1: %\documentstyle[aps,prl,multicol]{revtex}
2: \documentstyle[aps,prl,multicol,psfig]{revtex}
3: %\documentstyle[pra,aps]{revtex}
4:
5:
6: %\documentstyle[aps,preprint]{revtex}
7:
8: \begin{document}
9: \draft
10: \title{Probing superconducting phase fluctuations from the current noise spectrum
11: of pseudogaped metal-superconductor tunnel junctions}
12:
13: \author{Xi Dai, Tao Xiang$^{\dagger *}$, Tai-Kai Ng and Zhao-bin Su$^\dagger$}
14:
15: \address{Physics Department, Hong Kong University of
16: Science and Technology,
17: Clear Water Bay, Kowloon, Hong Kong}
18:
19: \address{$^\dagger$Institute of Theoretical Physics, Chinese Academy
20: of Sciences, P.O. Box 2735, Beijing 100080, the People's Republic of China}
21:
22: \address{$^*$Isaac Newton Institute for Mathematical Sciences, the University
23: of Cambridge, 20 Clarkson Road, Cambridge CB3 0HE, U.K.}
24:
25: \date{\today}
26:
27: \maketitle
28:
29: \begin{abstract}
30: We study the current noise spectra of a tunnel junction of a metal with
31: strong pairing phase fluctuation and a superconductor. It is shown that
32: there is a characteristic peak in the noise spectrum at the intrinsic
33: Josephson frequency $\omega _J=2eV$ when $\omega _J$ is smaller than the
34: pairing gap but larger than the pairing scattering rate. In the presence of
35: an AC voltage, the tunnelling current noise shows a series of characteristic
36: peaks with increasing DC voltage. Experimental observation of these peaks
37: will give direct evidence of the pair fluctuation in the normal state of
38: high-$T_c$ superconductors and from the half width of the peaks the pair
39: decay rate can be estimated.
40: \end{abstract}
41:
42: \input{psfig.sty}
43:
44: \begin{multicols}{2}
45:
46:
47: Direct measurements of the pair susceptibility in the normal state of a
48: superconductor are important for understanding the pseudogap phenomenon in
49: high-$T_c$ superconductors. For conventional superconductors, the pairing
50: fluctuation exists only in a very narrow temperature range above $T_c$.
51: However, in the underdoped high-$T_c$ materials,
52: it is believed that the pair fluctuation is
53: important in a substantial temperature range in the normal state \cite
54: {Emery95}. This fluctuation is also important in superconductor thin films,
55: where long ranged phase coherence is strongly suppressed \cite{thin}. The
56: pseudogap effect was often interpreted as a precursor to the high-$T_c$
57: superconductivity. It was observed in many different experiments \cite
58: {nmr1,Mar,Loe,HD1,loram,opt,tun1}. However, what was really found in
59: experiments is the reduction of the low-lying density of states associated
60: with the opening of an energy gap, which may or may not be a pairing gap,
61: and a direct observation of the pair phase fluctuation is still absent.
62:
63: The pair susceptibility in the normal state of a underdoped cuprate can be
64: measured from the linear response of the system to an external pair field,
65: like the measurement of the magnetic susceptibility. Such an external pair
66: field can be introduced, for example, by coupling this underdoped cuprate to
67: an optimally doped cuprate with a higher superconducting transition
68: temperature via a tunnelling junction. When the temperature lies between the
69: superconducting transition temperatures of these two materials, the
70: tunnelling process will undoubtedly be strongly affected by the pair
71: fluctuation in the normal side\cite{pair}.
72:
73: The Josephson effect occurs when both sides of the junction are
74: in the superconducting state. A DC voltage across the junction leads to an
75: oscillating Josephson current at frequency $\omega _J=2eV$, where $V$ is the
76: applied voltage. If one side of the junction is not in the superconducting
77: phase, the Josephson current is zero on average. However, the Josephson
78: effect does not disappear completely. It manifests in the current noise
79: spectrum and, as will be shown later, the current noise is greatly enhanced
80: at the Josephson frequency $\omega _J$ if the pairing fluctuation is strong.
81:
82: In this paper, we present a theoretical analysis of the tunnelling current
83: noise spectrum of the pseudogaped metal-superconductor junction as
84: discussed above. We assume
85: that the pairing on both sides of the junction has $d_{x^2-y^2}$-wave
86: symmetry. The extension to other pairing symmetry cases is straightforward.
87: The main results we obtained are: (1) With an applied DC voltage, there is
88: a characteristic Josephson peak in the noise spectrum at the intrinsic
89: Josephson frequency $\omega _J$, although the coherent Josephson current is
90: zero. (2) When an AC voltage $V_s\cos \left( \omega _st\right) $ is applied
91: in addition to the DC voltage, there are a series of characteristic peaks in
92: the zero frequency noise centred at $\omega _J=n\omega _s$ ($n$ an integer)
93: with increasing $\omega _J$. (3) the half width of the peak is proportional
94: to the pair decay rate $\Gamma $ of the system, but the peak intensity
95: drops very quickly with increasing $\Gamma $.
96:
97: The Hamiltonian of the junction we study is defined by $H=H_S+H_N+H_T$ with
98: \begin{eqnarray}
99: H_S &=&\sum_{p\sigma }\varepsilon _pd_{p\sigma }^{+}d_{p\sigma
100: }+\sum_p\left( \Delta \varphi _pd_{p\uparrow }d_{-p\downarrow }+h.c.\right) ,
101: \label{super} \\
102: H_N &=&\sum_{k\sigma }\varepsilon _kc_{k\sigma }^{+}c_{k\sigma
103: }+\sum_{kq}\left[ \Psi (q,t)c_{-k+q/2,\uparrow
104: }c_{k+q/2,\downarrow }+h.c.\right] \nonumber \\
105: &&+\sum_q\chi _{{\rm pair}}^{-1}(q,t,t^{\prime })\Psi ^{*}(q,t)\Psi
106: (-q,t^{\prime }), \label{normal} \\
107: H_T &=&\sum_{kp\sigma }V_{kp}e^{i\int_0^teV_s\cos (\omega _st^{\prime
108: })dt^{\prime }+i\omega _Jt}c_{k\sigma }^{+}d_{p\sigma }+h.c., \label{tunnel}
109: \end{eqnarray}
110: where $\varphi _p=\cos p_x-\cos p_y$ and $\Psi (x,t)=\Delta _n\exp \left[
111: i\phi (x,t)\right] $ is the pairing field of the normal lead. $H_S$ is the
112: Hamiltonian for a $d$-wave superconductor. $H_N$ is the effective
113: Hamiltonian of a metal with pairing fluctuations \cite{pair,levin1} and $
114: \chi _{{\rm pair}}$ is the pair propagator (or pair susceptibility) in the
115: normal state. In $H_T$, $V_{kp}$ is the tunnelling matrix and can be expanded
116: with the crystal harmonics \cite{pair,hirs}. To the second order
117: approximation, it is given by
118: \begin{equation}
119: |V_{pk}|^2=V_0^2+V_1^2\varphi _k\varphi _p.
120: \end{equation}
121: The constant term in $|V_{pk}|^2$ has contribution to the single particle
122: tunnelling current, but no contribution to the pair tunnelling current because
123: of the $d$ -wave pairing symmetry. The integration resulted from the applied
124: AC voltage on the exponent in $H_T$ is difficult to handle analytically. In
125: practical calculations, it is often expanded as a sum of a serial of
126: oscillating terms with the Bessel function:
127: \[
128: e^{i\int_0^teV_s\cos (\omega _st^{\prime })dt^{\prime }}=\sum_nJ_n\left( {
129: \frac{{eV_s}}{{\omega _s}}}\right) e^{in\omega _st},
130: \]
131: where $J_n(x)$ is the n'th order Bessel function \cite{mahan}.
132:
133: Let us first consider phenomenologically
134: the effect of pairing phase fluctuation on the current
135: noise. We assume
136: that the phase fluctuation follows a diffusion process and the spatial
137: correlation length is very short.
138: In this case, the correlation function of the phase field is given
139: by
140: \begin{equation}
141: \left\langle e^{i\phi (x,t)}e^{-i\phi (x^{\prime },0)}\right\rangle \sim
142: e^{-\Gamma |t|/2}\delta(x-x'), \label{pair}
143: \end{equation}
144: where $\Gamma $ is the pair fluctuation rate. From this equation, it can be
145: shown that
146: \begin{equation}
147: \chi _{{\rm pair}}(q,\omega )=1/(i\omega +\Gamma /2).
148: \end{equation}
149:
150: The current noise spectrum measures the current fluctuation in the junction.
151: It is defined by the symmetrized current-current function:
152: \begin{equation}
153: S(\omega )\equiv \int dte^{i\omega t}\left\langle \left\{ I(t)-\left\langle
154: I(t)\right\rangle ,I(0)-\left\langle I(0)\right\rangle \right\}
155: \right\rangle , \label{sw}
156: \end{equation}
157: where $I(t)$ is the current across the junction and $\left\{ A,B\right\}
158: \equiv AB+BA$. In the case only a DC voltage is applied, the above tunnelling
159: Hamiltonian can be treated semi-classically and the current-current
160: correlation function contributed from the pair tunnelling is simply given by
161: the thermal average of the phase fields:
162: \begin{eqnarray}
163: \left\langle I(t)I(0)\right\rangle &\sim &\int dx dx'
164: \left\langle \sin \left[ \phi
165: (x,t)+\omega _Jt\right] \sin \phi (x^{\prime },0)\right\rangle
166: \nonumber \\
167: &\sim & \cos \left( \omega _Jt\right) e^{-\Gamma |t|/2}
168: \end{eqnarray}
169: Substituting this equation into (\ref{sw}), we then obtain the noise
170: spectrum contributed from the pair tunnelling
171: \begin{equation}
172: S_{pair}(\omega )\sim {\frac{{\Gamma }}{{(\omega -\omega _J)^2+\Gamma ^2/4}}}
173: +{\frac{{\Gamma }}{{(\omega +\omega _J)^2+\Gamma ^2/4}}.} \label{noise}
174: \end{equation}
175: This result indicates that the Josephson current
176: cross the junction is coherent if the tunnelling time is much shorter than
177: the fluctuation time $\tau \sim \Gamma ^{-1}$ of virtual cooper pairs. In
178: the limit $\Gamma \rightarrow 0$, $S(\omega )$ becomes two delta-functions
179: centred at $\pm {\omega _J}$. For finite $\Gamma $, these two
180: delta-functions are broadened and the half width of the broadened peak is
181: proportional to $\Gamma $.
182:
183: Eq. (\ref{noise}) is similar as the result obtained by Martin and Balatsky
184: very recently \cite{Ivar}. However, as both sides of the junction in the
185: experimental setup they suggested are in the normal state, the total pair
186: decay rate is larger than the case discussed here. Since the
187: intensity of the peak compared with the average noise drops sharply with
188: increasing ${\Gamma }$ (Figure 2), it may be
189: difficult to observe the
190: characteristic Josephson fluctuation in finite frequency conductance
191: measurement suggested in their paper. While in our paper,
192: to determine the pair decay rate only the
193: zero frequency noise is required which is easier to measure.
194:
195: In a s-wave pairing system, since no quasiparticle excitations
196: exist within the
197: energy gap, the single particle tunnelling can be ignored when the noise
198: frequency is lower than the gap. For the d-wave pairing case, the
199: contribution of the single particle tunnelling cannot be ignored because of
200: the existence of gap nodes. To find out the contribution from both the
201: quasiparticle and pair tunnelling to the noise, we
202: perform a microscopic calculation the
203: current-current correlation function using the closed time Green's function
204: technique\cite{Kel,chou}. The tunnelling Hamiltonian $H_T$ is treated as a
205: perturbation. Since the normal Josephson effect is absent in the junction,
206: the second order contribution of $H_T$ contains only the quasiparticle
207: tunnelling terms. However, the noise caused by the quasiparticle tunnelling
208: varies very slowly with frequency. When the noise frequency is smaller than $
209: \Delta $, it adds only a featureless background to the noise spectra.
210:
211:
212:
213: The most important contribution of $H_T$ to the noise is from the fourth
214: order perturbation, in particular from the virtual pair tunnelling process.
215: When the pairing fluctuation in the normal state is strong, a Cooper pair
216: can tunnel from the superconductor to the normal lead and contributes a
217: significant term to the noise at low frequencies. Figure 1 shows the Feymann
218: diagram for this pair tunnelling term in the current-current correlation
219: function. The solid line is the Green's function of electrons. The dashed
220: line is the propagator of the pair field in the normal state.
221:
222: \begin{figure}[htb]
223: \begin{center}
224: %\framebox[55mm]{\rule[-21mm]{0mm}{43mm}}
225: \psfig{file=fig1.eps,height=40mm,width=85mm,angle=0}
226: \end{center}
227: \caption{The Feymann diagram for the pair tunnelling term, where
228: the solid line represents the electronic Green's function and
229: the dashed line represents the pairing fluctuation.}
230: \label{Fig.1}
231: \end{figure}
232:
233: To the leading order approximation in both the quasiparticle and pair
234: tunnelling channels, we find that the noise spectra is given by
235: \begin{equation}
236: S(\omega )\approx N_q(\omega )+N_p(\omega )+N_q(-\omega )+N_p(-\omega ),
237: \end{equation}
238: where $N_q$ and $N_p$ are respectively the contributions from the
239: quasiparticle and pair tunnelling:
240:
241: \end{multicols}
242:
243: \begin{center}
244: --------------------------------------------------------------
245: \end{center}
246:
247: \begin{eqnarray}
248: N_q(\omega ) &=&{V_0^2}\sum_nJ_n^2(x)\rho _s\int d\nu \rho _d(\nu )\left[
249: n_f(\nu )-n_f(\nu +\omega _n^{\prime \prime })\right] \coth \frac{\beta
250: \omega _n^{\prime \prime }}2, \label{nq} \\
251: N_p(\omega ) &=&V_1^4C\sum_{n,n_1,n_2}\frac{
252: J_{n_1}(x)J_{n-n_1}(x)J_{n_2}(x)J_{n-n_2}(x)}{{\omega _n^{^{\prime
253: }2}+\Gamma ^2/4}}\coth \frac{\beta \omega _n^{\prime }}2 \nonumber\\
254: &&\sum_q\left\{ {\omega _n^{\prime }}\left[ \phi _2^2(\omega _n^{\prime
255: },q)-\phi _1^2(\omega _n^{\prime },q)\right] +{\Gamma }\phi _2(\omega
256: _n^{\prime })\phi _1(\omega _n^{\prime },q)\right\} . \label{np}
257: \end{eqnarray}
258:
259: \begin{center}
260: ---------------------------------------------------------------
261: \end{center}
262:
263: \begin{multicols}{2}
264:
265: In the above equations, $x=eV_s/\omega _s$, $\omega _n^{\prime }=\omega
266: +\omega _J+n\omega _s$, $\omega _n^{\prime \prime }=\omega +eV+n\omega _s$, $
267: C=-\sum_p{\Delta \varphi _p^2/}\left( {\varepsilon _p^2+\Delta ^2\varphi _p^2
268: }\right) $, $\rho _s$ is the density of states of the superconductor in the
269: normal state, and $\rho _n(\nu )$ is the density of states of the normal
270: lead. In Eq. (\ref{np}),
271: \begin{eqnarray*}
272: \phi _1(\nu ,q) &=&\pi \sum_k\varphi _k^2\left( 1-n_k-n_{-k+q}\right) \delta
273: (\nu -\varepsilon _k-\varepsilon _{-k+q}), \\
274: \phi _2(\nu ,q) &=&-\sum_k\varphi _k^2{\frac{{1-n_k-n_{-k+q}}}{{\ \nu
275: -\varepsilon _k-\varepsilon _{-k+q}}},}
276: \end{eqnarray*}
277: and $n_k=1/(1+e^{\beta \varepsilon _k})$ is the Fermi-Dirac function .
278:
279:
280: \begin{figure}[htb]
281: \begin{center}
282: %\framebox[55mm]{\rule[-21mm]{0mm}{43mm}}
283: \psfig{file=fig2.eps,height=70mm,width=80mm,angle=-90}
284: \end{center}
285: \caption{The normalised current noise spectrum as a
286: function of the noise frequency for three different $\Gamma$.
287: $\omega_J=0.4\Delta$.}
288: \label{Fig.2}
289: \end{figure}
290:
291:
292: Figure 2 shows $S(\omega )$ normalised by $S(0)$ as a function of $\omega $
293: for the case $\omega _J=0.4\Delta $ and $\omega _s=0$. The parameters used
294: are $\Delta _n=\Delta $, $T=0.2\Delta $, $V_0=2V_1=2\Delta $, and $\rho
295: _s=1/10\Delta $. The band width of the normal lead is assumed to be $
296: 10\Delta $. From the figure, it is clear that if the fluctuation rate $
297: \Gamma $ is sufficiently small, the noise peak centred at $\omega =\pm
298: \omega _J$ is very sharp. The half width of the peak is determined by $
299: \Gamma $. When $\omega _J\gg \Gamma $, the thermally excited pair tunnelling
300: current will oscillate a few times before it loses the phase coherence. But
301: with increasing $\Gamma $, the noise peak at $\omega _J$ becomes broadened
302: and vanishes eventually.
303:
304: \begin{figure}[htb]
305: \begin{center}
306: %\framebox[55mm]{\rule[-21mm]{0mm}{43mm}}
307: \psfig{file=fig3.eps,height=70mm,width=80mm,angle=-90}
308: \end{center}
309: \caption{The normalised current noise spectrum at zero noise
310: frequency as a function of the intrinsic
311: Josephson frequency $\omega_J=2eV$ for three different $\Gamma$.
312: $\omega_s = 0.4\Delta$. }
313: \label{Fig.3}
314: \end{figure}
315:
316:
317: In the presence of both DC and AC voltages, the pair tunnelling is enhanced
318: at the frequency $\omega =\omega _J+n\omega _s$ and the noise spectrum shows
319: a serial of local maxima at these frequencies. If the applied microwave
320: frequency and the Josephson frequency satisfy the condition $\omega
321: _J+n\omega _s=0$, a local maximum will appear at the zero frequency. Figure
322: 3 shows the zero frquency
323: normalised current noise spectrum as a function of $\omega _J$
324: for the case $\omega _s=0.4\Delta $ (the other parameters are the same as in
325: Figure 2). There are a series of peaks for $\Gamma =0.2\Delta $. When $
326: \Gamma $ is increased, these peaks becomes broader and weaker.
327:
328:
329: The above results indicate that the pair fluctuation in the normal
330: lead has strong effects on the noise spectrum of the tunnelling current.
331: The broadened peaks at the characteristic frequencies in the noise spectrum
332: shown in Figures 2 and 3 are unique to the normal-superconductor junction
333: with strong pair fluctuations. Experimentally, it is not easy
334: to measure the finite
335: frequency noise, but when the AC voltage is applied we only need
336: to measure the zero frequency noise which is quite easy to do.
337: Thus experimental measurements of the current
338: noise allow us to judge how strong the pairing fluctuation is in the pseudogap
339: phase. This is apparently important for a direct test of the superconducting
340: precursor scenario.
341:
342:
343: In summary, we have shown that the noise spectrum of the
344: tunnelling current between a underdoped cuprate in the normal state and
345: an optimally doped cuprate in the superconducting state is a useful
346: probe of the pairing fluctuation in the pseudogap phase.
347: We have calculated the noise spectrum using the closed path Green's
348: functions and found that the current noise is strongly enhanced at
349: a series of characteristic frequencies determined by the
350: Josephson pair tunnelling processes. Notice that the system is
351: nonequilibrium when the DC voltage is applied, it is proper to use
352: the closed time green's function technique.
353: Experimental observations of the noise peaks at these characteristic
354: frequencies will help us to establish the correct picture of the
355: pair fluctuation in the pseudogap phase of high-$T_c$
356: cuprates.
357:
358: TX acknowledges the financial
359: support from the National Natural Science Foundation of China and from the
360: special funds for Major State Basic Research Projects of China.
361:
362:
363: \begin{references}
364: \bibitem{Emery95} V. Emery and S. Kivelson, {Nature} {\bf 374}, 434 (1995).
365:
366: \bibitem{thin} D. B. Haviland, Y. Liu and A. M. Goldman, Phys. Rev. Lett.
367: {\bf 62}, 2180 (1989); A. M. Goldman and Y. Liu, Physica D {\bf 83}, 163
368: (1995).
369:
370: \bibitem{nmr1} M. Takigawa, A. P. Reyes, P. C. Hammel, J. D. Thompson, R.
371: H. Heffner, Z. Fisk, and K. C. Ott, Phys. Rev. B {\bf 43}, 247 (1991).
372:
373: \bibitem{Mar} D. S. Marshall, D. S. Dessau, A. G. Loeser, C-H. Park, A. Y.
374: Matsuura, J. N. Eckstein, I. Bozovic, P. Fournier, A. Kapitulnik, W. E.
375: Spicer, and Z.-X. Shen, Phys. Rev. Lett. {\bf 76}, 4841 (1996).
376:
377: \bibitem{Loe} A. G. Loeser, Z. X. Shen, D. S. Dessau, D. S. Marshall, C. H.
378: Park, P. Fournier, A. Kapitulnik, Science {\bf 273}, 325 (1996).
379:
380: \bibitem{HD1} H. Ding, T. Yokoya, J. C. Campuzano, T. Takahashi, M,
381: Randeria, M. R. Norman, T. Mochiku, K. Kadowaki and J. Giapintzakis, Nature
382: {\bf 382}, 51 (1996).
383:
384: \bibitem{loram} J. W. Loram, K. A. Mirza, J. R. Cooper, and W. Y. Liang,
385: Phys. Rev. Lett. {\bf 71}, 1740 (1993).
386:
387: \bibitem{opt} D. N. Basov, R. Liang, B. Dabrowski, D. A. Bonn, W. N. Hardy,
388: and T. Timusk, Phys. Rev. Lett. {\bf 77} 4090 (1996).
389:
390: \bibitem{tun1} Ch. Renner, B. Revaz, J. Y. Genond, K. Kadowak and Fischer,
391: Phsy. Rev. Lett. {\bf 80}, 149 (1998).
392:
393: \bibitem{pair} B. Janko, I. Kosztin, K. Levin, M. R. Norman and D. J.
394: Scalapino, Phys. Rev. Lett. {\bf 82} 4304 (1999).
395:
396: \bibitem{Ivar} Ivar Martin and Alexander Balatsky, cond-mat/0002043.
397:
398: \bibitem{levin1} B. Janko, J.Maly and K. Levin, Phys. Rev. {\bf B56},
399: R11407 (1997); J. Maly, B.Janko and K. Levin, Phys. Rev. {\bf B59}, 1354,
400: (1999).
401:
402: \bibitem{mahan} G. D. Mahan, ``Many-Particle Physics'', Chapter {\bf 9 },
403: Plenum Press,New York,(1990).
404:
405: \bibitem{hirs} P. J. Hirschfeld, S. M. Quinlan and D. J. Scalapino, Phys.
406: Rev. {\bf B55}, 12742 (1997).
407:
408: \bibitem{Kel} L. V. Keldysh, JETP {\bf 20,} 1018 (1965).
409:
410: \bibitem{chou} K. C. Chou, Z. B. Su, B. L. Hao, and L. Yu, Phys. Rep. {\bf
411: 118}, 1 (1985).
412: \end{references}
413:
414:
415:
416: \end{multicols}
417:
418: \end{document}
419: