: k_z(\tau(k_2)+k_2)},
265: ~~~B_0(\vec{k}_2)=-A_0(\vec{k}_2)\frac{k_2}{\tau(k_2)},\nonumber \\
266: A_1(\vec{k}_2)&=&B_1(\vec{k}_2)=\frac{[\vec{H}_{\gamma}^{surf}(\vec{k}_2)]_x}{k_2[k_2+\tau(k_2)]}.
267: \end{eqnarray}
268: In the absence of vortices the magnetic field is given by
269: $\vec{H}_0$. It is plotted in Fig.~(\ref{fg:1a}) for $d=5 \lambda$.
270: \begin{figure}[h]
271: \begin{center}
272: \includegraphics*[width=40mm]{HJ.eps}
273: \caption{Field lines of the vortex-free solution $\vec{H}_0(x,z)$
274: for $d=5 \lambda$ [direction of field lines correspond to
275: anticlockwise rotation around $(x,z)=(-d,0)$].\label{fg:1a} }
276: \end{center}
277: \end{figure}
278:
279: The total free energy as function of $\Gamma$ is obtained by
280: substituting the magnetic field Eq.~(\ref{eq:H}) into the free
281: energy Eq.~(\ref{eq:F}). We obtain
282: \begin{equation}
283: \label{eq:Fterms} F =F_0+ F_v + F_{s} + F_{ext}.
284: \end{equation}
285: Here
286: \begin{eqnarray}
287: \label{eq:Ftot} F_0&=&\frac{\mu_0}{2} \int d^3 r
288: \bigl[\vec{H}_{0}^2+\theta(x) \lambda^2(\vec{\nabla} \times
289: \vec{H}_{0})^2 -2\vec{A}_0 \cdot \vec{j}_{\rm{ext}}\bigr], \nonumber \\
290: F_{i}&=&\frac{\mu_0}{2} \int d^3 r \bigl[\vec{H}_{i}^2+\theta(x)
291: \lambda^2 (\vec{\nabla} \times \vec{H}_{i})^2 \bigr],~~~i=v,s,
292: \nonumber
293: \\F_{ext}&=&-\mu_0 \int d^3 r (\vec{A}_v + \vec{A}_{s} )\cdot
294: \vec{j}_{\rm{ext}} .
295: \end{eqnarray}
296: Here $\vec{H}_i=\vec{\nabla} \times \vec{A}_i$, $i=0,v,s$. All mixed
297: terms between $\vec{H}_0,\vec{H}_v,\vec{H}_{s}$ vanish. For the
298: vanishing of mixed terms involving $\vec{H}_v$ see
299: Ref.~\onlinecite{Brandt80}, p.579. We prove the vanishing of the
300: remaining crossed terms between $\vec{H}_0$ and $\vec{H}_s$ in the
301: appendix.
302:
303:
304: The term $F_0$ is the energy of the system without vortices. To
305: evaluate it we introduce a finite wire radius $a \ll \lambda, d$,
306: and assume the external current flows in a thin shell of this
307: radius. Note that $F_0$ scales linearly with the length of the wire,
308: $L_y$. The result of a calculation, using the methods of the
309: appendix, is
310: \begin{eqnarray} \frac{F_0}{L_y}&=&-\frac{\mu_0 I^2}{2 \pi} \left[
311: \frac{1}{2} \log(2d/a)+g(d/\lambda) \right], \nonumber
312: \\
313: g(y)&=&\int_0^\infty dx \frac{e^{-2 x}}{x+\sqrt{x^2+y^2}}.
314: \end{eqnarray}
315: We can infer from it the repulsive force per unit length $\frac{
316: \partial_d F_0}{L_y}<0$ between the wire and the SC. It is plotted in Fig.~(\ref{fg:Force}) (for
317: $a/\lambda = 0.01$). Using $g(y \rightarrow \infty) =\frac{1}{2
318: y}$, $g(y \rightarrow 0)= \frac{1}{2 }\log\frac{1}{2 y}$, we may
319: identify two regimes. (i) $d \gg \lambda$: Here $g \to 0$ and $
320: \frac{
321: \partial_d F_0}{L_y} \rightarrow - \frac{\mu_0 I^2}{2 \pi (2 d)}$. In agrement with Ampere force law, this corresponds to a repulsive
322: force per unit length between two wires $2d $ apart carrying current
323: $I$ with opposite direction. This is the origin of the levitation
324: effect. The second wire corresponds to the term $\vec{H}_{{-I,-d}}$
325: in the solution for $\vec{H}_0$, see Eq.~(\ref{eq:H}); (ii) $d \ll
326: \lambda$: The $1/d$ divergence in the force is cutoff by $\lambda$.
327: The limiting repulsion force per unit length as the wire approaches
328: the surface is $\frac{
329: \partial_d F_0}{L_y} \rightarrow - \frac{\mu_0 I^2}{2 \pi \lambda} c_0$ where $c_0 \sim
330: 0.665$.
331: \begin{figure}[h]
332: \begin{center}
333: \includegraphics*[width=75mm]{Force.eps}
334: \caption{Repulsive force between the superconductor and the wire in
335: the absence of vortices. The dimensionless force $-(2 \pi
336: \lambda/L_y \mu_0 I^2)
337: \partial_d F_0$ behaves as $\lambda/(2d)$ for $d \gg \lambda$, and goes to a constant $c_0 \sim
338: 0.665$ for $d \ll \lambda$.\label{fg:Force} }
339: \end{center}
340: \end{figure}
341:
342: The term $F_{ext}$ accounts for the interaction between the vortex
343: and the external current. Using $\vec{j}_{\rm{ext}}=-I \delta(x+d)
344: \delta(z) \hat{y}$ we have
345: \begin{eqnarray}
346: F_{ext}=\mu_0 I \int_{-\infty}^\infty d y [\vec{A}_v(-d,y,0) +
347: \vec{A}_{s} (-d,y,0)]_y .
348: \end{eqnarray}
349: The contour of integration $(x,y,z)=(0,-\infty,0)$ $\rightarrow
350: (0,\infty,0)$ corresponds to the external current. Physically the
351: wire should be closed into a loop, and we may close the contour of
352: integration e.g. in the $xy$ plane from the $x \rightarrow -\infty$
353: side. Then, using Green's theorem we obtain
354: \begin{eqnarray}
355: F_{ext}=\mu_0 I \int_{-\infty}^{-d} dx \int_{-\infty}^\infty d y
356: [\vec{H}_v(x,y,0) + \vec{H}_{s} (x,y,0)]_z .\nonumber
357: \end{eqnarray}
358: Note that $\vec{H}_v(x,y,0)$ vanishes at $x<0$. Using the formula
359: for $\vec{H}_{s}$, Eq.~(\ref{eq:H}), we obtain
360: \begin{eqnarray}
361: \label{eq:FI} F_{ext}&=&- \frac{I \phi_0}{\pi} \int_{\Gamma}
362: d\vec{r}_z \int_0^\infty dk e^{- k d } \cos(k r_z) \times
363: \nonumber \\
364: &\times& \left( 1-e^{- \tau(k) r_x}\right) \left(
365: 1-\frac{k}{\tau(k)}\right).
366: \end{eqnarray}
367: We used the identity $\oint_\gamma d \vec{r} \cdot \vec{\nabla}
368: \mathcal{F}(\vec{r})=0$ which holds for any continuous function
369: $\mathcal{F}$ if $\gamma$ is a closed contour. In this calculation
370: $\mathcal{F}(\vec{r}) = e^{i k r_z}{\rm{sgn}}(r_x) \left( 1-e^{-
371: \tau(k) |r_x|}\right)$.
372:
373: The terms $F_v$ and $F_s$ have been derived in
374: Refs.~[\onlinecite{Brandt80,Shehata84}],
375: \begin{eqnarray}
376: \label{eq:Fvs} F_v&=&\frac{\phi_0^2}{2 \mu_0} \sum_{i=x,y,z}
377: \int_\gamma d\vec{r}_i \int_\gamma d\vec{r}'_i
378: \frac{\exp(-|\vec{r}-\vec{r}'|/
379: \lambda)}{8 \pi \lambda^2 |\vec{r}-\vec{r}'|} , \nonumber \\
380: F_s&=&\frac{\phi_0^2}{2 \mu_0} \int_\Gamma d\vec{r}_z
381: \int_{\Gamma_1} d\vec{r}'_z V^{(s)}(\vec{r} - \vec{r}').
382: \end{eqnarray}
383: The term $F_v$ is sensitive to the short distance cutoff $\xi$. To
384: account for the cutoff we restrict the contour integration to
385: $|\vec{r} - \vec{r}'|>\xi$. Here the anisotropic kernel for $F_s$ is
386: \begin{eqnarray}
387: V^{(s)} (\vec{r})= \frac{1}{2 \pi \lambda^2 } \int_0^\infty dk
388: \left( 1 - \frac{k}{\tau(k)}\right)e^{- \tau(k) |r_x|} J_0(k
389: |r_z|),\nonumber
390: \end{eqnarray}
391: where $J_0(x)$ is a Bessel function, and this integral can be done
392: and expressed in terms of other Bessel functions.\cite{Shehata84}
393: Note that $V^{(s)}(r_x \to 0,r_z \to 0)= (2 \pi \lambda^3)^{-1}$,
394: hence there is no need to regulate $F_s$ with a cutoff.
395:
396: Different than the usual case with a uniform magnetic field, in our
397: problem the energy $F =F_0+ F_v + F_{s} + F_{ext}$ is a function of
398: the contour $\Gamma$ and is minimized for a particular contour which
399: we need to find. To this end we minimize $F[\Gamma]$ numerically,
400: approximating $\Gamma$ by a polyline having $2M$ equal length sides
401: ($M=8$ in most simulations). We assume that $\Gamma$ has the
402: reflection symmetry $z\rightarrow -z$. This leads to a
403: $M+1$-dimensional parameter space in which we search for the minimum
404: of $F$. For an example see Fig.~(\ref{fg:contour}). In all our
405: calculations $\xi=.001 \lambda$.
406: \begin{figure}[h]
407: \begin{center}
408: \includegraphics*[width=30mm]{contour.eps}
409: \caption{Contours corresponding to local minimum of $F$ for $d=10
410: \lambda$, and for the specified currents. We assume that $\Gamma$
411: has the reflection symmetry $z\rightarrow -z$, and plot $\Gamma$
412: only for $z \ge 0$. \label{fg:contour} }
413: \end{center}
414: \end{figure}
415:
416: \section{Surface barrier}
417: \label{se:Barrier} We find that the free energy contains a surface
418: energy barrier. From this section we shall disregard the vortex
419: independent term of the free energy, $F \to F_v + F_{s} + F_{ext}$.
420: For later comparison we briefly discuss the case with homogeneous
421: magnetic field.\cite{Bean64} Consider a semi-infinite type-II
422: superconductor with a flux thread within it, parallel to the surface
423: and to the external magnetic field $H_{ext}$ ($\parallel \hat{z}$).
424: The line energy $f=F/L_z$ ($L_z$ is the length of the vortex taken
425: to be parallel to $\hat{z}$) as function of the distance from the
426: surface $x_0$, is given by~\cite{Bean64,DeGennes66}
427: \begin{equation}
428: \label{degenes} f(x_0)=\phi_0 \bigl[ H_{ext}
429: e^{-x_0/\lambda}-\frac{1}{2}h(2 x_0)+H_{c1}-H_{ext} \bigr].
430: \end{equation}
431: Here $h(r) = \frac{\phi_0}{2 \pi \mu_0 \lambda^2} K_0\bigl(
432: \frac{r}{\lambda} \bigr)$ is the function giving the field at
433: distance $r$ of a single straight flux line,
434: $H_{c1}=\frac{1}{2}h(\xi) \cong \frac{\phi_0}{4 \pi \mu_0 \lambda^2}
435: \log \frac{\lambda}{\xi}$, and $K_0$ is the zero-order Bessel
436: function. The term $\phi_0 H_{ext} e^{-x_0/\lambda}$ describes the
437: interaction of the line with the external field and associated
438: screening currents. It is a repulsive term. The term $-\phi_0 h(2
439: x_0) / 2$ represents the attraction between the line and its image.
440: When $H_0 \sim H_{c1}$ there is a strong barrier opposing the entry
441: of a line. We can understand this barrier as follows: When
442: $H_{ext}=H_{c1}$, $f(x_0=0)=f(x_0 = \infty)=0$. If we start from
443: $x_0$ large and bring the line closer to the surface, the repulsive
444: term ($\sim e^{-x_0/\lambda}$) dominates the image term ($\sim e^{-2
445: x_0/\lambda}$). Thus $f$ becomes positive and we have a barrier. The
446: barrier disappears, however, in high fields. When $H
447: > H_S = \phi_0/4 \pi \lambda \xi$, the slope $\partial f/\partial x_0|_{x_0 \sim \xi}$ becomes
448: negative. $H_S$ is of the order of the thermodynamic critical field
449: $H_{c2}$. The conclusion is that, at field $H < H_S$, the lines
450: cannot enter in an ideal specimen (although their entry is
451: thermodynamically allowed as soon as $H>H_{c1}$). However this
452: picture is modified in experiment due to surface inhomogeneities
453: producing local large magnetic fields, and allowing vortices to
454: enter the sample above $H_{c1}$.
455:
456: \subsection{Results for wire with zero width}
457: \label{se:BarrierA} We find a similar energy barrier for the
458: entrance or exit of a curved vortex in our geometry with an external
459: current rather than an homogeneous external magnetic field. This
460: barrier can be visualized in the curves in Fig.~(\ref{fg:EL})
461: (except for the diamonds). Note that typically the barrier hight
462: $\Delta$ is of order $\Delta \sim \phi_0^2/\mu_0 \lambda \gg T_c$,
463: where $T_c$ is the critical temperature of the SC. This implies
464: rather small tunneling probabilities $e^{-\Delta/T} \ll 1$ which
465: prevents entry of vortices for clean surfaces. However for strong
466: disorder, vortices can enter more efficiently via nucleation at
467: impurity sites. The contours corresponding to the minimum of the
468: curves with stars and squares are plotted in
469: Fig.~(\ref{fg:contour}). In all our calculations $\xi=.001 \lambda$.
470:
471: \begin{figure}[h]
472: \begin{center}
473: \includegraphics*[width=75mm]{EL.eps}
474: \caption{Evolution of surface barrier as function of external
475: current for $d/\lambda=10$. When $I<I_{c0}$ (diamonds, $I=19
476: \frac{\phi_0}{\mu_0 \lambda}$) the force on the line points always
477: towards the surface. When $I_{c0}<I<I_{c1}$ (stars, $I=22
478: \frac{\phi_0}{\mu_0 \lambda}$) there exists a meta-stable minima
479: with positive energy. When $I_{c1}<I<I_{S}$ (squares and triangles
480: $I=26,80 \times \frac{\phi_0}{\mu_0 \lambda}$) the minimum energy is
481: negative, but a barrier opposes the entry of the flux line. In each
482: point in this plot we have minimized numerically $F$ with respect to
483: the contour $\Gamma$ at fixed $x_0$. \label{fg:EL}}.
484: \end{center}
485: \end{figure}
486: Figure (\ref{fg:EL}) implies the following picture. For
487: infinitesimal current there is no stable vortex configuration. As
488: the current increases we identify three threshold currents
489: $I_{c0}<I_{c1}<I_{S}$: When the current exceeds $I_{c0}$ a
490: meta-stable minima with $F>0$ occurs. When the current exceeds
491: $I_{c1}$, the minimum energy changes sign, $F<0$, but still there is
492: an energy barrier for the entrance of a flux line. When the current
493: exceeds $I_{S}$ the barrier disappears.
494:
495: In Fig.~(\ref{fg:Ic01}) we investigated the dependence of $I_{c0}$
496: and $I_{c1}$ on the distance to the wire $d$. In the limit $d \gg
497: \lambda $ the results for $I_{c1}$ are consistent with the formula $I_{c1} \to \pi d H_{c1}$ [see diagonal dashed line in
498: Fig.~(\ref{fg:Ic01})]. The behavior of $I_{c0}$ in that limit shows
499: that the region of metastability $I_{c0}<I<I_{c1}$ is very narrow.
500: This behavior appears in sharp contrast to the case of uniform
501: magnetic field even in the limit $d \gg
502: \lambda$: We recall that Eq.~(\ref{degenes}) predicts metastable solutions
503: for infinitesimal homogeneous magnetic field $H_{ext}$. These states
504: live far from the surface as $H_{ext}$ becomes smaller. This effect
505: is absent in our geometry both due to the fact that the effective
506: external magnetic field created by the wire decays at long distances
507: from the surface and due to the line energy for penetration a long
508: distance into the SC. In the other extreme limit $d \ll \lambda$ we
509: observed from the numerical solution that the contour $\gamma$ can
510: be approximated by a circle centered at the origin. Making this
511: assumption we can calculate $I^{circle}_{c0}=10^{0.6981}
512: \frac{\phi_0}{\mu_0 \lambda}$ ($x_0 \sim 0.72 \lambda$, $F \mu_0
513: \lambda/\phi_0^2=0.1715$), and $I^{circle}_{c1}=10^{0.749}
514: \frac{\phi_0}{\mu_0 \lambda}$ ($x_0=1.27 \lambda$, $F=0$) in the
515: limit $d \rightarrow 0$. This approximation is in reasonable
516: agreement with the actual solution as the horizontal dashed lines
517: show.
518:
519: \begin{figure}[h]
520: \begin{center}
521: \includegraphics*[width=80mm]{Ic01.eps}
522: \caption{Dependence of threshold currents $I_{c0}$ and $I_{c1}$ on
523: $d/\lambda$. \label{fg:Ic01}}.
524: \end{center}
525: \end{figure}
526:
527: The shape of the contour changes as function of $d$. In
528: Fig.~(\ref{fg:xz}) we plot the extension of the contour in the $x$
529: and $z$ directions. We fitted the numerical results for $x_0$ with
530: an empirical formula $x_0/\lambda = c +\log(d /\lambda)$ with $c
531: \sim 1$, implying that the penetration of the vortex is of order
532: $\lambda$ for all $d$. On the other hand it appears that $z_0$ grows
533: linearly as function of $d$. In the limit $d \to 0$ we have
534: $x_0/\lambda \to 1.26$ and $z_0/\lambda \to 1.43$ (implying that
535: the circular contour is only an approximation).
536:
537: \begin{figure}[h]
538: \begin{center}
539: \includegraphics*[width=90mm]{xz.eps}
540: \caption{Extensions of the curved flux line along $x$ and $z$ as
541: function of $d/\lambda$, at $I = I_{c1}$. \label{fg:xz}}.
542: \end{center}
543: \end{figure}
544:
545:
546: For disordered surfaces, the present geometry can be useful for
547: application of large magnetic fields on a SC sample in a vortex free
548: state. The maximal magnetic field that can be applied in a vortex
549: free state using the wire geometry is $\vec{H}_0[x=z=0]$ [see
550: Eq.~(\ref{eq:H})] at current $I_{c1}$. In the limit $d \gg \lambda$
551: this magnetic field coincides with the bulk first critical field
552: $H_{c1} \approx \frac{\phi_0}{4\pi \mu_0
553: \lambda^2}\log(\lambda/\xi)$, however at smaller $d$ the magnetic
554: field at the surface increases. This is shown in
555: Fig.~(\ref{fg:Hratio}) where we plot the magnetic field
556: $H_{surface}=[\vec{H}_0(0,0,0)]_z = [\vec{H}_{s0}(0^+,0,0)]_z$ given
557: in Eq.~(\ref{eq:H}) at the current $I_{c1}$, which we calculated
558: above as function of $d$. Note that the field enhancement is small
559: for $d>3\lambda$ ($H_{surface} \cong 2 H_{c1}$ for $d=3 \lambda$).
560:
561:
562: \begin{figure}[h]
563: \begin{center}
564: \includegraphics*[width=80mm]{Hratio.eps}
565: \caption{Magnetic field at the surface (see definition in the text)
566: of a clean SC just before the entry of the first vortex for current
567: approaching $I_{c1}$. At $d \gg \lambda$ this magnetic field tends
568: to the bulk critical field $H_{c1}$. As $d$ becomes smaller the SC
569: can sustain larger magnetic fields in the vortex-free (Meissner)
570: state.\label{fg:Hratio}}
571: \end{center}
572: \end{figure}
573:
574: We turn to an estimation of the threshold current $I_S$ at which the
575: barrier disappears. A more precise calculation would involve the
576: Ginzbur-Landau theory. We follow the above analysis of
577: $H_S$.~\cite{DeGennes66} Since the London theory is applicable at
578: distances $\gg \xi$ we estimate $I_S$ using
579: \begin{equation}
580: \label{eq:zeroderivative} \frac{\partial F }{\partial x_0}
581: \bigl|_{x_0 \sim \xi}=0.
582: \end{equation}
583: We find numerically that at $x_0 \ll \lambda$ the closed contour
584: $\gamma$ is well approximated by a circle with radius $x_0$ centered
585: at $x=z=0$. In the limit $x_0 \sim \xi \ll \lambda$ we can evaluate
586: the functional $F$ analytically as function of $x_0$. In
587: Eq.~(\ref{eq:Fvs}) for $F_v$ we can set $\exp(-|\vec{r}-\vec{r}'|/
588: \lambda) \to 1$, hence
589: \begin{eqnarray}
590: F_v(x_0) &\sim& \frac{\phi_0^2 x_0}{32 \pi \mu_0 \lambda^2}
591: \int_0^{2\pi} d \theta_1 \int_0^{2\pi} d \theta_2
592: \frac{\cos(\theta_1 -\theta_2)}{\left| \sin \frac{\theta_1
593: -\theta_2}{2}\right|} \times
594: \nonumber \\
595: &\times& \theta(2 x_0 \left| \sin \frac{\theta_1
596: -\theta_2}{2}\right|-\xi).
597: \end{eqnarray}
598: Compared to $F_v$, the stray term is negligibly small, $F_s \sim
599: \frac{\phi_0^2}{\mu_0 \lambda}\left( \frac{x_0}{\lambda} \right)^2$.
600: The interaction energy with the external current reads
601: \begin{eqnarray}
602: F_{ext}(x_0)&=& - \frac{I \phi_0 x_0^2}{2 \lambda^2}
603: \tilde{f}(d/\lambda),\nonumber \\
604: \tilde{f}(x)&=&\int_0^\infty dy e^{-y x} (\sqrt{y^2+1}-y).
605: \end{eqnarray}
606: Using these formulas for $F_{v}$ and $F_{ext}$ we obtain from
607: Eq.~(\ref{eq:zeroderivative}) the estimate
608: \begin{equation}
609: I_{S} \sim \frac{\phi_0}{8 \mu_0 \xi \tilde{f}(d/\lambda)}.
610: \end{equation}
611: The dependence of $I_S$ on $d/\lambda$ is hidden in the function
612: $\tilde{f}(x)$, with $\tilde{f}(x\to \infty)\to x^{-1}$,
613: $\tilde{f}(x\to 0)\to \log( x^{-1/2})+c$, where $c \sim 0.3$. For
614: $d\gg \lambda$ we have $I_{S} \sim \frac{\phi_0 d}{8 \mu_0 \xi
615: \lambda} $. In this case the magnetic field due to the external
616: current at $x=z=0$ is $[H_0(\vec{r}=0)]_z \to \frac{I_{S}}{\pi d}$.
617: It is of the order of the second critical field $H_{c2}$. In the
618: other limit $d\ll \lambda$ we have $I_{S} \sim \frac{\phi_0 }{4
619: \mu_0 \xi \log[\lambda/d]}$. Note that this behavior holds for $ \xi
620: \ll d \ll \lambda$. In this regime we have $I_S \gg I_{c1} \sim
621: \frac{\phi_0 }{ \mu_0 \lambda }$. In Fig.~(\ref{fg:Ic2}) we plot the
622: phase diagram.
623: \begin{figure}[h]
624: \begin{center}
625: \includegraphics*[width=70mm]{Ic2.eps}
626: \caption{Phase diagram. For $I<I_{c1}$ vortices are
627: thermodynamically unfavorable. For $I>I_{c1}$ curved flux lines
628: become favorable, but an energy barrier opposes their entry until
629: the current exceeds $I_S$. A metastable phase exists in a narrow
630: strip below the boundary $I=I_{c1}$, shown in Fig.~(\ref{fg:Ic01}).
631: \label{fg:Ic2}}.
632: \end{center}
633: \end{figure}
634:
635: \subsection{Finite wire cross section}
636: \label{se:BarrierB} Experimentally the wire carrying the external
637: current has a finite cross section. Therefore it is important to
638: include this effect in our calculations. Consider the rectangular
639: cross section as shown in Fig.~(\ref{fg:rec}), and assume current
640: $I$ flows uniformly in this cross section.
641: \begin{figure}[h]
642: \begin{center}
643: \includegraphics*[width=60mm]{rec.eps}
644: \caption{Rectangular wire cross section. \label{fg:rec}}.
645: \end{center}
646: \end{figure}
647: We can write the external current as
648: \begin{equation}
649: \label{eq:jext} \vec{j}_{\rm{ext}} \to -\frac{I}{x_w z_w}
650: \int_{d-\frac{x_w}{2}}^{d+\frac{x_w}{2}} d \tilde{d}
651: \int_{-\frac{z_w}{2}}^{\frac{z_w}{2}} d \tilde{z}
652: \delta(x-\tilde{d}) \delta(z-\tilde{z}) \hat{y}.
653: \end{equation}
654: The modification to the magnetic field
655: $\vec{H}=\vec{H}_0+\vec{H}_v+\vec{H}_s$ occurs only in the first
656: term,
657: \begin{equation}
658: \vec{H}_0(\vec{r}) \to \frac{1}{x_w z_w}
659: \int_{d-\frac{x_w}{2}}^{d+\frac{x_w}{2}} d \tilde{d}
660: \int_{-\frac{z_w}{2}}^{\frac{z_w}{2}} d \tilde{z}
661: \left[\vec{H}_0(x,y,z-\tilde{z}) \right]_{d \to \tilde{d}},\nonumber
662: \end{equation}
663: where $\vec{H}_0(x,y,z)$ is given in Eq.~(\ref{eq:H}). Next we focus
664: on the modification of the vortex dependent part of the free energy
665: $F = F_v + F_{s} + F_{ext}$. Only the term $F_{ext}$ is modified.
666: Using Eq.~(\ref{eq:jext}) it is easy to find that
667: \begin{eqnarray}
668: \label{eq:Fext} F_{ext} &\to& - \frac{I \phi_0}{\pi} \int_{\Gamma} d
669: \vec{r}_z \int_0^\infty dk e^{- k d } \cos(k r_z) (1-e^{- \tau(k)
670: r_x})\times \nonumber \\
671: &\times&\left( 1-\frac{k}{\tau(k)} \right) \left( \frac{\sinh\frac{k
672: x_w}{2}}{\frac{k x_w}{2}} \frac{\sin\frac{k z_w}{2}}{\frac{k
673: z_w}{2}} \right).
674: \end{eqnarray}
675: Let us first specialize to the case of square cross section where
676: the wire touches the SC, $x_w=z_w=2d$, and compare this with a point
677: like cross section $x_w=z_w \to 0$ which we considered until now (We
678: ignore any electron or Cooper pair tunneling between the SC and
679: wire). First we repeated the calculation of $I_{c0}$ and $I_{c1}$.
680: The results are roughly the same for both cross sections for $d
681: \lesssim \lambda$, and deviations up to $10 \%$ are obtained for $d
682: \gg \lambda$ up to $d=100 \lambda$. In Fig.~(\ref{fg:xzrec}) we
683: compare the contours at $I_{c1}$ as function of $d$ for the two
684: cross sections. We can see that $z_0$ changes by a factor of up to
685: $1.6$ for $d \le 90 \lambda$.
686: \begin{figure}[h]
687: \begin{center}
688: \includegraphics*[width=90mm]{xzrec.eps}
689: \caption{Dependance on $d$ of the lengths $x_0$ and $z_0$,
690: characterizing the vortex contour, for zero versus finite wire cross
691: section [$x_x=z_w=0$ (stars) and $x_x=z_w=2d$ (squares)].
692: \label{fg:xzrec}}.
693: \end{center}
694: \end{figure}
695: The magnetic field at the surface just below $I_{c1}$ is compared
696: for the two cross sections in Fig.~(\ref{fg:HratioREC}). At $d\gg
697: \lambda$ it approaches $H_{c1}$ in both cases, while for $d \ll
698: \lambda$ it is larger for point like cross section by about $10 \%$.
699: \begin{figure}[h]
700: \begin{center}
701: \includegraphics*[width=70mm]{HratioREC.eps}
702: \caption{Magnetic field at the surface (see definition in the text)
703: just before the entry of the first vortex at $I \to I_{c1}$, for
704: either zero wire cross section ($x_x=z_w=0$, stars) or finite cross
705: section ($x_x=z_w=2d$, squares). \label{fg:HratioREC}}
706: \end{center}
707: \end{figure}
708:
709: We estimate the typical values of the threshold current $I_{c1}$.
710: For the regime of interest $d \sim \lambda$, we have $I_{c1}$ of the
711: order of $\frac{\phi_0}{\mu_0 \lambda} = \frac{1.6455
712: mA}{\lambda[\mu m]}$. For $\lambda = 1 \mu m$ this corresponds to
713: current density of $\sim 1 mA ~-~ \mu m^{-2}$.
714:
715: Next we consider the dependence on $z_w$ for $z_w \ge x_w = 2 d$
716: which can be experimentally relevant. The limit $z_w \to \infty$ can
717: be treated analytically since the external field $\vec{H}_0$ is
718: uniform at all $x>(-d+\frac{x_w}{2})$. In this limit the maximal
719: magnetic field at the surface before vortices penetrate is
720: $H_{surface} (I_{c1}) \to H_{c1}$. For finite $z_w$ we calculated
721: $H_{surface} (I_{c1})$ numerically with the result plotted in
722: Fig.~(\ref{fg:Hratiozw}). Accordingly $z_w$ should not be too large
723: in order to obtain the effect discussed here including the
724: enhancement of the surface field in the vortex free state for a
725: disordered surface.
726: \begin{figure}[h]
727: \begin{center}
728: \includegraphics*[width=60mm]{Hratiozw.eps}
729: \caption{Magnetic field at the surface at current approaching
730: $I_{c1}$ as function of $z_w$, for $x_w=2d$, $d = \lambda/2$.
731: \label{fg:Hratiozw}}
732: \end{center}
733: \end{figure}
734: \section{Conclusions}
735: \label{se:Conclusions} In this work we studied solutions of London
736: theory in a geometry where an external mesoscopic current flows
737: parallel to a surface of a SC. Only above a threshold current
738: $I_{c0}$ there exist solutions with curved flux lines entering and
739: leaving the SC at the surface. At a larger threshold current,
740: $I_{c1}$, these solutions become energetically favorable, however an
741: energy barrier separates them from a vortex free solution. At a
742: third threshold current, $I_{S}$, this barrier disappears. To
743: determine the current at which vortices actually penetrate the
744: sample one has to account for the degree of disordered of the
745: surface. For strong surface disorder the vortex can penetrate at
746: $I=I_{c1}$ despite the presence of the barrier, due to large local
747: magnetic fields produced at impurity sites allowing for nucleation
748: of vortices. On the other hand for clean surface the entrance of
749: vortices occurs at $I=I_S$.
750:
751: By calculating those currents using a numerical solution of the
752: problem we conclude that for strong surface disorder the present
753: geometry allows to achieve locally larger magnetic fields in the
754: vortex-free state, as compared to the case of homogeneous magnetic
755: field, provided that the wire thickness is of O($\lambda$). This can
756: be potentially relevant for experiments in high-temperature
757: superconductors which typically have extremely low values of
758: $H_{c1}$. We argued that the effect of enhancement of the magnetic
759: field in the vortex free (Meissner) state becomes more pronounced in
760: strongly anisotropic superconductors, which is particularly relevant
761: for layered high-temperature superconductors.
762:
763: We would like to thank Jordan Baglo, Walter Hardy and Cedric Lin for
764: stimulating discussions. This work was supported by NSERC (ES $\&$
765: IA) and CIfAR (IA).
766:
767: \appendix
768: \section{Mixed terms in free energy}
769: \label{se:F0} We shall prove the vanishing of crossed term in the
770: energy Eq.~(\ref{eq:F}) between $\vec{H}_0$ and $\vec{H}_{s}$ [see
771: Eq.~(\ref{eq:H})],\begin{widetext}
772: \begin{eqnarray}
773: \label{eq:mixed} F_{(H_0 H_s)} = \mu_0 \int d^3 r \bigl( \vec{H}_0
774: \cdot \vec{H}_s +\theta(x)\lambda^2 (\vec{\nabla} \times \vec{H}_0)
775: \cdot (\vec{\nabla} \times H_s) \bigr)=0.
776: \end{eqnarray}
777: From Eq.~(\ref{eq:H}) we have $\vec{H}_0=\vec{H}_0'+\vec{H}_{s0}$
778: where $\vec{H}_0'= \theta(-x)(\vec{H}_{I,d}+\vec{H}_{-I,-d}) =
779: \vec{\nabla} \times \vec{A}_1$ and
780: \begin{equation}
781: \vec{A}_1 =\frac{I \hat{y} }{4 \pi } \log
782: \frac{(x+d)^2+z^2}{(x-d)^2+z^2} , ~~x<0.
783: \end{equation}
784: Correspondingly, we have $F_{(H_0 H_s)} =F_{(H_0' H_s)} +F_{(H_{s0}
785: H_s)} $. Consider the term $F_{(H_0' H_s)}=\mu_0 \int_{x<0} d^3 r
786: \vec{H}_0' \cdot \vec{H}_s$. We will use the vector identity
787: $(\vec{\nabla} \times \vec{A}) \cdot \vec{B} = \vec{A} \cdot
788: (\vec{\nabla} \times \vec{B})+\vec{\nabla} \cdot (\vec{A} \times
789: \vec{B})$, with $\vec{A}=\vec{A}_1$, $\vec{B}=\vec{H}_s$, and the
790: fact that $\vec{\nabla} \times \vec{H}_s=0$. Then the volume
791: integral can be transformed to an integral on the surface $x=0^-$.
792: However this integral vanishes because $\vec{A}_1 (0^-,y,z)=0$,
793: hence $F_{(H_0' H_s)} =0$. Now let us consider the term $F_{(H_{s0}
794: H_s)}$ and define $\vec{H}_s = \vec{\nabla} \times \vec{A}_s$. For
795: the integral in the region $x<0$ we use the above vector identity
796: with $\vec{A}=\vec{A}_s$, $\vec{B}=\vec{H}_{s0}$, and for $x>0$ we
797: use the vector identity with $\vec{A}=\vec{H}_{s0}$,
798: $\vec{B}=\vec{\nabla} \times \vec{H}_{s}$. Taking into account that
799: $\vec{H}_s$ and $\vec{H}_{s0}$ satisfy the homogeneous equations we
800: obtain
801: \begin{eqnarray}
802: \label{eq:xxx} \int_{x<0} d^3 r (\vec{\nabla} \times \vec{A}_s)
803: \cdot \vec{H}_{s0}=\int dS \bigl(\vec{A}^-_s \times \vec{H}^-_{s0}
804: \bigr)_x, \nonumber \\ \int_{x>0} d^3 r \bigr( \vec{H}_{s0} \cdot
805: \vec{H}_s+\lambda^2 (\vec{\nabla} \times \vec{H}_{s0}) \cdot
806: (\vec{\nabla} \times H_s) \bigl)=-\lambda^2\int dS
807: \bigl(\vec{H}^+_{s0} \times (\vec{\nabla} \times \vec{H}^+_{s})
808: \bigr)_x.
809: \end{eqnarray}
810: Here $\int dS = \int_{-\infty}^\infty dy \int_{-\infty}^\infty dz$,
811: and $\vec{H}^\pm = \vec{H}(x=0^\pm,y,z)$. For $x>0$ we can use
812: $\vec{A}_s = -\lambda^2 \vec{\nabla} \times \vec{H}_s$, which
813: follows from London equation for $\vec{H}_s$. Next we use the fact
814: that, by construction, $\vec{H}_s^- = \vec{H}_s^+ + \vec{H}_v^+$.
815: This allows us to express $\vec{A}_s^-=-\lambda^2 \vec{\nabla}
816: \times (\vec{H}_v^++ \vec{H}_s^+)$, and combine the two terms of
817: Eq.~(\ref{eq:xxx}) as
818: \begin{equation}
819: \label{eq:simple?} F_{(H_{s0} H_s)} = -\mu_0 \lambda^2 \int dS
820: \left[ (\vec{\nabla} \times \vec{H}^+_{s}) \times (\vec{H}_{s0}^-
821: -\vec{H}_{s0}^+ )+(\vec{\nabla} \times \vec{H}_{v})^+ \times
822: \vec{H}_{s0}^- \right]_x.
823: \end{equation}
824: Now we use explicit forms of these factors: $(\vec{H}_{s0}^-
825: -\vec{H}_{s0}^+ ) =-\frac{I d \hat{z}}{\pi (d^2+z^2)}$,
826: $(\vec{H}_{s0}^-)_y=0$,
827: \begin{eqnarray}
828: (\vec{\nabla} \times \vec{H}^+_{s})_y&=&-\frac{\phi_0}{2 \mu_0
829: \lambda^2} \int_\gamma d\vec{r}`_x \int \frac{d ^2 k_2}{(2 \pi)^2}
830: e^{- i \vec{k}_2 \cdot \vec{r}`+i(k_y y+k_z z)-\tau(k_2)
831: |\vec{r}`_x|} \frac{-i k_z(\tau(k_2) - k_2)}{k_2 \tau(k_2)}
832: ,\nonumber \\
833: (\vec{\nabla} \times \vec{H}^+_{v})_y&=&\frac{\phi_0}{2 \mu_0
834: \lambda^2} \int_\gamma d\vec{r}`_x \int \frac{d ^2 k_2}{(2 \pi)^2}
835: e^{- i \vec{k}_2 \cdot \vec{r}`+i(k_y y+k_z z)-\tau(k_2)
836: |\vec{r}`_x|} \frac{i k_z}{\tau(k_2)}
837: \left(1-\frac{\tau^2(k_2)}{k_z^2}
838: \right),\nonumber \\
839: (\vec{H}_{s0}^-)_z&=&-I \int \frac{d ^2 k_2}{(2 \pi)^2} e^{ i
840: \vec{k}_2 \cdot \vec{r}-|k_z|d} \frac{k_2 2 \pi \delta(k_y)
841: }{\tau(k_2)+k_2}.
842: \end{eqnarray}
843: Plugging these expressions in Eq.~(\ref{eq:simple?}), one can
844: readily obtain $F_{(H_{s0} H_s)}=0$ (without performing any
845: integration), completing the proof for $F_{(H_0 H_s)}
846: =0$.\end{widetext}
847:
848: \begin{thebibliography}{99}
849: \bibitem{Bean64} C.P.~Bean and J.~D.~Livingston, Phys. Rev. Lett.~\textbf{12}, 14 (1964).
850: \bibitem{DeGennes66}P. G. De Gennes, Superconductivity of Metals and Alloys (Benjamin; 1966).
851: \bibitem{Joseph64} A.~S.~Joseph and W.~J.~Tomasch, Phys. Rev. Lett.~\textbf{12}, 219 (1964).
852: \bibitem{Blois64} R.~W.~De~Blois and W.~De~Sorbo, Phys. Rev. Lett.~\textbf{12}, 499 (1964).
853: \bibitem{Brandt80} E.~H.~Brandt, J. Low Temp. Phys. \textbf{ 42}, 557 (1981).
854: \bibitem{Sudbo91} A.~Sudb{\o} and E.~H.~Brandt, Phys. Rev. B \textbf{43}, 10482 (1991).
855: \bibitem{Yeshurun91} M.~Konczykowski, L.~I.~Burlachkov, Y.~Yeshurun, and F.~Holtzberg, Phys. Rev. B \textbf{43}, 13707 (1991).
856: \bibitem{Gregory01} J. K. Gregory \emph{et. al.}, Phys. Rev. B \textbf{64}, 134517 (2001).
857: \bibitem{Iniotakis08} C.~Iniotakis, T.~Dahm, and N.~Schopohl, Phys. Rev. Lett.~\textbf{100}, 037002 (2008).
858: \bibitem{Zeldov94} E.~Zeldov \emph{et. al.}, Phys. Rev. Lett.~\textbf{73}, 1428 (1994).
859: \bibitem{Brandt99} E.~H.~Brandt, Phys. Rev. B \textbf{60}, 11939 (1999).
860: \bibitem{Morozov97} N.~Morozov \emph{et. al.}, Physica C \textbf{291}, 113 (1997).
861: \bibitem{Altshuler95} E.~Altshuler and R.~Mulet, Journal of Superconductivity \textbf{8}, 779 (1995).
862: \bibitem{Shehata84} L.~N.~Shehata and A.~G.~Saif, J. Low Temp. Phys. \textbf{56}, 113 (1984).
863: \bibitem{Liang94} R.~Liang \emph{et. al.}, Phys. Rev. B \textbf{50}, 4212 (1994).
864: \bibitem{London50} F.~London, \emph{Superfluids} (Wiley, New York,
865: 1950), Vol. I, Sec. VIII.
866: \bibitem{Bok68} J.~Bok and J.~Klein, Phys. Rev. Lett.~\textbf{20}, 660 (1968).
867: \bibitem{Brown68} J.~B.~Brown and T.~D.~Morris, \emph{Proc. 11th Int. Conf. Low. Temp.
868: Phys}., Vol. 2, 768 (St. Andrews, 1968).
869: \bibitem{Morris71} T.~D.~Morris and J.~B.~Brown, Physica (Amsterdam) \textbf{55}, 760
870: (1971).
871: \bibitem{Lipavsky04} P. Lipavsk\'{y} \emph{et. al.}, Phys. Rev. B \textbf{70}, 104518 (2004).
872: \bibitem{Blatter93} G.~Blatter and V.~Geshkenbein, Phys. Rev. B \textbf{47}, 2725 (1993),
873: \emph{ibid}. E.~H.~Brandt \textbf{48}, 6699 (1993).
874: \bibitem{Sudbo93} A.~Sudb{\o}, E.~H.~Brandt and D.~A.~Huse, Phys. Rev. Lett. \textbf{71}, 1451 (1993).
875: \end{thebibliography}
876:
877: \end{document}
878: