cond-mat0510065/smi.tex
1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: \documentstyle[aps]{revtex}
3: %----------------------------------------------------------------
4: \begin{document}
5: \draft
6: \title{
7: Superconductor-Metal-Insulator Crossover 
8: in Disordered Thin Films
9: \footnote{This paper has been submitted to 
10: the Short Note section of J. Phys. Soc. Jpn.}
11: }
12: \author{
13: O. Narikiyo
14: }
15: \address{
16: Department of Physics, 
17: Kyushu University, 
18: Fukuoka 810-8560, 
19: Japan
20: }
21: 
22: \vskip 10pt
23: 
24: \date{
25: Oct. 4, 2005
26: }
27: 
28: \vskip 10pt
29: 
30: \maketitle
31: %----------------------------------------------------------------
32: \begin{abstract}
33: Superconductor-Metal-Insulator crossover 
34: in disordered thin films 
35: is discussed on the basis of a random 
36: junction-network model. 
37: \vskip 10pt
38: \noindent
39: Key Words: {\bf superconductor-metal-insulator crossover, 
40: disordered thin films, random resister network, 
41: Josephson junction network}
42: \end{abstract}
43: %----------------------------------------------------------------
44: \vskip 30pt
45: 
46: 
47: The superconductor-insulator (S-I) transition in thin films 
48: is one of the typical phenomena of quantum phase transition.~\cite{RMP} 
49: Recently a metallic phase 
50: intervening between superconducting and insulating phases 
51: has been drawing attention.~\cite{PD} 
52: Although several theories~\cite{PD} 
53: modifying the idealized theory~\cite{RMP} for the S-I transition 
54: have been proposed, 
55: present understanding of the metallic phase 
56: is far from conclusive. 
57: In this Short Note we also propose a simple phenomenological model 
58: and try to understand the temperature dependence of the resistance 
59: from which we judge the phase at zero temperature, 
60: superconductor (S) or metal (M) or insulator (I). 
61: 
62: Our phenomenological model is a combination of two models. 
63: One is the random Josephson-junction-network model~\cite{Dynes} 
64: for disordered superconductors 
65: and the other is the random resister-network model~\cite{Meir} 
66: for disordered metals or insulators. 
67: We consider a network 
68: consisting of superconducting islands 
69: linked by two types of junctions, J (Josephson) and N (normal). 
70: At the type-J junction 
71: the current is carried by Cooper pairs, 
72: while at the type-N junction 
73: by normal electrons. 
74: The type-N junction in our model is expected 
75: for the case where the distance between islands is so large 
76: that the Cooper pairs are broken in the junction. 
77: In real materials 
78: the type-N junction is also expected 
79: for the case where the size of the island is so small 
80: that the Cooper pairs are not formed there. 
81: In our model 
82: only the resistance caused by the junctions 
83: is taken into account 
84: and the resistance drop 
85: at around the superconducting transition temperature $T_{\rm c}$ 
86: of the islands is neglected. 
87: In the following we set $\hbar = k_{\rm B} = 1$. 
88: 
89: At a type-J junction~\cite{Dynes} 
90: the resistance $R(T)$ as a function of the temperature $T$ 
91: is assumed to be given by the Ambegaokar-Halperin formula 
92: \begin{equation}
93: R(T) = R_{\rm N} / [ I_0(E_{\rm J}(T)/T) ]^2, 
94: \end{equation}
95: where $I_0(x)$ is the modified Bessel function of order $0$. 
96: The Josephson-coupling energy $E_{\rm J}(T)$ 
97: is given by the Ambegaokar-Baratoff form 
98: \begin{equation}
99: E_{\rm J}(T) = {1 \over 4} {R_0 \over R_{\rm N}} 
100:                \Delta(T) \tanh{\Delta(T) \over 2T}, 
101: \end{equation}
102: where $R_0$ is the quantum unit of the resistance, 
103: $R_0 = \pi / e^2$, 
104: and $\Delta(T)$ is the BCS gap function. 
105: This model is appropriate for relatively thick films 
106: where contacts among islands are relatively strong. 
107: Since the normal resistance $R_{\rm N}$ between islands 
108: saturates at low temperatures 
109: and exhibits only weak temperature dependence 
110: for such a strong contact, 
111: we neglect the temperature dependence of $R_{\rm N}$ 
112: in accordance with the former study.~\cite{Dynes} 
113: In contrast to this good metallic behavior 
114: a poor metallic junction disscussed below 
115: exhibits some temperature dependence. 
116: 
117: At a type-N junction~\cite{Meir} 
118: the resistance is assumed to be given by the Landauer formula 
119: for quantum point contact as 
120: \begin{equation}
121: R(T) = R_0 \cdot [ 1 + \exp(E_{\rm c}/T) ], 
122: \end{equation}
123: where $E_{\rm c}$ is the threshold energy for the electron transmission 
124: measured from the chemical potential. 
125: The junction is insulating for $E_{\rm c} > 0$ 
126: and metallic for $E_{\rm c} <0$. 
127: This model is appropriate for relatively thin films 
128: where contacts among islands are relatively weak. 
129: For such a weak contact 
130: the resistance exhibits poor metallic or insulating temperature dependence. 
131: In Fig. 1 the temperature dependence of the resistance 
132: of a type-N junction is shown for a poor metallic case. 
133: While the resistance saturates at low temperatures, 
134: it exhibits almost exponential temperature dependence 
135: in some temperature range. 
136: 
137: In the following we consider the circuit 
138: consisting of 3 conductors connected in parallel. 
139: Each conductor consists of 10 composites 
140: of resistors connected in series. 
141: Each composite consists of 3 resistors 
142: connected in parallel. 
143: Each resistor is modeled by 
144: either type-J or type-N junction. 
145: Although circuits in general cannot be reduced 
146: to series-parallel combinations of resistors, 
147: we adopt this model, 3-10-3 circuit, for simplicity 
148: to discuss qualitative aspects of the resistance of a random network. 
149: 
150: In Fig. 2 
151: the temperature dependence of the resistance 
152: for 3-10-3 circuit is shown in the case 
153: where every resistor in the circuit is type-J junction. 
154: The values of $R_{\rm N}$ for junctions are randomly chosen. 
155: The resistance vanishes faster than single exponential function of $T$ 
156: as $T$ is decreased. 
157: This temperature dependence 
158: is similar to that obtained in the former study.~\cite{Dynes} 
159: 
160: In Fig. 3 
161: the temperature dependence of the resistance 
162: for 3-10-3 circuit is shown in the case 
163: where every resistor in the circuit is type-N junction. 
164: The values of $E_{\rm c}$ are randomly chosen. 
165: A crossover from metallic to insulating behavior is seen 
166: as $E_{\rm c}$ is increased. 
167: This crossover 
168: is similar to that obtained in the former study.~\cite{Meir} 
169: 
170: In Fig. 4 
171: the temperature dependence of the resistance 
172: for 3-10-3 circuit is shown in the case 
173: where a resistor is either superconducting or metallic. 
174: Almost exponential temperature dependence is seen 
175: reflecting the behavior of metallic junctions. 
176: This behavior is consistent with experiments, 
177: while it was unexplained in the former study~\cite{Dynes} 
178: where only superconducting junctions were taken into accounts. 
179: 
180: In experiments 
181: a S-M-I crossover~\cite{crossover} is observed 
182: when the film thickness is tuned. 
183: Such a crossover is also seen in Figs. 2-4. 
184: In our phenomenological analysis 
185: it is reduced to the nature of each junction 
186: and has nothing to do with a macroscopic phase transition. 
187: It should be also noted that the present metallic state is 
188: irrelevant to the recent issue~\cite{MIT} 
189: of the presence of a metallic state 
190: in two-dimensional disorderd systems. 
191: The issue corresponds to the presence of a coherence 
192: at macroscopic scale in uniform system after averaging. 
193: On the other hand, 
194: a metallic state is possible at mesoscopic scale. 
195: Our system is a mesoscopic one in terms of junctions, 
196: while superconducting islands are macroscopic objects. 
197: The system is highly heterogeneous 
198: where the coherence is maintained within the islands 
199: and the loss of the coherence occurs only at the junctions. 
200: 
201: The author is grateful to 
202: T. Kawaguti, B. Shinozaki and K. Makise 
203: at Kyushu University Ropponmatsu 
204: for valuable discussions. 
205: 
206: \vskip 30pt
207: 
208: %----------------------------------------------------------------
209: \begin{references}
210: \bibitem{RMP} 
211: S. L. Sondhi, S. M. Girvin, J. P. Carini and D. Shahar: 
212: Rev. Mod. Phys. \textbf{69} (1997) 315 for a review. 
213: \bibitem{PD} 
214: P. Phillips and D. Dalidovich: 
215: Science \textbf{302} (2003) 243 for a review. 
216: \bibitem{Dynes} 
217: A. Frydman, O. Naaman and R. C. Dynes: 
218: Phys. Rev. B \textbf{66} (2002) 052509. 
219: \bibitem{Meir} 
220: Y. Meir: 
221: Phys. Rev. Lett. \textbf{83} (1999) 3506. 
222: \bibitem{crossover}
223: C. Christiansen, L. M. Hernandez and A. M. Goldman: 
224: Phys. Rev. Lett. \textbf{88} (2002) 037004 for an example. 
225: \bibitem{MIT}
226: E. Abrahams, S. V. Kravchenko and M. P. Sarachik: 
227: Rev. Mod. Phys. \textbf{73} (2001) 251 for a review. 
228: \end{references}
229: %----------------------------------------------------------------
230: \vskip 15pt
231: %----------------------------------------------------------------
232: \begin{figure}
233: \caption{
234: Temperature dependence of the resistance 
235: for a single normal junction. 
236: The threshold energy is chosen as $E_{\rm c}/T_{\rm c}= -1.5$. 
237: }
238: \label{fig1}
239: \end{figure}
240: 
241: \vskip -15pt
242: 
243: \begin{figure}
244: \caption{
245: Temperature dependence of the resistance 
246: for a network of superconducting junctions. 
247: The normal resistances are randomly distributed in the range, 
248: $0.1 < R_{\rm N}/R_0 < 0.6$, for the bottom case. 
249: The range for the middle case is $0.2 < R_{\rm N}/R_0 < 0.7$ and 
250: $0.3 < R_{\rm N}/R_0 < 0.8$ for the top case. 
251: }
252: \label{fig2}
253: \end{figure}
254: 
255: \vskip -15pt
256: 
257: \begin{figure}
258: \caption{
259: Temperature dependence of the resistance 
260: for a network of normal junctions. 
261: The threshold energies are randomly distributed in the range, 
262: $-0.4 < E_{\rm c}/T_{\rm c} < 0.6$, for the top case. 
263: The range for the other cases is shifted as 
264: $-0.5 < E_{\rm c}/T_{\rm c} < 0.5$, 
265: $-0.6 < E_{\rm c}/T_{\rm c} < 0.4$, 
266: $-0.7 < E_{\rm c}/T_{\rm c} < 0.3$ and 
267: $-0.8 < E_{\rm c}/T_{\rm c} < 0.2$ in order (form top to bottom). 
268: }
269: \label{fig3}
270: \end{figure}
271: 
272: \vskip -15pt
273: 
274: \begin{figure}
275: \caption{
276: Temperature dependence of the resistance 
277: for a network of the random mixture of superconducting and normal junctions. 
278: The normal resistances for superconducting junctions 
279: are randomly distributed in the range 
280: $0.1 < R_{\rm N}/R_0 < 0.6$. 
281: The threshold energies for normal junctions 
282: are randomly distributed in the range 
283: $-1.5 < E_{\rm c}/T_{\rm c} < -1.0$. 
284: For the top case 3\% of the junctions are superconducting 
285: and 97\% are normal. 
286: For the middle case 5\% are superconducting. 
287: For the bottom case 7\% are superconducting. 
288: }
289: \label{fig4}
290: \end{figure}
291: %----------------------------------------------------------------
292: \end{document}
293: %----------------------------------------------------------------
294: \bye
295: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
296: