1:
2: %\documentstyle{article}
3: %\documentstyle[aps,prl,array,multicol,psfig]{revtex4}
4: %\documentstyle[preprint,aps,psfig]{revtex}
5: \documentclass[aps,prl,twocolumn,superscriptaddress,floatfix,showpacs]{revtex4}
6: %\usepackage{graphicx}
7: %\bibstyle{apsrev.bib}
8: %\documentclass[preprint,showpacs,aps,prb,psfig]{revtex4}
9: \usepackage{amssymb}
10: \usepackage{graphicx}
11: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
12: \begin{document}
13: \title{Superconducting Transition Temperature in Heterogeneous
14: Ferromagnet-Superconductor Systems}
15: \author{Valery L. Pokrovsky}
16: \affiliation{Department of Physics, Texas A\&M University, College Station,
17: TX 77843-4242}
18: \affiliation{ Landau Institute for Theoretical Physics,
19: Chernogolovka, Moscow Distr. 142432, Russia}
20: \author{Hongduo Wei}
21: \affiliation{Department of Physics, Texas A\&M University, College Station,
22: TX 77843-4242}
23: \begin{abstract}
24: We study the shift of the the superconducting transition
25: temperature $T_c$ in ferromagnetic-superconducting bi-layers and
26: in a superconducting film supplied a square array of ferromagnetic
27: dots. We find that the transition temperature in these two cases
28: change presumably in opposite direction and that its change is not
29: too small. We extend these results to multilayer structures. We
30: predict that rather small external magnetic field $\sim 10$ Oe can
31: change the transition temperature of the bilayer by $10$\% .
32: \end{abstract}
33: \pacs{74.60.Ge, 05.40.-a, 74.62.Dh}
34: \maketitle
35:
36:
37:
38: \section{1. Introduction}
39:
40: Heterogeneous ferromagnetic-superconducting (FM-SC) systems have
41: attracted much attention recently \cite{yot}. In majority of these
42: systems the proximity effect is suppressed by the oxide layer
43: between FM and SC. Inhomogeneous magnetization produces a magnetic
44: field penetrating into the superconductor and inducing
45: supercurrents. Supercurrents in turn produce the magnetic field
46: acting on the magnetization. Thus, the FM and SC of systems
47: strongly interact via magnetic field. Systems in which both, FM
48: and SC parts are thin films represent a special interest for the
49: experiment and can be analyzed theoretically. In these systems,
50: spontaneous vortices appear due to the magnetic interaction
51: \cite{igor}. Erdin et al. \cite {Erdin1} have developed a method
52: to calculate the arrangement of the magnetization in the FM film
53: and supercurrents including vortices in the SC film in the
54: London's approximation. The London's approximation is quite good
55: for these mesoscopic systems because characteristic length scales
56: for magnetic field (the effective penetration depth, the linear
57: size of the textures) are much larger than the coherence length
58: $\xi$ of the superconductor. This method was applied recently
59: \cite{Erdin2} to study topological textures in a FM-SC bilayer
60: (FSB). It was shown that the homogeneous state of the FSB with the
61: magnetization perpendicular to the layer is unstable with respect
62: to the formation of vortices. The ground state of the FSB
63: represents a periodic array of stripe domains in which the
64: direction of the magnetization in the FM film and the vorticity in
65: the SC film alternate together.
66:
67: In this article we study how the combined FM-SC textures change
68: the superconducting transition temperature in bilayers and
69: multilayers. For this purpose we extend theory of spontaneous
70: SC-FM structures developed in \cite {Erdin2} to the case of
71: multilayers. We find that spontaneous domain-vortex structures
72: increase the transition temperature, whereas vortex structures
73: induced by a periodic array of magnetic dots decreases the
74: transition temperature. The magnitude of this effect is large
75: enough to enable its experimental observation. Though the
76: influence of the textures on the transition temperature is akin to
77: the influence of the homogeneous magnetic field, there are
78: important differences between these two phenomena: first, the
79: average magnetic field may be zero for magnetic textures; second,
80: the reciprocal action of the magnetic field generated by vortices
81: onto magnetization is substantial.
82:
83: The plan of this article is as follows. In the next section we consider the
84: change of the transition temperature in the spontaneous stripe structure
85: formed in the FSB. In Section 3 we analyze how this stripe structure and the
86: transition temperature change in the presence of an external magnetic field.
87: In Section 4 we study the shift of the transition temperature in a square
88: array of magnetic dots. Section 5 is devoted to theory of the FM-SC
89: spontaneous textures in multilayers and to the transition temperature shift
90: in them. Our conclusion are given in Section 6.
91:
92: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
93: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
94: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
95:
96: \section{2. Transition Temperature in spontaneous Stripe Structure of FSB}
97:
98: As it was mentioned earlier, the homogeneous state of the FSB with
99: the magnetization perpendicular to the layer is unstable with
100: respect to formation of a stripe domain structure, in which both,
101: the the direction of the magnetization in the FM film and the
102: circulation of the vortices in the SC film alternate together. Let
103: the stripe width be $L_{s}$. The magnetization can be written as
104: $\mathbf{m}=ms(x)\hat{z}$, where the coordinate $x$ is along the
105: direction perpendicular to the domain walls, $\hat{z}$ denotes the
106: unit vector perpendicular to the layers, $s(x)$ is the periodic
107: step function with period $2L_{s}$:
108: \[
109: s(x)=\left\{
110: \begin{array}{ll}
111: +1 & \mbox{ $0<x<L_s$,} \\
112: -1 & \mbox{ $L_s<x<2 L_s$.}
113: \end{array}
114: \right.
115: \]
116: The energy of the stripe structure per unit area $U$ and the stripe
117: equilibrium width $L_{s}$ were calculated in \cite{Erdin2}. Here we correct
118: a calculational mistake of this work \cite{error}:
119: \begin{equation}
120: U=\frac{-16{\tilde{m}}^{2}}{\lambda _{e}}\exp (\frac{-\epsilon _{dw}}{4{%
121: \tilde{m}}^{2}}+C-1)\,,
122: \label{U}
123: \end{equation}
124: \begin{equation}
125: L_{s}=\frac{\lambda _{e}}{4}\exp (\frac{\epsilon _{dw}}{4{\tilde{m}}^{2}}%
126: -C+1)\,. \label{Ls} \\
127: \end{equation}
128: The notations in eqs. (\ref{U}), (\ref{Ls}) are as follows: $\lambda _{e}=%
129: \frac{\lambda _{L}^{2}}{d_{s}}$ is the effective penetration depth
130: in the SC film, whose thickness is denoted $d_{s}$; $\epsilon
131: _{dw}$ is the linear tension of the domain wall;
132: $\tilde{m}=m-\epsilon _{v}/\phi _{0}$; $m$ is the magnetization
133: per unit area of the FM film; $\epsilon _{v}=\frac{\phi
134: _{0}^{2}}{16\pi ^{2}\lambda _{e}}\ln {\frac{\lambda _{e}}{\xi }}$
135: is the single vortex energy in the absence of the FM film;
136: $C=0.57721\cdots$ is the Euler constant. To find the transition
137: temperature, we combine the energy given by eq. (\ref{U}) with the
138: Ginzburg-Landau free energy. The total free energy per unit area
139: reads:
140: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
141: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
142: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
143: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
144: \begin{eqnarray}
145: F =U+F_{GL}
146: =\frac{-16{\tilde{m}}^{2}}{\lambda _{e}}\exp (\frac{-\epsilon _{dw}}{4{%
147: \tilde{m}}^{2}}+C-1)\nonumber \\
148: +n_{s}d_{s}[\alpha (T-T_{c})+\frac{\beta }{2}n_{s}]\,.
149: \label{U-tot}
150: \end{eqnarray}
151: Here $\alpha $ and $\beta $ are the Ginzburg-Landau parameters. We
152: omit the gradient term in the Ginzburg-Landau equation since the
153: gradients of the phase are included in the energy (\ref{U}),
154: whereas the gradients of the superconducting electrons density can
155: be neglected everywhere beyond the vortex cores. Recalling that
156: $\lambda _{L}^{2}=\frac{m_{s}c^{2}}{4\pi n_{s}e^{2}}$ and plugging
157: it into eq. (\ref{U}), we find:
158: \begin{eqnarray}
159: F &=&\frac{-256{\pi }^{2}\chi _{0}n_{s}}{\phi _{0}}(m+n_{s}\chi _{0}\ln
160: \frac{16{\pi }^{2}\xi n_{s}\chi _{0}}{\phi _{0}})^{2} \nonumber \\
161: &&\cdot \exp [\frac{-\epsilon _{dw}}{4(m+n_{s}\chi _{0}\ln {\frac{16{\pi }%
162: ^{2}\xi n_{s}\chi _{0}}{\phi _{0}}})^{2}}+C-1] \nonumber \\
163: &&+n_{s}d_{s}[\alpha (T-T_{c})+\frac{\beta d_{s}}{2}n_{s}]\,. \label{free}
164: \end{eqnarray}
165: where $\chi _{0}=\frac{\phi _{0}d_{s}e^{2}}{4\pi m_{s}c^{2}}$.
166: Minimizing the total free energy over $n_{s}$, and using the
167: condition that $n_{s}=0$ at new transition temperature $T_{c}^{*}$
168: we obtain:
169: \begin{equation}
170: \Delta T_{c}\equiv T_{c}^{*}-T_{c}=\frac{64\pi m^{2}e^{2}}{\alpha m_{s}c^{2}}%
171: \exp (\frac{-\epsilon _{dw}}{4m^{2}}+C-1)\,. \label{shift}
172: \end{equation}
173: Eq. (\ref{shift}) demonstrates that the interaction between FM and SC layers
174: in the spontaneous stripe structure increases the transition temperature $%
175: T_{c}$. Theory \cite{Erdin2} assumes that the ratio
176: $\epsilon_{dw}/m^{2}$ is larger than $1$, so that $\exp (\frac{\epsilon _{dw}}{4\tilde{m}%
177: ^{2}})\gg 1$. However, this ratio can not be too large. Otherwise the width
178: of domain becomes larger than the sample linear size. The maximum possible
179: shift of transition temperature corresponds to $\frac{\epsilon _{dw}}{4%
180: \tilde{m}^{2}}\sim 1$. It is
181: \begin{equation}
182: \Delta T_{c}\sim \frac{64\pi m^{2}e^{2}}{\alpha m_{s}c^{2}}\,,
183: \end{equation}
184: The value $\alpha $ can be estimated as $\alpha \sim
185: T_{c}/\epsilon _{F}$, where $\epsilon_F$ is the Fermi energy.
186: It is about $10^{-4}$ for low-temperature superconductors and about $%
187: 10^{-3}-10^{-2}$ for high-temperature superconductors. If we take $4\pi
188: M\sim 1T$, $T_{c}\sim 3K$, $\epsilon _{F}\sim 30,000K$
189: and $d_{m}\sim 300$\AA ,
190: we obtain $\Delta T_{c}/T_{c}\sim 0.1$. The dependence $\Delta
191: T_{c}\propto m^{2}=M^{2}d_{m}^{2}$ on the thickness of the FM film and
192: magnetization can be checked experimentally.
193: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
194: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
195: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
196: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
197: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
198:
199: \section{3. Spontaneous stripe structure in external field}
200:
201: In this section we study the spontaneous stripe system in the
202: FM-SC bilayer in the presence of an external perpendicular
203: magnetic field $B$ (along $\hat{z}$ direction). Since the
204: magnetization tends to align along the external magnetic field we
205: anticipate that the width $L_{1}$ of stripes with the
206: magnetization parallel to the external magnetic field increases,
207: whereas the width $L_{2}$ of the stripes with the antiparallel
208: magnetization decreases. Let us define a step function with the
209: period $L=L_{1}+L_{2}$ as follows:
210: \[
211: s(x)=\left\{
212: \begin{array}{ll}
213: +1 & \mbox{ ($0<x<L_1$),} \\
214: -1 & \mbox{ ($L_1<x<2 L_2$).}
215: \end{array}
216: \right.
217: \]
218: The Fourier transform of $s(x)$ is:
219: \begin{eqnarray}
220: s_{G}=\left\{
221: \begin{array}{ll}
222: 2i(1-e^{iGL_{1}})/(LG) & \mbox{ ($G \neq 0$),} \\
223: (L_{1}-L_{2})/L & \mbox{ ($G=0$).}
224: \end{array}
225: \right.
226: \label{Fourier-step}
227: \end{eqnarray}
228: Here $G=2\pi r/L$ and $r$ are integers. For the sake of brevity, we denote $%
229: t=L_{1}-L_{2}$. At large distance from the bilayer the magnetic field
230: asymptotically becomes equal to the external magnetic field. The total
231: magnetic flux is the same in any cross- section of the space. Thus, the
232: average magnetic field through superconducting layer is
233: \begin{equation}
234: \frac{1}{L}\int_{0}^{L}n(\mathbf{x})\phi _{0}dx=B_{ext}\,.
235: \end{equation}
236: This constraint can be incorporated by the standard Lagrange
237: multiplier method. According to it, the problem is reduced to
238: minimization of the effective energy
239: $\tilde{U}_{v}=U_{v}+U_{\omega }$, where $U_{v}$ is the energy of
240: the vortex system, and $U_{\omega }=\omega \frac{\phi _{0}}{L}\int
241: n(\mathbf{x})x$, $\omega $ is the Lagrange multiplier. The general
242: expression for the free energy of a periodic stripe system of
243: magnetization and vortices is given by equation (10) of the work
244: \cite{Erdin2}. Employing this equation and the Fourier expansion
245: for the step function $s(x)$ (see equation (\ref{Fourier-step}))
246: and denoting $n_{G}$ the Fourier-transform of the vortex density
247: $n(x)$, we obtain:
248: \begin{eqnarray}
249: \tilde{U}_{v} &=&\sum_{G}\tilde{\epsilon}_{v}s_{G}n_{-G}+\frac{1}{2}%
250: \sum_{G\neq 0}V_{G}n_{G}n_{-G} \nonumber \\
251: &&+\frac{\tilde{\epsilon}_{v}B_{ext}t}{\phi _{0}L}-\omega (\phi
252: _{0}n_{G}-B_{ext})\,, \label{U-v}
253: \end{eqnarray}
254: where $V_{G}=\phi _{0}^{2}/(2\pi |G|)$ is the Fourier-transform of the
255: vortex interaction. An infinitely large interaction term $%
256: V_{G=0}n_{G=0}n_{G=0}$ has been omitted since it corresponds to the fixed
257: average magnetic field. From equation( \ref{U-v}) we readily find that the
258: constraint condition implies:
259: \begin{equation}
260: n_{G=0}=\frac{B_{ext}}{\phi _{0}}\,. \label{constr}
261: \end{equation}
262: This equation confirms that $V_{G=0}n_{G=0}n_{G=0}$ is the energy of the
263: uniform external field. Minimization of the total vortex energy $\tilde{U}%
264: _{v}$ over the vortex density $n_{\mathbf{G}}$ results in a system of
265: equations:
266: \[
267: \left\{
268: \begin{array}{ll}
269: \tilde{\epsilon}_{v}s_{G}+V_{G}n_{G}=0 & \mbox{($G \neq 0$),} \\
270: \tilde{\epsilon}_{v}s_{G=0}=\Lambda \phi _{0} & \mbox{($G=0$).}
271: \end{array}
272: \right.
273: \]
274: Plugging the solutions of $n_{\mathbf{G}}$($G\neq 0)$, $\omega $ and $n_{G=0}
275: $ from equation (\ref{constr}), we finally obtain
276: \begin{eqnarray}
277: U &=&\frac{-8{\tilde{\epsilon}_{v}^{2}}}{\phi _{0}^{2}L}[C+\ln {\frac{L}{%
278: \lambda }}+\frac{1}{2}\ln (2+2\cos {\frac{\pi t}{L}})] \nonumber \\
279: & &+\frac{\tilde{\epsilon}_{v}B_{ext}t}{\phi _{0}L}
280: +\frac{2\epsilon _{dw}}{L}\,.
281: \end{eqnarray}
282: We are now in a position to minimize the total energy $U$ over $L$ and $t$.
283: After doing that, we obtain two equations:
284: \begin{eqnarray}
285: &&C-1+\ln {\frac{2L}{\lambda }}-\ln (1+\tan ^{2}{u})^{1/2}-\frac{\tilde{%
286: \epsilon}_{dw}}{4{\tilde{m}}^{2}}=0\,, \label{L-t.a} \\
287: &&\tan {u}=\frac{LB_{ext}}{4\pi \tilde{m}}\,. \label{L-t.b}
288: \end{eqnarray}
289: Here $u=\frac{\pi t}{2L}$. From equations (\ref{L-t.a}) and
290: (\ref{L-t.b}) we find solutions for $L$ and $t$:
291: \begin{eqnarray}
292: L &=&\frac{2L_{s}}{\sqrt{1-(\frac{L_{s}B_{ext}}{2\pi \tilde{m}})^{2}}}\,,
293: \label{L} \\
294: t &=&\frac{2L}{\pi }\arctan {\frac{LB_{ext}}{4\pi \tilde{m}}}\,. \label{t}
295: \end{eqnarray}
296: where $L_{s}$ is given by equation (\ref{Ls}). The results. (\ref{L}) and (%
297: \ref{t}) are similar to those for a purely ferromagnetic stripe structure in
298: a single ferromagnetic film \cite{Kashuba}. The critical external field $%
299: B_{ext}^{c}$ at which the domain structure vanishes is
300: \begin{equation}
301: B_{ext}^{c}=2\pi \tilde{m}/L_{s}\,, \label{ecri}
302: \end{equation}
303: which is in the range of $1-10$ Oe. \newline
304:
305: In conclusion of this section, we consider how the temperature of
306: superconducting transition of the bilayer changes in the presence of
307: external magnetic field. Since at the field $B_{ext}^{c}\thicksim 1-10$ Oe
308: the stripe structure vanishes, the superconducting transition proceeds in
309: the homogeneous state of ferromagnetic film excluding very small vicinity of
310: $T_{c}$. Therefore, it is determined by the nucleation process as in the
311: case of a single superconducting film . The nucleation in a
312: thin film for the field perendicular to it was considered by Tinkham \cite
313: {tinkham}. Though the geometry is different from the bulk geometry
314: considered by Abrikosov\cite{abrikosov}, his solution can be directly applied. The order
315: parameter coincides with the Landau wave function for the first Landau
316: level. In the case of bilayer the energy of the nucleus reads:
317: \begin{equation}
318: U=\int \left[ \frac{1}{2m}\left| \left( \frac{\hbar }{i}\nabla -\frac{2e}{c}%
319: \mathbf{A}_{0}\right) \psi \right| ^{2}+a\left| \psi \right| ^{2}\right]
320: d^{2}x+\Delta U \,. \label{nucleus}
321: \end{equation}
322: It differs from the energy in the absence of magnetic film by the value $%
323: \Delta U=-m\int B_{z}^{(n)}d^{2}x$, where $B_{z}^{(n)}$ is the magnetic
324: field generated by the nucleus at the ferromagnetic film. The magnetic field
325: generated by the nucleus reads:
326: \begin{equation}
327: \mathbf{B}^{(n)}(\mathbf{r)}=\frac{1}{c}\int \triangledown \frac{1}{|\mathbf{%
328: r-r}^{\prime }|}\times \mathbf{j}_{n}(\mathbf{r}^{\prime })d^{3}x^{\prime } \,.
329: \label{field}
330: \end{equation}
331: We assume that the current flows in the $x-y$ plane. Since it has zero
332: divergence, it can be represented as $\mathbf{j}_{n}=\stackrel{\symbol{94}}{z%
333: }\times \nabla f$ , where $f(x,y)$ is a function localized in a finite part
334: of the plane. The flux of the induced field is:
335: \begin{eqnarray}
336: &&\int B_{z}^{(n)}d^{2}x \nonumber \\
337: &&=\frac{1}{c}\int \left( \stackrel{\symbol{94}}{z}%
338: \times \triangledown \frac{1}{|\mathbf{r-r}^{\prime }|}\right) \left(
339: \stackrel{\symbol{94}}{z}\times \nabla f(\mathbf{r}^{\prime })\right)
340: d^{2}xd^{3}x^{\prime } \,.\label{flux}
341: \end{eqnarray}
342: Simple transformations turn this integral into a following form:
343: \begin{equation}
344: \int B_{z}^{(n)}d^{2}x=-\frac{1}{c}\int f(\mathbf{r}^{\prime })\nabla ^{2}%
345: \frac{1}{|\mathbf{r-r}^{\prime }|}d^{2}xd^{3}x^{\prime } \,. \label{flux1}
346: \end{equation}
347: This integral is equal to zero. Indeed, the 3d Laplasian of the inverse
348: distance is proportional to $\delta \left( \mathbf{r}-\mathbf{r}^{\prime
349: }\right) $. The radius-vector $\mathbf{r}$ belongs to the magnetic layer,
350: whereas the radius-vector $\mathbf{r}^{\prime }$ belongs to the
351: superconducting one. What stays in equation (\ref{flux1}) is the 2d Laplasian
352: operator, which differs from the 3d one by the second derivative $\left(
353: \frac{\partial }{\partial z}\right) ^{2}\frac{1}{|\mathbf{r-r}^{\prime }|}$,
354: but the integral from this function over the plane $\mathbf{r}$ is equal to
355: zero when $\mathbf{r}^{\prime }$ does not belong to the same plane.
356:
357: Thus, the interaction of the superconducting nucleus and the
358: homogeneously magnetized film is zero independently on the wave
359: function of the localized nucleus. Therefore, the transition
360: temperature is the same as in the absence of the ferromagnetic
361: film. Earlier we have found that, without magnetic field the
362: ferromagnetic layer increases the transition temperature and its
363: change may be about 1/10 of initial value. Surprisingly this shift
364: disappears at extremely small external field about 1-10 Gs. The
365: reason of this paradox consist of unlimited expansion of domains
366: upon approaching the transition point. An interesting and
367: counterintuitive result of this consideration is that a very small
368: magnetic field about $10$Oe is enough to change the transition
369: temperature relatively by 1/10. After this fast change the
370: transition temperature changes substantially when magnetic field
371: becomes comparable to $H_{c2}$, normally in the range 1$T$.
372:
373: \section{4. Transition Temperature in a Superconducting Film with a Square
374: Array of Ferromagnetic dots}
375:
376: Recently Erdin considered theoretically the vortex-antivortex
377: textures in a superconducting film supplied with a regular square
378: lattice of ferromagnetic dots \cite{Erdin3}. Unfortunately, there
379: occurred a mistake in his code for numerical calculations.
380: Therefore, we reproduce here a part of his analysis, but our
381: conclusions are completely different. We denote the dot lattice
382: constant be $a$ and assume each dot to be a circular thin disk
383: with the radius $R$ and constant magnetization $m$ per unit area
384: directed perpendicular to the plane (along $z$-axis).
385: \begin{figure}[htb]
386: \includegraphics[angle=0,width=85mm]{fm-dots.eps}
387: \caption{\label{f1} Schematics representaion of FM dots with
388: spontaneous vortices and antivortices. The circles
389: drawn by solid line represent FM dots. The dash half-circles
390: with clockwise and anticlockwise arrow indicate
391: vortices and antivortices respectively.}
392: \end{figure}
393: The total energy per unit area
394: of the system is \cite{Erdin3}:
395: \begin{equation}
396: U=u_{vv}+u_{mv}+u_{mm}\,. \label{dots-energy1}
397: \end{equation}
398: The three terms in the right-hand side of the above equation can be written
399: as follows:
400: \begin{eqnarray}
401: u_{vv} &=&\frac{\phi _{0}}{4\pi a^{2}}\sum_{\mathbf{G}}\frac{|F_{\mathbf{G}%
402: }|^{2}}{G(1+2\lambda G)}\,, \\
403: u_{mv} &=&-\frac{\phi _{0}}{a^{2}}\sum_{\mathbf{G}}\frac{m_{z\mathbf{G}}F_{{-%
404: \mathbf{G}}}}{1+2\lambda G}\,, \\
405: u_{mm} &=&-2\pi \lambda \sum_{\mathbf{G}}\frac{G^{2}|\mathbf{m}_{z\mathbf{G}%
406: }|^{2}}{1+2\lambda G} \,.\label{e24}
407: \end{eqnarray}
408: where $\mathbf{G}=\frac{2\pi }{a}(r,s)$ ($r,s$ are integers) are the
409: reciprocal lattice vectors; $F_{\mathbf{G}}=\sum_{i}n_{i}e^{i\mathbf{G}\cdot
410: \mathbf{r}_{i}}$ is the structure factor of the vortex lattice; $n_{i}$, $%
411: \mathbf{r}_{i}$ indicate the vorticity and the position of a $i$-th vortex.
412: Only the change of energy $\tilde{u}_{mm}$ in superconducting state compared
413: to the normal state matters:
414: \begin{equation}
415: \tilde{u}_{mm}=u_{mm}(\lambda )-u_{mm}(\lambda \rightarrow \infty )\,
416: \label{difference}
417: \end{equation}
418: The last term in the r.-h.s. of equation (\ref{difference}) is the
419: dipolar energy of the FM dots above the superconducting
420: transition. Below the superconducting transition temperature
421: $T_{c}$ the magnetic field generated by the dots penetrates into
422: the SC film and can create vortices and antivortices if the
423: magnetization and the size of the dots are large enough
424: \cite{Erdin1}. Erdin \cite{Erdin3} considered the case when only
425: one vortex and one antivortex appear per a magnetic dot. He stated
426: that there is a symmetry violation in the lowest energy state. Our
427: numerical results contradict to this statement: we find that the
428: vortex centers are located precisely under the centers of the
429: magnetic dots, whereas the antivortex centers are located at the
430: centers of the magnetic dot lattice. Employing this fact and
431: keeping in mind that $\lambda \gg L$ near the new transition
432: temperature $T_{c}^{*}$, we can rewrite the total energy
433: (\ref{dots-energy1}) as follows:
434: \begin{eqnarray}
435: u &=&\frac{\phi _{0}e^{2}d_{s}n_{s}}{2\pi m_{s}c^{2}a^{2}}\ln {\frac{a}{\xi }%
436: }-\frac{\phi _{0}^{2}e^{4}d_{s}^{2}n_{s}^{2}}{4\pi ^{2}m_{s}^{2}c^{4}a}I_0 \nonumber \\
437: &&-\frac{\phi _{0}^{2}e^{2}d_{s}n_{s}}{4\pi ^{2}m_{s}c^{2}a^{2}}%
438: (I_1+\frac{4\pi ^{2}mR}{\phi _{0}}I_2) \nonumber \\
439: &&+\frac{2\pi ^{2}m^{2}e^{2}d_{s}n_{s}R^{2}}{m_{s}c^{2}a^{2}}I_3%
440: \,. \label{dots-energy}
441: \end{eqnarray}
442: where $\sum^{\prime }$ means that the term $r=s=0$ are omitted.
443: $I_1$, $I_2$ and $I_3$ are defined as series:
444: \begin{eqnarray}
445: I_0&=&\sum_{n,s=-\infty }^{+\infty \,\prime }\frac{1}{(n^{2}+s^{2})^{3/2}} \nonumber \\
446: I_1&=&\sum_{n,s=-\infty }^{+\infty \,\prime }\frac{(-1)^{n}+(-1)^{s}}{n^{2}+s^{2}} \nonumber \\
447: I_2&=&\sum_{n,s=-\infty }^{+\infty \,\prime }\frac{
448: J_{1}(\frac{2\pi R}{a}\sqrt{n^{2}+s^{2}})[1-(-1)^{n+s}]}{n^{2}+s^{2}} \nonumber \\
449: I_3&=&\sum_{n,s=-\infty }^{+\infty \,\prime }\frac{J_{1}^{2}(\frac{2\pi R}{a}\sqrt{%
450: n^{2}+s^{2}})}{n^{2}+s^{2}} \,.
451: \end{eqnarray}
452: We combine this energy with the Ginzburg-Landau free energy for
453: the SC film as it was done in the stripe structure case:
454: \begin{eqnarray}
455: F &=&\frac{\phi _{0}e^{2}d_{s}n_{s}}{2\pi m_{s}c^{2}a^{2}}\ln {\frac{a}{\xi }%
456: }-\frac{\phi _{0}^{2}e^{4}d_{s}^{2}n_{s}^{2}}{4\pi ^{2}m_{s}^{2}c^{4}a}I_0
457: \nonumber \\
458: & &-\frac{\phi _{0}^{2}e^{2}d_{s}n_{s}}{4\pi ^{2}a^{2}m_{s}c^{2}}(I_1+\frac{%
459: 4\pi ^{2}mR}{\phi _{0}}I_2)
460: +\frac{2\pi ^{2}m^{2}e^{2}d_{s}n_{s}R^2}{m_{s}c^{2} a^2}I_3 \nonumber \\
461: & &+\alpha (T-T_{c})n_{s}d_{s}+\frac{\beta }{2}n_{s}^{2}d_{s}\,.
462: \label{dots-free}
463: \end{eqnarray}
464: The condition of minimum over $n_{s}$ fro the free energy
465: \ref{dots-free} reads:
466: \begin{eqnarray}
467: & &\frac{\phi _{0}e^{2}d_{s}}{2\pi m_{s}c^{2}a^{2}}\ln {\frac{a}{\xi }}-\frac{%
468: \phi _{0}^{2}e^{4}d_{s}^{2}n_{s}}{2\pi ^{2}m_{s}^{2}c^{4}a} I_0 \nonumber \\
469: & &-\frac{\phi _{0}^{2}e^{2}d_{s}}{4\pi ^{2}a^{2}m_{s}c^{2}}(I_1+\frac{%
470: 4\pi ^{2}mR}{\phi _{0}}I_2) +\frac{2\pi ^{2}m^{2}e^{2}d_{s}R^2}{m_{s}c^{2}a^2}I_3 \nonumber \\
471: & &+\alpha(T-T_{c})d_{s}+\beta n_{s}d_{s}=0\,. \label{dots-minimize}
472: \end{eqnarray}
473: At a new critical temperature $T_{c}^{*}$ the density of superconducting
474: carriers must be zero. Plugging $n_{s}(T_{c}^{*})=0$ into eq. (\ref
475: {dots-minimize}), we obtain the shift of the critical temperature:
476: \begin{eqnarray}
477: \Delta T_{c} &=&\frac{\phi _{0}e^{2}}{4\pi ^{2}m_{s}c^{2}a^{2}}(\frac{4\pi
478: ^{2}mR}{\phi _{0}}I_2+I_1 \nonumber \\
479: &&-2\pi \ln {\frac{a}{\xi }}-\frac{8\pi^{4}m^{2}R^{2}}{\phi _{0}^{2}}I_3) \,.
480: \label{dots-shift}
481: \end{eqnarray}
482: Figure (2) shows the relation between $\Delta T_c$ and $R$
483: respectively for $\xi=0.21a$.
484: \begin{figure}[htb]
485: \includegraphics[angle=270,width=85mm]{cir2.eps}
486: \caption{\label{f2} $\Delta T_c$ vs. $R$ for $\xi=0.21a$
487: respectively for $r=10.0$, $12,5$ and $15.0$, here $r=\frac{4\pi^2
488: m L}{\phi_0}$. $\Delta T_c$ is in the unit $\frac{\phi_0^2 e^2}{4
489: \pi^2 a^2 m_s c^2}$.}
490: \end{figure}
491: To ensure spontaneous occurrence of the vortices the inequality
492: $u_{mv}+u_{vv}<0$ must be satisfied. It is equivalent to the
493: following relation:
494: \begin{equation}
495: \frac{4\pi^{2}mR}{\phi _{0}}I_2+I_1-2\pi \ln{\frac{a}{\xi }}<0 \,.
496: \end{equation}
497: The validity of the London's approximation implies $\xi\ll a$.
498: This condition is violated in a close vicinity of the transition
499: temperature, the smaller the larger is the dot lattice constant
500: $a$. For $a\sim 1\mu m$ this vicinity is less or of the order of
501: $0.01 T_c$ and further we neglect it. Fig. (2) shows that the shift
502: of transition temperature is a rather complicated function of the
503: dots radius $R$ and the ratio $r=4\pi^2mL/\phi_0$. For each
504: value $r$, there exists a threshold radius $R_0$, at which the
505: vortices first appear. The shift of the transition temperature
506: grows near the threshold with $R/a$ until maximum and then decreases,
507: it remain negative in the interval $R_0/a$ and $1/2$. For each fixed $R/a$,
508: the shift of the transition temperature increases with the ratio $r$ and is
509: negtive.
510:
511: \section{5. Ferromagnetic Textures in multilayers}
512:
513: Let us with a FM-SC multilayer consisting of $N$ bilayers, each
514: having the thickness $d$ in the limit $Nd\gg L_{s}$, where $%
515: L_{s}$ is the lateral size of the layers. If the magnetic films
516: are magnetized perpendicularly, the average induction inside the
517: multilayer is $B=4\pi m/d$ and it is directed perpendicularly to
518: the layers. The situation is the same as in the layered
519: superconductors placed into an external magnetic field
520: \cite{Blatter}. Therefore, pancake vortices in each
521: superconducting layer must appear. Together they form the
522: Abrikosov linear vortices and satisfy a condition: $m\phi
523: _{0}/d>\epsilon _{l}$, which guarantees that the vortex line is
524: energy favorable. Here $\epsilon _{l}=\epsilon _{0}\ln
525: {\frac{\lambda }{\xi }}$ \cite{Clem} is the usual vortex line
526: energy per unit length, $\epsilon _{0}=\phi _{0}^{2}/(4\pi \lambda
527: )^{2}$ and $\lambda $ is the planar bulk penetration depth. There
528: is no need to consider the Josephson coupling effect in this case
529: since the phase difference between superconducting layers is zero
530: if the vortex lines are perpendicular to the layers. On the other
531: hand, the Josephson vortices appear along the layers if the
532: magnetization $\mathbf{m}$ is parallel to the layers and satisfy a
533: condition $m\phi _{0}/d>\epsilon _{J}$ where $\epsilon _{J}=\gamma
534: \epsilon _{0}\ln {\frac{\lambda }{d}}$ is the Josephson vortex
535: line energy and $\gamma $ is the anisotropy parameter for the
536: layered superconductor\cite {Blatter}. These ideas were applied by
537: M. Houzet et al. \cite{Houzet} to explain the magnetic properties
538: of the magnetic superconductor RuSr${}_{2}$ GdCu${}_{2}$O${}_{8}$.
539: In this article we presumably study the opposite limit $Nd\ll
540: \Lambda $, where $\Lambda =\lambda ^{2}/d$ is the effective
541: penetration depth for layered superconductors.
542:
543: In this section we first focus on few-layer superconductors
544: without ferromagnetic texture, which will be discussed later.
545: Pancake vortices in a finite stack of layers were discussed by
546: Mints et al. \cite{mints}. We modify a method developed by
547: Efetov \cite{Efetov} for a plane superconductor and by K.
548: Fischer \cite{Fischer} for a layered superconductor with infinite
549: number of layers. We consider a superconductor consisting of $N$
550: layers coupled only by their magnetic field. To simplify the
551: calculation, we assume that layers are infinitely thin and located
552: at the planes $z_{n}=nd$ ($n$ is an integer). The vector potential
553: $\mathbf{A}$ due to the pancake vortices at superconducting layers
554: satisfy a following equation:
555: \begin{eqnarray}
556: &&-\Delta \mathbf{A}+\frac{1}{\Lambda }\sum_{n}\delta (z-z_{n})\mathbf{A}
557: \nonumber \\
558: &=&\frac{\phi _{0}}{2\pi \Lambda }\sum_{n}\delta (z-z_{n})\sum_{\mathbf{\rho
559: }}\delta _{\mathbf{\rho }}\mathbf{\nabla }^{(2)}\varphi _{n}(\mathbf{r-\rho
560: )\,.} \label{multi-equation}
561: \end{eqnarray}
562: The currents in equation (\ref{multi-equation}) are induced by
563: pancake vortices with the vorticity $\delta _{\mathbf{\rho }}=\pm
564: 1$ placed at the position $\mathbf{\rho }$ in the $n$-th plane.
565: The Coulomb gauge $\mathbf{ \nabla }\cdot \mathbf{A}=0$ was used.
566: In addition, $A_{z}=0$ because the direction of $\mathbf{\nabla
567: }^{(2)}\varphi _{n}$ is along the layers. It is useful to
568: introduce an auxiliary potential ${\tilde{\mathbf{A}}}(\mathbf{r,}
569: z)=\sum_{n}\delta (z-z_{n})\mathbf{A}(\mathbf{r},z)$ confined to
570: the layers, the ''London vector'' \cite{Fischer}
571: $\mathbf{\phi}_{n}(\mathbf{r})=\sum_{ \mathbf{\rho }}\delta
572: _{\mathbf{\rho }}\frac{\phi _{0}}{2\pi }\mathbf{\nabla
573: }^{(2)}\varphi _{n}(\mathbf{r}-\mathbf{\rho })$ and corresponding
574: auxiliary vector
575: $\mathbf{\tilde{\phi}}_{n}(\mathbf{r,}z)=\sum_{n}\mathbf{\phi
576: }_{n}( \mathbf{r})\delta (z-z_{n})$. In terms of these variables
577: equation (\ref {multi-equation}) can be rewritten as follows:
578: \begin{equation}
579: -\Delta \mathbf{A}+\frac{1}{\Lambda }{\tilde{\mathbf{A}}=}\frac{\phi _{0}}{%
580: 2\pi \Lambda }\mathbf{\tilde{\phi}}_{n} \,. \label{multi-transf}
581: \end{equation}
582: Equation (\ref{multi-transf}) can be solved by Fourier-transformation. The
583: partial result is:
584: \begin{equation}
585: \mathbf{A}(\mathbf{q},k)=\sum_{n}e^{-ikz_{n}}\frac{\phi _{n}(\mathbf{q})-%
586: \mathbf{A}_{n}(\mathbf{q})}{\Lambda (q^{2}+k^{2})} \,.\label{multi-Fourier}
587: \end{equation}
588: where $\mathbf{A}(\mathbf{q},k)$ is the Fourier-transform of the
589: vector-potential $\mathbf{A}(\mathbf{r},z)$ and $\mathbf{A}_{n}(\mathbf{q})$
590: is the plane Fourier-transform of the vector-potential $\mathbf{A}(\mathbf{r}%
591: ,z_{n})$ taken at the $n-$th superconducting plane. Expressing
592: this value by the inverse Fourier-transformation, we find a system
593: of equations for $ \mathbf{A}_{n}(\mathbf{q})$:
594: \begin{eqnarray}
595: &&\sum_{n}(\frac{1}{2\Lambda q}e^{-q|m-n|d}+\delta _{mn})\mathbf{A}_{n}(%
596: \mathbf{q}) \nonumber \\
597: &=&\frac{1}{2\Lambda q}\sum_{n}\mathbf{\phi }_{n}(\mathbf{q})e^{-q|m-n|d}\,.
598: \label{multi-system}
599: \end{eqnarray}
600: We apply eq. (\ref{multi-system}) to study the simplest case of
601: two superconducting layers. Let only one pancake vortex to be
602: placed in the center of the layer $z=0$ at $\mathbf{\rho }=0$. The
603: other layer is located at $z=d$. The solution of the system
604: (\ref{multi-system}) for this situation reads:
605: \begin{eqnarray}
606: \mathbf{A}_{1}(\mathbf{q}) &=&\frac{1+2\Lambda q-e^{-2qd}}{1+4\lambda
607: q+4\Lambda ^{2}q^{2}-e^{-2qd}}\mathbf{\phi }_{1}(\mathbf{q)} \nonumber \\
608: \mathbf{A}_{2}(\mathbf{q}) &=&\frac{2\Lambda qe^{-2qd}}{1+4\lambda
609: q+4\lambda ^{2}q^{2}-e^{-2qd}}\mathbf{\phi }_{1}(\mathbf{q)\,.}
610: \label{multi-two}
611: \end{eqnarray}
612: In the limit $qd\ll 1$ this solution becomes simpler:
613: \begin{eqnarray}
614: \mathbf{A}_{1}(\mathbf{q}) &=&\frac{1}{2+2\Lambda q}\mathbf{\phi }_{1}(%
615: \mathbf{q)} \nonumber \\
616: \mathbf{A}_{2}(\mathbf{q}) &=&\frac{1}{2+2\Lambda q}\mathbf{\phi }_{1}(%
617: \mathbf{q)\,.} \label{multi-two-1}
618: \end{eqnarray}
619: The current density in each layer is given by:
620: \begin{eqnarray}
621: \mathbf{J}_{1}(\mathbf{q}) &=&\frac{c}{4\pi \Lambda }(\mathbf{\phi }_{1}(%
622: \mathbf{q)-A}_{1}\mathbf{(q))} \nonumber \\
623: \mathbf{J}_{2}(\mathbf{q}) &=&-\frac{c}{4\pi \Lambda }\mathbf{A}_{2}(\mathbf{%
624: q})\,. \label{multi-two-current}
625: \end{eqnarray}
626: Returning to the space coordinates, we find from eqs. (\ref{multi-two-1}), (%
627: \ref{multi-two-current}) in different asymptotic regions:
628:
629: \begin{equation}
630: \begin{tabular}{lll}
631: & $r\gg \lambda $ & $d\ll r\ll \Lambda $ \\
632: $\mathbf{J}_{1}(r)$ & $\frac{\phi _{0}c}{16\pi ^{2}\Lambda r}\hat{\varphi}$
633: & $\frac{\phi _{0}c}{8\pi ^{2}\Lambda r}\hat{\varphi}$ \\
634: $\mathbf{J}_{2}(r)$ & $-\frac{\phi _{0}c}{16\pi ^{2}\Lambda r}\hat{\varphi}$
635: & $-\frac{\phi _{0}c}{4\Lambda ^{2}}\hat{\varphi}$%
636: \end{tabular}
637: \label{multi-coord}
638: \end{equation}
639: The force between two pancake vortices is $\mathbf{F}=-\frac{\phi
640: _{0}}{c} \hat{z}\times \mathbf{J}$, where $\mathbf{J}$ is the
641: current produced by one pancake at the center of another one. It
642: follows from eqs. (\ref{multi-coord} ) that the energy of
643: interaction between two pancakes with the same vorticity at the
644: same layer is logarithmic and repulsive at the distance $ R\gg d$.
645: A peculiarity of the few-layer structure is that the interaction
646: energy of two pancake vortices with the same vorticity at
647: different layers, separated by the distance at $R\gg \Lambda $, is
648: logarithmic and attractive. It has the same absolute value as the
649: repulsion of the pancakes in the same layer. This result
650: dramatically differs from the interaction energy of two vortices
651: at different layers, when the total number of layers is infinite:
652: in this case the interaction in different layers differs from the
653: interaction in the same layer by a small pre-factor $d/\lambda $ .
654: The result on the logarithmic attraction of two pancakes in
655: different layers and its amplitude we have obtained persists at
656: any number of layers $N$ provided $Nd\ll \lambda $. It can be
657: interpreted as the attraction of two ''half-vortices'' in the two
658: plane, one carrying the flux $+\phi _{0}/2$, other carrying $-\phi
659: _{0}/2$.
660:
661: In the two-layer system the asymptotics for the components of
662: magnetic field produced by the pancake vortex located in the plane
663: $z=0$ at its origin directly follow from equation
664: (\ref{multi-two-1}). In the range $r\gg \Lambda $ they are:
665: \begin{eqnarray}
666: B_{z} &=&\frac{\phi _{0}(|z|+|z-d|)}{8\pi (z^{2}+r^{2})^{3/2}} \label{Bz1}
667: \\
668: B_{r} &=&\frac{\phi _{0}}{8\pi \Lambda r}sgn(z)(1-\frac{|z|}{\sqrt{%
669: r^{2}+z^{2}}})\hat{r} \nonumber \\
670: &&-\frac{\phi _{0}}{8\pi \Lambda r}sgn(z-d)(1-\frac{|z-d|}{\sqrt{r^{2}+z^{2}}%
671: })\hat{r} \nonumber \\
672: &&+\frac{\phi _{0}(2z-d)}{8\pi (r^{2}+z^{2})^{3/2}}\hat{r}\,. \label{Br1}
673: \end{eqnarray}
674: In another asymptotic region $d\ll r\ll \Lambda $ we find:
675: \begin{eqnarray}
676: B_{vz} &=&\frac{\phi _{0}}{4\pi \Lambda \sqrt{r^{2}+z^{2}}} \label{Bz2} \\
677: \mathbf{B}_{v}^{(2)}(\mathbf{r},z) &=&\frac{\phi _{0}}{4\pi \Lambda r}%
678: sgn(z)(1-\frac{|z|}{\sqrt{r^{2}+z^{2}}})\hat{r}\,. \label{Br2}
679: \end{eqnarray}
680: Due to the strong screening effect exerted by one layer to another
681: the magnetic field decays more quickly in the $z$-direction than
682: in $r$ -direction. The total magnetic flux through the plane $z=0$
683: and $z=d$ is $ \Phi (z=0)=B_{vz}(\mathbf{q}=0,z=0)=\frac{\Lambda
684: +d}{2\Lambda +d}\phi _{0}\approx \phi _{0}/2$, and $\Phi
685: (z=d)=B_{vz}(\mathbf{q}=0,z=d)=\frac{ \Lambda }{2\Lambda +d}\phi
686: _{0}\approx \phi _{0}/2$. The two fluxes are not exactly equal,
687: and the net flux $d/(2\Lambda +d)$ escapes through the remote side
688: surface.
689:
690: The self-energy of a single pancake vortex reads:
691: \begin{eqnarray}
692: E_{sv} &=&\frac{1}{8\pi \Lambda }\int \frac{d^{2}q}{(2\pi )^{2}}[|\mathbf{%
693: \phi }_{1}(\mathbf{q})|^{2}-\mathbf{\phi }_{1}(-\mathbf{q})\cdot \mathbf{A}%
694: _{1}(\mathbf{q})] \nonumber \\
695: &=&\frac{1}{8\pi \Lambda }\int \frac{d^{2}q}{(2\pi )^{2}}[\frac{\phi _{0}^{2}%
696: }{q^{2}}-\frac{\phi _{0}^{2}}{2q^{2}(1+\Lambda q)}] \nonumber \\
697: &=&\frac{\phi _{0}^{2}}{%
698: 16\pi \Lambda }\ln \frac{R_{s}}{\xi }\,. \label{single}
699: \end{eqnarray}
700: Here $R_{s}$ is the lateral linear size of the sample. Due to divergence of $%
701: E_{sv}$, it is energy unfavorable to produce single pancake in a
702: layer below the Berezinsky-Kosterlitz-Touless transition. The
703: energy of a pair of pancake vortices located one opposite another
704: at different planes is:
705: \begin{eqnarray}
706: E_{lv} &=&\frac{2}{8\pi \Lambda }\int \frac{d^{2}q}{(2\pi )^{2}}[|\mathbf{%
707: \phi }_{1}(\mathbf{q})|^{2}-\mathbf{\phi }_{1}(-\mathbf{q}) \nonumber \\
708: & & \cdot (\mathbf{A}%
709: _{v1}(\mathbf{q})+\mathbf{A}_{v2}(\mathbf{q})] \nonumber \\
710: &=&\frac{1}{4\pi \Lambda }\int \frac{d^{2}q}{(2\pi )^{2}}[\frac{\phi _{0}^{2}%
711: }{q^{2}}-\frac{\phi _{0}^{2}}{q^{2}(1+\Lambda q)}] \nonumber \\
712: &=&\frac{\phi _{0}^{2}}{8\pi
713: \Lambda }\ln \frac{\Lambda }{\xi }\,. \label{pair-energy}
714: \end{eqnarray}
715: The interaction of two such a pairs separated by a distance $R\gg
716: d$ is:
717: \begin{eqnarray}
718: V_{ll}(R) &=&\frac{2}{8\pi \Lambda }\int \frac{d^{2}q}{(2\pi )^{2}}[|\mathbf{%
719: \phi }_{1}(\mathbf{q})(1+e^{-i\mathbf{q}\cdot \mathbf{R}})|^{2} \nonumber \\
720: &&-\mathbf{\phi }_{1}({-\mathbf{q}})\cdot (\mathbf{A}_{v1}(\mathbf{q})+%
721: \mathbf{A}_{v2}(\mathbf{q})|1+e^{-i\mathbf{q}\cdot \mathbf{R}}|^{2}] \nonumber \\
722: & &-2E_{lv} \nonumber \\
723: &=&\frac{\phi _{0}^{2}}{4\pi ^{2}}\int \frac{J_{0}(qR)}{1+\Lambda q}dq \nonumber \\
724: &=&\frac{%
725: \phi _{0}^{2}}{8\pi \Lambda }[\mathbf{H}_{0}(\frac{R}{\Lambda })-Y_{0}(\frac{%
726: R}{\Lambda })]\,. \label{pair-int}
727: \end{eqnarray}
728: In the last step we used the formula \cite{pru}:
729: \begin{equation}
730: \int_{0}^{\infty }\frac{1}{x+z}J_{0}(cx)dx=\frac{\pi }{2}[\mathbf{H}%
731: _{0}(cz)-N_{0}(cz)]\,,
732: \end{equation}
733: where $\mathbf{H}_{0}(x)$ is the zeroth Struve function, and $N_{0}(x)$ is
734: the zeroth Neumann function. The asymptotics of the interaction energy (\ref
735: {pair-int}) at small and large distances are as follows:
736: \begin{equation}
737: V_{ll}(R)=\left\{
738: \begin{array}{ll}
739: \frac{\phi _{0}^{2}}{4\pi ^{2}\Lambda }\ln {\frac{\Lambda }{R}} & (r\ll
740: \Lambda ) \\
741: \frac{\phi _{0}^{2}}{4\pi ^{2}R} & (r\gg \Lambda )\,.
742: \end{array}
743: \right. \label{pair-int-as}
744: \end{equation}
745:
746: Equation (\ref{multi-system}) can be solved by the same method for
747: any number of layers, though calculations become more cumbersome.
748: However, in the region $R\gg Nd$ equation (\ref{multi-system}) can
749: be solved quite easily. The vector potential of a line vortex,
750: identical at all layers reads:
751: \begin{equation}
752: \mathbf{A}_{1}=\cdots =\mathbf{A}_{N}=\frac{iN\phi _{0}\hat{q}\times \hat{z}%
753: }{q(N+2\Lambda q)}\,. \label{vortexp}
754: \end{equation}
755: Equation (\ref{vortexp}) allows to calculate the magnetic field ,
756: the current, and the interaction energy. Specifically, the single
757: line self-energy and the interaction energy of two linear vortices
758: for an $N$-layer superconductor are:
759: \begin{equation}
760: E_{lv}=\frac{N\phi _{0}^{2}}{16\pi ^{2}\Lambda }\ln {\frac{\Lambda }{\xi }}%
761: \,; \label{line-energy}
762: \end{equation}
763: \begin{equation}
764: V_{ll}(R)=\left\{
765: \begin{array}{ll}
766: \frac{N\phi _{0}^{2}}{4\pi ^{2}\Lambda }\ln {\frac{\Lambda }{R}} & (r\ll
767: \Lambda ) \\
768: \frac{\phi _{0}^{2}}{4\pi ^{2}R} & (r\gg \Lambda )\,.
769: \end{array}
770: \right. \label{line-int}
771: \end{equation}
772: From these equations we see that the energy of the vortex line in
773: a few-layer system is the same as that of the Pearl vortex in a
774: thin-film superconductor (if we replace $Nd$ by $d_{s}$), but
775: their interaction at short distances is $N$ times stronger than
776: the corresponding Pearl vortex interaction. At long distance, the
777: interaction energy is the same as for the Pearl vortices.
778:
779: Next, we discuss ferromagnetic textures in a few-layer system. We
780: assume that the SC and FM layers form very thin bi-layers
781: separated by a finite distance $d$. The London-Pearl equations for
782: the vector potential $\mathbf{A}_{m}$ induced by the magnetic
783: layers and screened by superconducting layers are:
784: \begin{eqnarray}
785: &&-\Delta \mathbf{A}_{m}+\frac{1}{\Lambda }\sum_{n}\delta (z-z_{n})\mathbf{A}_{m} \nonumber \\
786: &&=4\pi \sum_{n}\mathbf{\nabla }\times [\mathbf{m}\delta (z-z_{n})]\,.
787: \label{multi-ferro}
788: \end{eqnarray}
789: Comparing it with equation (\ref{multi-equation}), we find that they become
790: identical if we replace ${i\phi _{0}{\hat{q}}\times {\hat{z}}}/{q}$ by $%
791: i4\pi m_{q}\Lambda \mathbf{q}\times {\hat{z}}$. Therefore, it is
792: straightforward to obtain the result for the magnetic vector
793: potential from that for the vector potential induced by
794: superconducting vortices. The Fourier-transform of vector
795: potential at each layer produced by an FM texture, identical in
796: each plane, reads:
797: \begin{equation}
798: \mathbf{A}_{m1}=\cdots =\mathbf{A}_{mn}=\frac{i4N\pi m_{q}\Lambda \mathbf{q}%
799: \times \hat{z}}{N+2\Lambda q}\,. \label{magp}
800: \end{equation}
801: \newline
802: Equations (\ref{vortexp}) and (\ref{magp}) allow to calculate the
803: interaction of ferromagnetic textures and vortex-ferromagnet interaction
804: energy given the configuration of the magnetic texture.
805:
806: Let us consider the spontaneous stripe structure in a few-layer
807: ferro-superconducting system. Under the same assumption about the
808: stripe width $L\gg \Lambda $ and the average distance $\bar{R}$
809: between vortices $ \bar{R}\gg \Lambda $, we find from equations
810: (\ref{line-energy}) and (\ref {line-int}) that the interaction
811: energy between two vortex lines is the same as that in single
812: layer, but the single line energy increases $N$ times. The total
813: vortex-ferromagnet interaction energy also increases $N$ times,
814: because the magnetic vector potential and vortex vector potential
815: both increase $N$ times if $L\gg \Lambda $. That means that the
816: condition $m\phi _{0}>\epsilon _{v}$ required for spontaneous
817: formation of vortices and anti-vortices does not change. The
818: domain width for a few-layer is:
819: \begin{equation}
820: L_{s}^{\prime }=\frac{\Lambda }{4}\exp (\frac{\epsilon _{dw}}{4{N\tilde{m}}%
821: ^{2}}-C+1)\,. \label{few-width}
822: \end{equation}
823: The factor $1/N$ in the exponent (\ref{few-width})significantly
824: reduces the domain width in the few-layer system. The sum of
825: widths of parallel and antiparallel domains in an external
826: magnetic field (the period of the domain structure) is:
827: \begin{equation}
828: L^{\prime }(B_{ext})=\frac{2L_{s}^{\prime }}{\sqrt{1-(\frac{L_{s}^{\prime
829: }B_{ext}}{2N\pi \tilde{m}})^{2}}}\,; \label{few-width-magn}
830: \end{equation}
831: whereas the ratio of the width \medskip parallel domain to the
832: period reads:
833: \begin{equation}
834: t^{\prime }=\frac{2L^{\prime }}{\pi }\arctan \frac{L^{\prime }B_{ext}}{4N\pi
835: \tilde{m}}\,. \label{few-ratio}
836: \end{equation}
837: The critical field, at which the stripe structure vanishes
838: follows from eq. (\ref{few-width-magn}):
839: \begin{equation}
840: B_{ext}^{c \prime}=\frac{2N\pi \tilde{m}}{L_{s}^{\prime }} \,;
841: \end{equation}
842: Note that it is proportional to the number of layers. The shift of
843: the transition temperature $\Delta T_{c}^{\prime }$ is:
844: \begin{equation}
845: \Delta T_{c}^{\prime }=\frac{64N\pi m^{2}e^{2}}{\alpha m_{s}c^{2}}\exp (%
846: \frac{-\epsilon _{dw}}{4Nm^{2}}+C-1)\,. \label{few-shift}
847: \end{equation}
848: It also grows with increasing $N$. For the case of few SC films
849: with the square array of of FM columnar dots, the shift of the SC
850: transition temperature is the same as for a single SC film with
851: the square array of o ferromagnetic dots. This result can be
852: readily seen from the observation observe that $\Lambda \gg Na$
853: near the transition temperature. Then equation (\ref{line-energy})
854: implies that the vortex line energy in a few-layer system is $N$
855: times larger than for one layer. The total Ginzburg-Landau free
856: energy for few layers is $N$ times larger than that for one
857: layer (\ref{dots-free}). Thus, equation for $n_{s}$ will not
858: change and the shift of the transition temperature will not change
859: either.
860:
861: \section{6. conclusions}
862: We studied the SC transition temperature in heterogeneous FM-SC
863: system in the London's approximation. The stripe structure of FSB,
864: if exists, leads to the positive shift of the transition
865: temperature.It can reach the value $\Delta T_c\sim 0.1 T_c$, when
866: the width of the stripe becomes comparable with the effective
867: penetration depth. Theory predicts that the shift of the
868: transition temperature is proportional to the square of
869: magnetization per unit area $m^2$. This fact can be checked
870: experimentally. It was demonstrated that that the stripe structure
871: must vanish at a very small external magnetic field between 1 and
872: few tens Oersted. Simultaneously the transition temperature
873: changes by the same value $\Delta T_c\sim 0.1 T_c$. This
874: theoretical prediction opens way to may be the strongest magneto-
875: resistive effect.
876:
877: In the multilayers the value of the value of the magnetic field at
878: which the stripes disappear increases proportionally to the number
879: of layers and the shift of the transition temperature grows even
880: faster.
881:
882: The shift of transition temperature in the superconducting layers
883: supplied with a periodic array of magnetic dots may be of the same
884: order of magnitude, but is negative at reasonable values of
885: parameters.
886:
887: \section{Acknowledgements}
888:
889: This work was supported by the NSF under the grants DMR 0072115,
890: DMR 0103455, by the DOE under the grant DE-FG03-96ER45598 and by
891: Telecommunications and Informatics Task Force at Texas A\&M
892: University. We appreciate the fruitful discussion with Dr. Serkan
893: Erdin on the numerical calculation codes.
894:
895: \begin{thebibliography}{99}
896: \vskip-0.5cm
897: \bibitem{yot} Y. Otani, B. Pannetier, J.P. Nozieres and D. Givord, J. Mag.
898: Mat. \textbf{126}, 622(1993); O. Geoffroy, D. Givord, Y. Otani, B.
899: Pannetier and F. Ossart, J. Magn. mag. met. \textbf{121},233(1993); Y.
900: Nozaki, Y. Otani, K. Runge, H. Miyaima, B. Pannetier, J.P. Nozieres and G.
901: Fillion, J. Appl. Phys. \textbf{79},8571(1996); I.K. Marmorkos, A. Matulis
902: and F.M. Peeters, Phys. Rev. B\textbf{53},2677(1996); J. Martin, M. Velez,
903: J. Nogues and I.K. Schuller, Phys. Rev. lett. \textbf{79}, 1929(1997); D.J.
904: Morgan and J.B. Ketterson, Phys. Rev. Lett. \textbf{80},3614(1998); I.F.
905: Lyuksyutov and V.L. Pokrovsky, Phys. Rev. Lett. \textbf{81}, 2344(1998);
906: I.F. Lyuksyutov and D.G. Naugle, Modern Phys. Lett. B \textbf{13},491(1999);
907: A. Terentiev, D.B. Watkins, L.E. De Long, D.J. Morgan and J.B. Ketterson,
908: Physica C \textbf{332}, 5(2000); M.J. Van Bael, L.Van Look, K. Temst, M.
909: Lange, J. Bekaert, U. May, G. Guntherodt, V.V. Moshchalkov and Y.
910: Bruynseraede, Physica C \textbf{332}, 12(2000); Feldman D.E., Lyuksyutov
911: I.F., Pokrovsky V.L. and V.M. Vinokur, Europhys. Lett. \textbf{51},
912: 110(2000); J.E. Santos, E. Frey and F. Schwabl, Phys. Rev. B \textbf{63},
913: 4439(2001); L.E. Helseth, Phys. B \textbf{66}, 104508(2002); M.A. Kayali,
914: Phys. Lett. A \textbf{298}, 432(2002).
915:
916: \bibitem{igor} I. E. lyuksyutov and V. L. Pokrovky, Mod. Rev. Lett. B
917: \textbf{14}, 409(2000).
918:
919: \bibitem{Erdin1} S. Erdin, A. M. Kayali, I.F. Lyuksyutov, and V.L.
920: Pokrovsky, Phys. Rev. B \textbf{66},014414(2002).
921:
922: \bibitem{Erdin2} S. Erdin, I.F. Lyuksyutov, V.L. Pokrovsky, V.M. Vinokur,
923: Phys. Rev. Lett. \textbf{88},017001(2002).
924:
925: \bibitem{Blatter} G. Blatter, M.V. Feigel'man, V.B. Geshkenbein, A.I.
926: Larkin, V.M. Vinokur, Rev. Mod. Phys. \textbf{66},1125(1994).
927:
928: \bibitem{Clem} J.R. Clem, Phys. Rev. B\textbf{43}, 7837(1991).
929:
930: \bibitem{Houzet} M. Houzet, A. Buzdin, and M.I. Kuli$\acute{c}$, Phys. Rev.
931: B\textbf{64},184501(2001).
932:
933: \bibitem{error} The renormalized energy of vortices is equal to $\tilde{%
934: \epsilon}_{v}=\epsilon _{0}-m\phi _{0}$ instead of $\tilde{\epsilon}%
935: _{v}=\epsilon _{0}-\frac{3}{2}m\phi _{0}$ found in \cite{Erdin2}). This
936: leads to a change of the domain width $L$ and energy $U$ (eqs. (18) and (19)
937: in [\cite{Erdin2}]). The corrected valuess are given by eqs. (U) and (\ref
938: {Ls}) of this article.
939:
940: \bibitem{Gennes} P.G. de Gennes, \textsl{Superconductivity of Metals and
941: Alloys} (Addison-Wesley, New York, 1989).
942:
943: \bibitem{Kashuba} A.B. Kashuba and V.L. Pokrovsky, Phys. Rev. Lett.\textbf{%
944: 70}, 3155(1993); A. Abanov, V. Kalatsky and V.L. Pokrovsky, Phys. Rev. B
945: \textbf{51},1023(1995).
946:
947: \bibitem{Erdin3} S. Erdin, con-mat/0211117.
948:
949: \bibitem{tinkham} M. Tinkham, \textsl{Introduction to Superconductivity}, 2nd
950: ed. (Mc-Graw-Hill, New York, 1996).
951:
952: \bibitem{abrikosov} A.A. Abrikosov, JETP. \textbf{5}, 1174(1957).
953:
954: \bibitem{mints} R.G. Mints, V.K. Kogan, and J.R. Clem, Phys. Rev. B \textbf{%
955: 61},1623(2000).
956:
957: \bibitem{Efetov} K.B. Efetov, Sov. Phys. JETP\textbf{49}, 905(1979).
958:
959: \bibitem{Fischer} K.H. Fischer, Physica C, \textbf{178}, 161(1991).
960:
961: \bibitem{pru} A.P. Prudnikov, Yu.A. Brychkov, O.I. Marichev, \textsl{%
962: Integrals and Series},Vol.\textbf{2} (Gordon and Breach Science Publishers).
963: \end{thebibliography}
964:
965: \end{document}
966: