1: %\documentclass[aps,prl,preprint,superscriptaddress,showpacs,showkeys]{revtex4}
2: \documentclass[prl,twocolumn,,superscriptaddress,showpacs]{revtex4}%
3:
4: \usepackage{amsfonts}%
5: \usepackage{amsmath}%
6: \usepackage{amssymb}%
7: \usepackage{graphicx}%
8:
9: \begin{document}
10:
11: \title{Topological Hysteresis in the Intermediate State of Type-I Superconductors}
12:
13: \author{Ruslan Prozorov}%
14: %\email{prozorov@mailaps.org}%
15: \affiliation{Ames Laboratory and Department of Physics \& Astronomy, Iowa State University, Ames,
16: Iowa 50011, U.S.A.}
17:
18: \author{Russell W. Giannetta}%
19: %\email{russg@uiuc.edu}%
20: \affiliation{Loomis Laboratory of Physics, University of Illinois at Urbana - Champaign, 1110 West
21: Green Street, Urbana, IL 61801, U.S.A.}
22:
23: \author{Anatolii A. Polyanskii}%
24: %\email{polyansk@engr.wisc.edu}%
25: \affiliation{Applied Superconductivity Center, University of Wisconsin, 1500 Engineering Drive,
26: Madison, WI 53706, U.S.A.}
27:
28: \author{Garry K. Perkins}%
29: \affiliation{Blackett Laboratory, Imperial College of Science
30: Technology and Medicine, London SW7 2BZ, United Kingdom}
31:
32: \keywords{type-I superconductor, intermediate state, hysteresis,
33: topology}
34:
35: \pacs{74.25.Ha,74.25.Op,89.75.Kd}
36:
37: \begin{abstract}
38: Magneto-optical imaging of thick stress-free lead samples reveals two distinct topologies of the
39: intermediate state. Flux tubes are formed upon magnetic field penetration (closed topology) and
40: laminar patterns appear upon flux exit (open topology). Two-dimensional distributions of shielding
41: currents were obtained by applying an efficient inversion scheme. Quantitative analysis of the
42: magnetic induction distribution and correlation with magnetization measurements indicate that
43: observed topological differences between the two phases are responsible for experimentally
44: observable magnetic hysteresis.
45: \end{abstract}
46:
47: \date{7 July 2005}
48:
49: \maketitle
50:
51: The structure of the intermediate state in type-I superconductors has a long history beginning with
52: the pioneering papers of Landau \cite{1,2} and continuing to the present day
53: \cite{3,4,5,6,7,8,9,10,11,12}. Intermediate state flux patterns closely resemble those found in a
54: wide variety of hydrodynamic, chemical and solid state systems \cite{9,9a,9b}. Study of the
55: intermediate state is therefore vital to a general understanding of pattern formation. Flux
56: structures can be readily tuned with a magnetic field and imaged with magneto-optical (MO)
57: techniques \cite{3,4}. The observed patterns can then be correlated with underlying thermal,
58: magnetic and resistive properties. Early MO images of the intermediate state revealed a variety of
59: phenomena not predicted by the simple theory \cite{3,4}. The initial models were then refined to
60: include domain branching and corrugation. Still, it is widely believed that a true
61: thermodynamically stable configuration of the intermediate state is the famous Landau laminar
62: structure \cite{2}.
63:
64: Magnetic hysteresis is routinely observed in type-I superconductors and has generally been
65: attributed to impurities, grain boundaries, dislocations and other imperfections of the crystal
66: structure \cite{4}. In this Letter we focus on the relationship between the topology of flux
67: structures and their macroscopic magnetic properties. We find that a small residual hysteresis
68: remains even in the most carefully prepared samples and present evidence that this hysteresis
69: arises from the differences in the topology of the intermediate state between flux entry and flux
70: exit.
71:
72: %
73: \begin{figure}[tb]
74: \centerline{\includegraphics[width=8cm]{fig1.eps}}%
75: \caption{Magnetization loops in stressed (top) and stress-free sample (bottom). Also shown MO
76: images obtained at magnetic fields indicated by arrows (
77: the numbers indicate images in succession).}%
78: \label{fig1}%
79: \end{figure}
80: %
81:
82: Samples were prepared from 99.9999{\%} lead \footnote{various sources, including
83: \emph{Puratronic}$^\circledR$ from \emph{Alfa Aesar}} foils and rods. More than a dozen
84: samples were prepared by using various annealing protocols or deliberately introducing
85: stress by cold rolling. The most reversible samples were obtained by melting lead between
86: two \textit{Pyrex} slides. Samples had thickness between $d=$ 0.1 and 1.5 mm and were
87: about $1.5\times 1.5\;\mathrm{mm}^\mathrm{2}$ in planar dimension. The topological
88: features described here were thickness independent above $d\approx 0.5\,\mathrm{mm}$,
89: which indicates that they are not due to surface-related effects. We show data for
90: samples which had demagnetization factors of about $N=0.5$ (determined both from initial
91: magnetisation and direct calculations \cite{20}.) \textit{Quantum Design} MPMS
92: magnetometer was used for DC magnetization measurements. MO imaging was performed in a
93: pumped flow-type optical $^{4}$He cryostat using Faraday rotation of polarized light in
94: Bi-doped iron-garnet films with in-plane magnetization \cite{14}. In all images the
95: bright regions correspond to the normal state and dark regions to the superconducting
96: state.
97:
98: %
99: \begin{figure}[tb]
100: \centerline{\includegraphics[width=8cm]{fig2.eps}}%
101: \caption{Magnetization loop in a stress-free sample accompanied by MO and zfc-fc measurements (see
102: text). Circles are obtained after applying field after cooling in zero field. Squares show results
103: of field-cooled measurements.}%
104: \label{fig2}%
105: \end{figure}
106: %
107:
108: Figure \ref{fig1} shows typical magnetization loops for the cold - rolled sample (top
109: panel) and the most reversible sample of similar dimensions (bottom). The stressed sample
110: shows considerably more hysteresis than the stress-free sample. The hysteresis increases
111: for decreasing field in the stressed sample as expected from pinning, whereas the
112: hysteresis disappears approaching $H=0$ in the reversible sample, indicating complete
113: Meissner expulsion. In that sample pinning is also absent at larger fields as well. Also
114: shown in Fig.\ref{fig1} are MO images taken at the same temperature and magnetic fields
115: indicated by arrows. The flux structure for the stressed sample is dendritic and a
116: significant amount of flux is trapped at $H=0$. In the stress-free sample the patterns
117: are noticeably different, revealing flux tube phase upon flux penetration and
118: well-defined laminar pattern upon flux exit. This behavior was observed at all accessible
119: temperatures.
120:
121: Figure \ref{fig2} shows details of the evolution of flux patterns in the stress-free
122: sample at $T=6$ K. After cooling in zero field, a full magnetization loop was measured
123: with a maximum magnetic field exceeding $H_c\approx 260$ Oe. Shown by the solid line, the
124: $M(H)$ loop exhibits magnetic hysteresis at intermediate fields. The crucial question is
125: whether this hysteresis is due to extrinsic factors (defects or residual stress) or it is
126: an intrinsic property of the intermediate state. To clarify this, we performed zero-field
127: and field - cooling experiments. The results are shown by the symbols in Fig.~\ref{fig2}.
128: The circles are obtained after cooling in zero field to 6 K and then increasing field,
129: whereas squares indicate measurements after cooling the sample to 6 K in a particular
130: field. If the hysteresis were due to pinning, the zero field cooled circles should
131: coincide with the ascending branch of the $M\left( H \right)$ loop but field-cooled
132: squares should not. Instead, Fig.\ref{fig2} clearly shows that both circles and squares
133: coincide exactly with the directly measured magnetization loop, implying that the
134: hysteresis is not due to pinning. The fact that the ascending and descending branches
135: merge at small and large fields is also inconsistent with pinning. Also, we observed no
136: magnetic relaxation with either flux penetration or flux exit. These results strongly
137: suggest that the hysteresis in stress-free samples is due to the topological difference
138: between the closed flux tube phase and the open laminar phase.
139:
140: MO images shown in Fig.~\ref{fig2} reveal that after pure Meissner screening, the
141: intermediate state appears, not as laminae, but as an assembly of normal tubes carrying
142: magnetic flux and separated by superconducting regions. (MO image $\sharp 1$ shows the
143: entire sample, others zoom in to reveal the structure). These images show that flux tubes
144: have a variety of structures - from simple monodomain to complex objects threaded with
145: superconducting tubes. The flux tube phase favors hexagonal symmetry, almost exactly as
146: modeled by Goren and Tinkham \cite{15}. In all cases, this tubular phase has a closed
147: topology that allows screening currents to circulate \cite{4}. Similar patterns were
148: directly observed in In \cite{3,8}, Re \cite{3}, Sn \cite{3,21} and Hg \cite{4} and it
149: seems that closed topology tubular pattern is a generic feature of the intermediate state
150: of pinning-free type-I superconductors upon flux penetration.
151:
152: Another possibility for the hysteresis is the edge barrier for flux penetration (including both,
153: Bean-Livingston and geometric barriers) \cite{17,18,19}. However, such a barrier would result in
154: delayed flux penetration and more negative values of magnetization compared to the thermodynamic
155: values. Figure \ref{fig2} shows that in our samples penetration occurs at the thermodynamic field
156: $H_c \left( {1-N} \right) \approx 120$ Oe and magnetization at that point is as supposed to be in
157: thermodynamic equilibrium, $4\pi M=-H_c $, independent of the demagnetization factor. We attribute
158: the weak influence of the edge barrier to the large thickness of our samples. On the other hand,
159: the edge barrier could be involved in the \emph{formation} of the observed topologies. When small
160: fingers of the normal phase are formed at the sample edge, the surface barrier will prevent their
161: continuous penetration into the interior and will break them into small flux tubes as suggested in
162: Ref.\cite{17}.
163:
164: %
165: \begin{figure}[tb]
166: \centerline{\includegraphics[width=8cm]{fig3.eps}}%
167: \caption{$T=5$ K, $H=100$ Oe: (\textbf{left}) Distribution of the magnetic induction upon flux
168: penetration (top) and exit (bottom);(\textbf{right}) corresponding patterns of shielding currents
169: density obtained by numerical inversion. Intensity is proportional to the current density. (Color online)}%
170: \label{fig4}%
171: \end{figure}
172: %
173:
174: Furthermore, quantitative imaging of the magnetic induction can be used to visualize
175: spatial distribution of shielding currents. This experimental information is important
176: for theoretical analysis involving current-loop models \cite{10,11} as well as for
177: general understanding of pattern formation in type-I superconductors. We used recently
178: developed fast inversion scheme \cite{perkins}. Figure \ref{fig4} shows the result
179: obtained for $H=100$ Oe. Left images correspond to flux penetration (top) and flux exit
180: (bottom) in our most reversible sample. The right panel shows corresponding distributions
181: of the shielding currents density (proportional to the brightness). Clearly, current
182: distribution exhibits two topologically distinct patterns. The closed topology of small
183: current loops vs. open topology of branched current streams.
184:
185: %
186: \begin{figure}[tb]
187: \centerline{\includegraphics[width=8cm]{fig4.eps}}%
188: \caption{$T=5$ K, $H=300$ Oe: (\textbf{left}) Distribution of the magnetic induction upon flux
189: penetration (top) and exit (bottom);(\textbf{right}) corresponding contour plots of shielding
190: currents density obtained by the numerical inversion. Arrows show the direction of currents. (Color online)}%
191: \label{fig5}%
192: \end{figure}
193: %
194:
195: The direction of the currents flow is seen in Fig.\ref{fig5}, which shows the reconstruction at
196: $H=300$ Oe. At this field the features are larger and contour lines with directional arrows can be
197: used to better visualize the flow patterns. In both topologies, currents flow counterclockwise -
198: against the direction of Meissner currents flowing along the sample edges. At low fields Meissner
199: currents dominate and hysteresis is negligible. It appears only at intermediate fields when the
200: intermediate state consists of mobile ensemble of flux tubes on field entry and laminar structure
201: that forms escape paths for flux expelled by the Meissner effect upon flux exit.
202:
203: To evaluate quantitative correspondence of the MO and $M\left(H\right)$ measurements we calculate
204: total magnetic moment from the MO images by using $4\pi
205: M=\int{\left[\textbf{H}-\textbf{B}(\textbf{r})\right]}d^3\textbf{r}$. Magnetic induction is
206: linearly proportional to the intensity, hence integrating images and using initial slope of the
207: measured $M(H)$ loop for calibration, $M$ is obtained. Figure \ref{fig3} shows that $M(H)$ loop
208: from the MO images (solid symbols) is in a good agreement with the direct measurements (open
209: symbols).
210:
211: %
212: \begin{figure}[tb]
213: \centerline{\includegraphics[width=7.5cm]{fig5.eps}}%
214: \caption{Comparison of DC magnetization measured by magnetometer and reconstructed from MO images.
215: Arrows show that at the minimum $4\pi M=-H_c$ as expected at thermodynamic equilibrium
216: without edge barriers.}%
217: \label{fig3}%
218: \end{figure}
219: %
220:
221: Observed topological hysteresis is also clearly seen on the profiles of the magnetic
222: induction. Following Landau \cite{2}, it has generally been assumed that the magnetic
223: field inside the normal phase in the intermediate state is close to $H_{c }$
224: \cite{3,4,9,17}. Figure \ref{fig3} shows profiles of the magnetic induction measured at
225: the same external field in the flux tube phase (obtained on flux entry) and in a laminar
226: phase (obtained on flux exit). While the field in the laminae is comparable to the
227: critical field $H_{c}$ as expected, the field above the flux tubes is much smaller.
228: Indeed, the measurements are carried out $\approx 10 \mu$m above the sample, so the
229: measured field is reduced compared to the values inside the tubes. However, simple
230: numerical analysis with appropriate dimensions does not reproduce such substantial
231: reduction. It is possible that flux tubes widen when approaching the surface \cite{3},
232: but why it is not seen in the laminar phase?
233:
234: Alternatively, it is possible that when a flux tube initially appears at the sample edge
235: with critical magnetic field inside. Due to closed topology, the total magnetic flux in
236: such tube is now conserved. The nucleated flux tube is driven by Meissner currents toward
237: sample interior (these currents flow everywhere on the surfaces perpendicular to the
238: magnetic field \cite{20}) until it is stopped at the center or later by other tubes
239: piling up from the center outward. When the flux tube reaches the interior, its radius
240: may increase to minimize the magnetic field energy. Real time imaging showed that flux
241: tubes produced at the sample edge continue to travel toward the center and form an
242: apparently outwardly expanding phase \cite{21}, as seen in the lower-right panel of
243: Fig.\ref{fig6}. The real-time observations were first made by Solomon and Harris in 1971
244: \cite{21}. The rigorous evaluation of the free energy of even a single tube is not
245: simple. The difficulty is the lack of a sharp interface between the superconductor and
246: tube interior, which decreases the surface energy. The tubes repel each other at the
247: large distances due to interaction of screening currents, but they attract each other
248: when their "cores" overlap. When tubes merge, average magnetic field in the tube
249: increases until it reaches $H_c$. At this stage, the honeycomb lattice is formed and it
250: persists almost up to $H=H_c$. The observed topological hysteresis is observed only at
251: the stage when magnetic field inside the tubes is less than $H_c$ and before the
252: formation of rigid hexagonal lattice, Fig.\ref{fig2}. A detailed study of tube nucleation
253: and expansion is needed to quantify this issue.
254:
255: %
256: \begin{figure}[tb]
257: \centerline{\includegraphics[width=8cm]{fig6.eps}}%
258: \caption{Profiles of the magnetic induction measured at $T=4$ K in the flux tube phase
259: (flux penetration)and laminar phase (flux exit). Right panel shows corresponding
260: MO images.}%
261: \label{fig6}%
262: \end{figure}
263: %
264:
265: We thank J.~R.~Clem, A.~T.~Dorsey, T.~A.~Girard, N.~D.~Goldenfeld, R.~P.~Huebener,
266: R.~V.~Kohn, D.~C.~Larbalestier and V.~K.~Vlasko-Vlasov for useful discussions.
267:
268: \begin{thebibliography}{99}
269: \bibitem{1}L. D. Landau, Sov. Phys. JETP \textbf{7}, 371 (1937).
270:
271: \bibitem{2}L. Landau, Nature \textbf{141}, 688 (1938).
272:
273: \bibitem{3}J. D. Livingston and W. DeSorbo, in \textit{Superconductivity}, edited by R. D. Parks
274: (Marcel Dekker, Inc., New York, 1969), Vol. 2, p. 1235.
275:
276: \bibitem{4}R. P. Huebener, \textit{"Magnetic Flux Structures of Superconductors"} (Springer-Verlag, New-York, 2001).
277:
278: \bibitem{5}F. Liu, M. Mondello, and N. Goldenfeld, Phys. Rev. Lett. \textbf{66}, 3071 (1991).
279:
280: \bibitem{6}H. Frahm, S. Ullah, and A. T. Dorsey, Phys. Rev. Lett. \textbf{66}, 3067 (1991).
281:
282: \bibitem{7}C. R. Reisin and S. G. Lipson, Phys. Rev. B \textbf{61}, 4251 (2000).
283:
284: \bibitem{8}V. Jeudy, C. Gourdon, and T. Okada, Phys. Rev. Lett. \textbf{92}, 147001 (2004).
285:
286: \bibitem{9}R. Choksi, R. V. Kohn, and F. Otto, J. Nonlinear Science \textbf{14}, 119 (2004).
287:
288: \bibitem{10}R. E. Goldstein, D. P. Jackson, and A. T. Dorsey, Phys. Rev. Lett. \textbf{76}, 3818 (1996).
289:
290: \bibitem{11}A. T. Dorsey and R. E. Goldstein, Phys. Rev. B \textbf{57}, 3058 (1998).
291:
292: \bibitem{12}O. Narayan, Phys. Rev. Lett. \textbf{81}, 5035 (1998).
293:
294: \bibitem{9a} P. Coullet and P. Huerre, "New Trends in Nonlinear Dynamics and Pattern Forming Phenomena", (Plenum Press, New York, 1991).
295:
296: \bibitem{9b} D. Walgraef, "Spatio-Temporal Pattern Formation", (Springer, New York, 1997).
297:
298: \bibitem{14}L.~A.~Dorosinskii \emph{et al.}, Physica C \textbf{203}, 149 (1992).
299:
300: \bibitem{15}R. N. Goren and M. Tinkham, J. Low. Temp. Phys. \textbf{5}, 465 (1971).
301:
302: \bibitem{17}A. Fortini and E. Paumier, Phys. Rev. B \textbf{14}, 55 (1976).
303:
304: \bibitem{18}J. R. Clem, R. P. Huebener, and D. E. Gallus, J. Low Temp. Phys. \textbf{12}, 449 (1973).
305:
306: \bibitem{19}H. Castro \emph{et al.}, Phys. Rev. B \textbf{59}, 596 (1999).
307:
308: \bibitem{20}R. Prozorov \emph{et al.}, Phys. Rev. B \textbf{62}, 115 (2000).
309:
310: \bibitem{perkins}G.~K.~Perkins, Yu.~V.~Bugoslavsky and A.~D.~Caplin, Supercond. Sci. Technol. \textbf{15}, 1140
311: (2002).
312:
313: \bibitem{21}P. R. Solomon and R. E. Harris, (United Aircraft Research Laboratories, (videotape provided by Dr. N. D. Goldenfeld), 1971).
314:
315: \end{thebibliography}
316:
317: \end{document}
318: