1: %\documentstyle[preprint,aps]{revtex}
2: %\documentstyle[aps]{revtex}
3: %\documentstyle[preprint,eqsecnum,aps]{revtex}
4: %\documentstyle[eqsecnum,aps]{revtex}
5:
6: \documentstyle[aps,prl,floats,graphicx]{revtex}
7: \begin{document}
8: \draft
9: \twocolumn[\hsize\textwidth\columnwidth\hsize\csname
10: @twocolumnfalse\endcsname
11:
12: %\preprint{Preprint }
13: \title{ Domain structure of superconducting ferromagnets}
14: \author{E.B. Sonin}
15:
16: \address{Racah Institute of Physics, Hebrew University of
17: Jerusalem, Jerusalem 91904, Israel }
18:
19: \date{\today} \maketitle
20:
21:
22: \begin{abstract}
23: In superconducting ferromagnets the equilibrium domain structure is
24: absent in the Meissner state, but appears in the spontaneous vortex
25: phase (the mixed state in zero external magnetic field), though with a
26: period, which can essentially exceed that in normal ferromagnets.
27: Metastable domain walls are possible even in the Meissner state. The
28: domain walls create magnetostatic fields near the sample surface,
29: which can be used for experimental detection of domain walls.
30:
31: \end{abstract}
32:
33: \pacs{PACS numbers: 74.25.Ha, 74.90.+n, 75.60.-d}
34: %\eject
35: ]
36: %\narrowtext
37: %\twocolumn
38:
39:
40: Recently there has been a growing interest to materials, in which
41: superconductivity and ferromagnetism coexist
42: \cite{felner,Nat1,Nat2,Chu,Tallon}. A number of unusual phenomena and
43: structures have been predicted and observed, spontaneous vortex phase as an
44: example\cite{GBV,SF}. But the theory mostly addressed macroscopically uniform
45: structures, whereas ferromagnetic materials, even ideally uniform, inevitably
46: have a domain structure, which is a ground-state property of ferromagnets. So
47: a further progress in studying materials with coexisting ferromagnetism and
48: superconductivity requires an analysis of the domain structure. The present
49: work is the first step in this direction.
50:
51: An object of the study is a material, in which the
52: magnetic transition occurs earlier, i.e. at a higher temperature, than
53: the superconductivity onset. This was called ``superconducting
54: ferromagnet'' \cite{SF}, in contrast to ``ferromagnetic (or magnetic)
55: superconductors'' where the superconductivity sets in {\em before} the
56: magnetic transition, which have been studied mostly in the past
57: \cite{BulBuz}. Competition of ferromagnetism and superconductivity may
58: result in various structures with the magnetic moment rotating
59: in space (spiral structures, cryptoferromagnetism and so on). This
60: also can be considered as a ``domain structure'', but with a
61: period determined by intrinsic properties of materials. However, our
62: goal is the domain structure due to magnetostatic fields
63: generated by nonzero average bulk magnetization $\vec M$. In this
64: case the domain size depends on a sample size. We shall consider
65: type-II superconductivity, bearing in mind ruthenocuprates
66: \cite{felner,Chu,Tallon}, which are type II high-$T_c$ superconductors.
67:
68: Before analyzing the domain structure it is useful to summarize the
69: magnetic properties of a single-domain superconducting ferromagnet. The
70: total free energy of the superconducting ferromagnet can be written as
71: \cite{SF}
72: \begin{equation}
73: F(\vec M, \vec B)=f_E+K +{
74: (\vec B- 4\pi \vec M)^2
75: \over 8\pi } +{2\pi \lambda^2 \over c^2}j_s^2 ~,
76: \label{F-tot}\end{equation}
77: where $\lambda$ is the London penetration depth and $\vec B$ is the
78: magnetic induction. The energy $f_E(M,\nabla \vec M)$ is the exchange
79: energy, which depends on the absolute value of $M$ and on gradients of
80: $\vec M$. As a rule \cite{LL}, in magnetic
81: materials this is the largest energy, which fixes $M$. The anisotropy
82: energy $K(\vec M/M)$ is smaller and depends on the direction of $\vec M$.
83: We shall consider a stripe magnetic structure, which is possible only if
84: $K$ essentially exceeds the magnetostatic energy $\sim M^2$ \cite{LL}.
85: The latter is determined by the magnetic field
86: $\vec H=\vec B- 4\pi \vec M$ [the third term in Eq. (\ref{F-tot})]. The
87: expression Eq. (\ref{F-tot}) includes also the kinetic energy related to
88: the superconducting current
89: \begin{equation}
90: \vec j_s={c\Phi_0 \over 8\pi^2 \lambda^2}\left(\vec \nabla \varphi -
91: {2\pi \vec A \over \Phi_0}\right) ~,
92: \label{j-s}\end{equation}
93: where $\Phi_0$ is the magnetic-flux quantum, $\varphi$ is the phase
94: of the superconducting order parameter, and the vector potential
95: $\vec A$ determines the magnetic induction $\vec B =\vec \nabla
96: \times \vec A$. The kinetic
97: energy of superconducting currents is absent in a normal
98: ferromagnet.
99:
100: Minimization of the energy with respect to the
101: vector potential
102: $\vec A$ yields the Maxwell equation
103: \begin{equation}
104: {4\pi \over c}\vec j_s
105: =\vec \nabla \times (\vec B -4\pi \vec M)
106: =\vec \nabla \times \vec B ~.
107: \label{Max} \end{equation}
108: Together with the equation
109: \begin{equation}
110: \vec \nabla \times \vec j_s=- \frac{c}{4\pi \lambda^2} \vec B
111: \label{rot-j}\end{equation}
112: this yields the London equation which determines $\vec B$:
113: \begin{equation}
114: \lambda^2 \vec \nabla \times [\vec \nabla \times \vec B]
115: + \vec B =0~.
116: \label{Lon-mu} \end{equation}
117: Here we took into account that $\vec \nabla \times \vec M=0$ inside
118: domains. In contrast to Ref. \cite{SF}, we neglect
119: the differential susceptibility ($M$ does not depend on a magnetic
120: field), which renormalizes the London penetration depth.
121:
122:
123: These equations and the boundary conditions at the sample boundary
124: (continuity of the tangential component of $\vec H$ and of the normal
125: component of $\vec B$) yield the distribution of $\vec B $ and
126: $\vec H$. This distribution is shown in Fig. \ref{fig1}
127: for the case of $\vec M$ parallel to the sample boundary and for zero
128: external magnetic field. The magnetic induction and the related magnetic
129: flux exist only in the layer of the thickness $\lambda$. Meissner
130: currents in this layer screen the internal field $4\pi \vec M$, as well
131: as they screen the external magnetic field in a nonmagnetic
132: superconductor.
133:
134: Let us consider now the mixed state of the superconducting
135: ferromagnetic, in which vortices (magnetic fluxons) are present in
136: the bulk. Since ferromagnetism does not affect the London equation
137: (\ref{Lon-mu}), one expect the same magnetic-induction distribution in
138: the mixed state as for nonmagnetic type II superconductors \cite{SF},
139: and the free energy is given by
140: \begin{equation}
141: F_m(\vec M, \vec B)=f_E+K
142: +2\pi M^2 -\vec B\cdot \vec M +F_0(B) ~,
143: \label{F-mix}\end{equation}
144: where $F_0(B)$ is the free energy of a nonmagnetic type II
145: superconductor, and $\vec B$ now is the magnetic
146: induction averaged over the vortex-array cell. The energy $F_0(B)$
147: contains both the magnetic energy $B^2/8\pi$ and the kinetic energy of the
148: superconducting currents inside the vortex cell. Determining the
149: magnetic field
150: \begin{equation}
151: \vec H = 4\pi {\partial F_m \over \partial \vec B}=
152: 4\pi {\partial F_0 \over \partial \vec B}-4\pi \vec M~,
153: \label{h-b} \end{equation}
154: we see that the magnetization curve of a superconducting ferromagnet
155: is described by
156: $B=B_0(|\vec H +4\pi
157: \vec M|)$ where $B_0(H)$ is the equilibrium magnetization curve for
158: a nonmagnetic type II superconductor \cite{SF} (Fig. \ref{fig2-1}a).
159: Note that in this relation the magnetic field $\vec H $ has a
160: different physical meaning from that used in the
161: Meissner state. For the Meissner state we introduced
162: $\vec H =\vec B -4\pi \vec M$, where the moment $\vec M$ originates
163: from ``molecular'' currents responsible for ferromagnetism, the
164: superconducting currents being treated as external currents. In the
165: mixed state, which is considered now, it is more convenient to
166: define the magnetic field as
167: $\vec H =\vec B -4\pi (\vec M +
168: \vec M_s)$, i.e. the definition includes also the diamagnetic moment
169: $\vec M_s=(\vec B_0-\vec H)/4\pi$ of the superconducting currents
170: circulating around vortex lines in the mixed state. Thus these
171: currents are treated in the same manner as molecular currents
172: responsible for ferromagnetism.
173:
174: Figure \ref{fig2-1}b shows that in a superconducting ferromagnet the
175: Meissner state ($B=0$) exists until
176: $H+4\pi M < H_{c1}$, where $H_{c1}= (\Phi_0 / \lambda ^2)\ln
177: (\lambda/\xi)$ is the lower critical field in a nonmagnetic superconductor and $\xi$ is the
178: coherence length, which determines the vortex core size. So ferromagnetism
179: decreases the lower critical field $\tilde H_{c1}= H_{c1}-4\pi M$. If
180: $4\pi M > H_{c1}$, the Meissner state is absent (Fig.
181: \ref{fig2-1}c) and the superconducting ferromagnet is in the mixed state
182: with vortices penetrating into it even in zero external field $H=0$. This
183: is {\em spontaneous vortex phase} with nonzero magnetic induction $B=
184: B_0(4\pi M)$ in the bulk.
185:
186:
187: Now let us consider formation of the domain structure in the standard
188: geometry \cite{LL}: a slab of the thickness $d$ along the anisotropy
189: easy axis $y$ and infinite in directions of the axes $x$ and $z$ (Fig.
190: \ref{fig2}). We start from a normal ferromagnet. In the absence of an
191: external magnetic field the average magnetic induction inside the slab
192: must vanish. Therefore, $B=0$ in a single-domain structure (Fig.
193: \ref{fig2}a), and there exists an uniform magnetostatic field $\vec H
194: =-4\pi \vec M$ in the entire sample, an analog of the electrostatic field
195: in a charged plane capacitor. This results in a high magnetostatic energy
196: $H^2/8\pi
197: \sim M^2$. However, the domain structure with period $l$ ($l \ll d$)
198: suppresses this energy in the domain bulk: $\vec H \approx 0$ and
199: $\vec B =-4\pi \vec M$, except for the area $\sim l^2$ near the sample
200: boundary (Fig.
201: \ref{fig2}b). But the average induction still vanishes, since $\vec M$
202: changes its sign from a domain to a domain. For the stripe
203: structure one can solve the equations of magnetostatics, $\vec \nabla
204: \times \vec H =0$ and $\vec \nabla \cdot \vec H =4\pi \rho_M$, exactly
205: \cite{LL,S}. Here $\rho_M =-\vec \nabla \cdot \vec M$ is the magnetic
206: charge. The magnetostatic energy per unit volume
207: of the slab is
208: \begin{equation}
209: E_s =0.852 M^2 l^2\times {1\over ld} = 0.852 M^2 {l\over d} ~.
210: \end{equation}
211: This energy is by a factor $l/d$ less than the magnetostatic energy in
212: a single-domain structure. However, the domain walls increase the energy.
213: The energy
214: of one domain wall (per unit length along the slab) is $\alpha K\delta
215: d$, where $\delta$ is the wall thickness and the numerical factor
216: $\alpha$ depends on the detailed definition of $K$
217: and $\delta$. Its specification is not essential for the
218: present analysis. The domain-wall energy per unit volume of the sample is
219: \begin{equation}
220: E_w =\alpha Kd\delta \times {1\over ld}= \alpha K{\delta \over l}~.
221: \end{equation}
222: The equilibrium value of the period $l$ is determined
223: by minimization of the energy $E_w +E_s$ \cite{LL}:
224: \begin{equation}
225: l = \sqrt{{\alpha K\over 0.852 M^2}\delta d}~.
226: \label{norm-l} \end{equation}
227:
228: Let us return back to a superconducting ferromagnet. In the Meissner state
229: the magnetic induction must vanish in the bulk, which is compatible
230: only with the single-domain structure. Thus {\em the equilibrium domain
231: structure is impossible in the Meissner state}. However, domains with
232: the changing direction of $\vec M$ can
233: appear in the spontaneous vortex phase with nonzero $B= B_0(4\pi M)$.
234: Like in a normal ferromagnet, the magnetic flux
235: $\propto B$ in domains should produce the magnetostatic fields in the
236: area $\sim l^2$, but in a
237: superconducting ferromagnet these fields are by the factor
238: $B_0(4\pi M)/4\pi M$ smaller. We can take it into account introducing
239: the effective magnetization $\tilde M = B_0(4\pi M)/4\pi$. Then
240: the period of the domain structure is given by Eq.
241: (\ref{norm-l}), where $M$ must be replaced with $\tilde M$.
242: In the limit of large $4\pi M \gg H_{c1}$, one has $\tilde M \rightarrow
243: M$ and the effect of superconductivity on the domain structure vanishes.
244: In the opposite limit of small $M$, when $4\pi M \rightarrow H_{c1}$,
245: $\tilde M$ vanishes and the period $l$ becomes infinite, as it
246: should be in the Meissner state $4\pi M < H_{c1}$.
247: However, this calculation of $l$ assumes that the penetration of the
248: magnetostatic field into a superconducting ferromagnet is similar to the
249: penetration into a normal ferromagnet. The assumption is correct if
250: rigidity of the vortex array is negligible and the effective penetration
251: depth is infinite. We can also consider the opposite limit of a very
252: rigid vortex array, when the magnetostatic fields penetrate only into the
253: layer of the thickness $\lambda$. If $\lambda \ll l$, the
254: penetration of the magnetic flux into a superconductor becomes
255: insignificant. This increases the magnetic fields outside the sample,
256: as well as the total magnetostatic energy, by a factor of 2, while
257: the correspondingly period $l$ decreases by a factor of
258: $\sqrt{2}$ (Fig. \ref{fig2}c), in analogy with the effect of a
259: superconducting substrate on a domain size in a ferromagnetic slab
260: \cite{S}. Thus avoiding a detailed analysis of the vortex and field
261: pattern in the domains close to the sample border we lose only a
262: numerical factor of not more than
263: $\sqrt{2}$. In any case, superconductivity, which coexists with
264: ferromagnetism in the same bulk, always increases the domain size, in
265: contrast to the superconductor-ferromagnet bilayer, where
266: superconductivity shrinks ferromagnetic domains \cite{S}.
267:
268: The absence of the equilibrium domain structure in the Meissner state
269: does not rule out a possibility of {\em metastable} domain walls, as
270: topologically stable planar defects. Domains can appear also
271: because of disorder, or grain structure. The structure of the domain
272: wall should be found by solution of the coupled equations of
273: magnetostatics and the London electrodynamics. We restrict ourselves to
274: the simplest case, when the London penetration depth $\lambda$
275: essentially exceeds the domain wall thickness $\delta$. This means that
276: at the spatial scales of order $\delta$ the domain-wall structure is
277: governed by large energies [the exchange energy and the anisotropy
278: energy, see Eq.~(\ref{F-tot})] and is not affected by the magnetostatic
279: and kinetic energy. On the other hand, at scales $\sim \lambda$ one can
280: find the distribution of $\vec B$ and $\vec H$ from the London equation
281: at constant $\vec M$. This is shown for the Bloch domain wall
282: (the magnetization $\vec M$ rotates in the plane of the wall and does not
283: produce the magnetostatic charges) in Fig. \ref{fig3}a. Though
284: our picture corresponds to a 180$^\circ$ wall, a similar
285: picture is expected for any domain wall. The jump of the tangential
286: component of the moment $\vec M$ at the wall defines the current sheet,
287: responsible for a jump of the magnetic induction parallel to the wall,
288: whereas a possible jump of the normal component of $\vec M$ (a
289: ``charged'' domain wall) would produce a jump of the normal component of
290: the field $\vec H$.
291:
292: The magnetic flux on the opposite sides from the domain
293: wall creates the magnetostatic fields outside the sample, where the
294: wall meets the sample boundary (Fig. \ref{fig3}b). The magnetic fluxes, which exit from
295: the sample at two sides from the wall, are equal in magnitude
296: ($4\pi M \lambda $ per unit length along the wall) but opposite in
297: direction. The magnetostatic field from
298: domain walls could be used for their experimental detection. At
299: distances $r \gg \lambda$ from a line, where the wall exits to the
300: sample boundary, this field is a dipole field of
301: the order of
302: $M\lambda^2 /r^2$.
303:
304: In summary, the letter presents the first analysis of the domain
305: structure in superconducting ferromagnets. There is no equilibrium
306: domain structure in a superconducting ferromagnet in the Meissner
307: state. In the spontaneous vortex phase the period of the domain structure
308: may essentially exceed that in the normal ferromagnet. But metastable
309: domain walls can exist even in the Meissner state. They generate the
310: magnetic flux in layers of a thickness
311: $\lambda$, which can be revealed by magneto-optical methodic.
312:
313: The author thanks Yu. Barash, I. Felner, and N. Kopnin for useful
314: discussions. The work has been supported by the grant of the Israel
315: Academy of Sciences and Humanities.
316:
317:
318: \begin{references}
319: \bibitem{felner} I. Felner, U. Asaf, Y. Levi, and O.~Milo, Phys. Rev.
320: B {\bf 55}, 3374 (1997).
321: \bibitem{Nat1} M.R. Eskildsen {\sl et al.}, Nature
322: {\bf 393}, 242 (1998).
323: \bibitem{Nat2}D. Aoki {\sl et al.},
324: Nature {\bf 413}, 613 (2001).
325: \bibitem{Chu} C.W. Chu {\sl et al.}, Journ. of
326: Superconductivity {\bf 13}, 679 (2000).
327: \bibitem{Tallon} C. Bernard, J.L. Tallon, E.
328: Brucher, and R.K. Kremer, Phys. Rev. B {\bf 61}, R14960 (2000).
329: \bibitem{GBV} H.S. Greenside, E.I. Blount, and C.M. Varma, Phys.
330: Rev. Lett. {\bf 46}, 49 (1981).
331: \bibitem{SF} E.B.~Sonin and I. Felner, Phys. Rev. B {\bf 57},
332: R14000 (1998).
333: \bibitem{BulBuz} L.N. Bulaevskii, A.I. Buzdin, M.L. Kulic, and S.V.
334: Panyukov, Adv. Phys. {\bf 34}, 176 (1985).
335: \bibitem{LL} L.D. Landau and E.M. Lifshitz, {\sl Electrodynamics of Continuous
336: Media} (Pergamon Press, Oxford, 1984).
337: \bibitem{S} E.B.~Sonin, cond-mat/0102102.
338: \end{references}
339:
340: \begin{figure}[!b]
341: \begin{center}
342: \leavevmode
343: \includegraphics[width=0.9\linewidth]{FigDS1.eps}
344: \bigskip
345: \caption{Magnetic induction $B$ (solid line), magnetic field $H$
346: (dashed line), and $4\pi M$ (dotted line) at the boundary between
347: a superconducting ferromagnet ($x<0$) and vacuum ($x>0$). }
348: \label{fig1}
349: \end{center}
350: \end{figure}
351:
352:
353: \begin{figure}%[!h]
354: \begin{center}
355: \leavevmode
356: \includegraphics[width=0.9\linewidth]{FigDS2.eps}
357: \bigskip
358: \caption{Magnetization curve: a) nonmagnetic type-II
359: superconductor; b) superconducting ferromagnet, $4\pi M < H_{c1}$;
360: c) superconducting ferromagnet, $4\pi M > H_{c1}$. }
361: \label{fig2-1}
362: \end{center}
363: \end{figure}
364:
365:
366:
367:
368: \begin{figure}%[!h]
369: \begin{center}
370: \leavevmode
371: \includegraphics[width=0.9\linewidth]{FigDS3.eps}
372: \bigskip
373: \caption{Domain structure in normal and superconducting
374: ferromagnets. The thick arrows show directions of the magnetic
375: moment $\vec M$, the thin lines with arrows are force lines of the
376: magnetostatic field $\vec H$. The magnetic charges are shown by + and
377: -. a) A single-domain structure. In the whole bulk $B=0$ and
378: $\vec H=-4\pi \vec M$. b) A stripe domain structure in a normal
379: ferromagnet. The magnetostatic fields are present in areas
380: $\sim l^2$ inside and outside the sample. In the rest parts of
381: domains $H=0$ and $\vec B=4\pi \vec M$. c) A superconducting ferromagnet
382: in the spontaneous vortex phase with a rigid vortex array. The
383: magnetostatic fields appear only in areas
384: $\sim l^2$ outside the sample. In the bulk of domains $H=0$
385: and
386: $B=B_0(4\pi M)$. }
387: \label{fig2}
388: \end{center}
389: \end{figure}
390:
391:
392: \begin{figure}%[!h]
393: \begin{center}
394: \leavevmode
395: \includegraphics[width=0.9\linewidth]{FigDS4.eps}
396: \bigskip
397: \caption{Domain wall in the Meissner state: a) Magnetic induction $B$ (solid line),
398: magnetic field $H$
399: (dashed line), and $4\pi M$ (dotted line) near the domain wall in the
400: superconducting ferromagnet. b) Magnetic flux lines around the exit of the domain wall
401: (of thickness $\delta$) to the sample surface. }
402: \label{fig3}
403: \end{center}
404: \end{figure}
405:
406:
407: \end{document}
408: \begin{equation}
409: \end{equation}
410: