Current-phase relation of the SNS junction in a superconducting loop
Extracted source text ยท arXiv cond-mat/0406445
\documentclass{epl}
\usepackage{amssymb}
\title{Current-phase relation of the SNS junction in a superconducting loop}
\author{Mun Dae Kim\inst{1}\thanks{E-mail: \email{mdkim@kias.re.kr}} \and Jongbae Hong\inst{2}}
\institute{
\inst{1} Korea Institute for Advanced Study, Seoul 130-722, Korea\\
\inst{2} School of Physics, Seoul National University, Seoul 151-742, Korea
}
\pacs{74.72.Jt}{Other cuprates, including Tl and Hg-based cuprates}
\pacs{74.45.+c}{Proximity effects; Andreev effect; SN and SNS junctions}
\pacs{73.23.Ra}{Persistent currents}
\begin{document}
\maketitle
\begin{abstract}
We study the current-phase relation of the superconductor/normal/superconductor (SNS)
junction imbedded in a superconducting loop. Considering the current conservation and
free energy minimum conditions, we obtain the persistent currents of the SNS
loop. At finite temperature we can explain the experimentally observed highly
non-sinusoidal currents which have maxima near the zero external flux.
\end{abstract}
\section{Introduction}
Physically interesting superconducting loops with junctions are the
superconducting quantum interference device (SQUID) and the
superconductor/normal/superconductor loop with long normal sector (${\rm SN_{long}S}$ loop).
Both are mesoscopic size and
yield persistent currents when a magnetic flux threads the loop.
%These systems attract further interest recently
%in connection with promising current qubits for quantum computer.
%
The thickness of the Josephson junctions in SQUID loop is much smaller
than the superconducting coherence length, but the thickness of
normal segment $d$ of ${\rm SN_{long}S}$ loop can be larger than that.
%
The persistent current in the former system flows by tunneling
the Josephson junction, while that in the latter by the long range
proximity effect.
%
Since these two loops are very different in nature, we can anticipate
a different current-phase relation for ${\rm SN_{long}S}$ loop, but it was not explicitly demonstrated in
previous studies for the superconductor/normal/superconductor hybrid junction
\cite{Gennes,Kulik,Ivanov} and for the ${\rm SN_{long}S}$ loop \cite{Buttiker,Mtb}
in connection with the Andreev reflection process \cite{Andreev}.
%The periodicity of the persistent current of NS loop
%has been studied theoretically \cite{Buttiker,Mtb}, which
%shows that the period of the persistent current is the superconducting flux quantum
%$\Phi_0/2$ with $\Phi_0\equiv h/e$ for very long superconducting sector
%while the behavior of the persistent current resembles that of the normal loop with
%the periodicity of $\Phi_0$ as the superconducting sector diminishes.
%However the crossover of the periodicity has not yet been reported in experiments.
The current described by the motion of a pair
of electron and hole in the normal sector changes into that by a
Cooper pair in the superconducting sector. There is an
intermediate region near the edge of superconductor, where the
current is described by quasiparticles and quasiholes.%\cite{Gennes}
Then the current carried by the electrons and the
holes in the normal sector should be the same as that carried by
quasiparticles and quasiholes in the intermediate region.
%This is just the current conservation condition at the NS junction.
%A notable point in the previous work is that the wave vectors of
%quasiparticle and quasihole are set to be the same.
%We, however, think that there is no reason why two wave vectors are equal.
%Therefore, we here introduce a general condition that
This current conservation condition can be satisfied by
considering that two wave vectors of the quasiparticle and the quasihole can be different
from each other like those of electron and hole in the normal sector.
%
In solving the Bogoliubov-de Gennes (BdG) equation,
we consider the average energy of a pair of particles
is a dynamic variable rather than a constant chemical potential.
Thus the values of the dynamic variables are determined
by minimizing the free energy of the ${\rm SN_{long}S}$ loop.
%
Recently an experiment for the high-$T_c$ supercondutor (HTS) junction with interlayer much thicker
than the superconducting correlation length has been reported \cite{IlichevNS},
where the HTS junction was incorporated in a superconducting loop with threading external magnetic flux $\Phi_{\rm ext}$.
In this single-junction interferometer experiment, the current-phase relation shows the highly
non-sinusoidal behaviors so that the slope of the current for zero external flux becomes larger
than that for $\Phi_{\rm ext}=\Phi_0/2$ with the superconducting unit flux quantum
$\Phi_0\equiv h/2e$ as temperature goes down.
%the maximum of the amplitude of current approaches the point of
%zero external flux as temperature goes down.
%
The tunneling currents are supposed to take place through the long range proximity effect across the thick
$\rm{PrBa_2Cu_3O_7}$ (PBCO) interlayer. Since the $\rm{CuO}$ chains in the PBCO layer are metallic
at low temperature, the above junction can be considered as a ${\rm SN_{long}S}$ junction incorporated in a superconduction loop.
%
In this study we calculate the persistent current of the above ${\rm SN_{long}S}$ loop considering
the current conservation and free energy minimum conditions and
show that the experimentally observed non-sinusoidal type current can emerge through the long-range
proximity effect in the ${\rm SN_{long}S}$ loop.
%We also demonstrated that the persistent currents in ${\rm SN_{long}S}$ loop
%are modulated by the phase shift 2$\eta$ induced by the Andreev reflection process at NS boundaries.
%In the limit of low excitation $E\ll\Delta$, the persistent currents take after the Josephson type formula,
%$I/\sin 4\eta \propto \sqrt{\Delta}$, in contrast to the case with Josephson junction \cite{Tinkham}
%where $I/\sin\phi \propto \Delta$ where $\phi$ is the phase difference
%across the Josephson junction and $\Delta$ is the pair potential.
%
%Furthermore, we show that the present analysis results in
%the constant periodicity of superconducting unit flux quantum $\Phi_0/2$ %with $\Phi_0\equiv h/e$
%for the persistent current in ${\rm SN_{long}S}$ loop, regardless of the system
%parameters such as the circumference of the loop $L$, the length of the normal sector $d$
%and $\Delta$. This result is consistent with
%the experiments of the Andreev interferometer.\cite{Reulet}
%
%Finally we point out that the two persistent current states for nearly vanishing external flux
%can be a good candidate for a qubit due to the low noise from the external magnetic field source.
\section{Andreev reflections in a superconductor/normal/superconductor loop}
The quasiparticles in an SNS loop with
threading external magnetic flux can be described by the Bogoliubov-de Gennes (BdG) equation
\begin{eqnarray}
\label{BdG}
\left(\matrix{H_0-\xi & \Delta (z) \cr
\Delta^* (z) & -H^*_0+\xi}\right)
\left(\matrix{u(z) \cr v(z)}\right)
=E\left( \matrix{u(z) \cr v(z)}\right),
\end{eqnarray}
where $H_0=(-i\hbar\partial/\partial z-eA/c)^2/2m_e$ with the electron mass $m_e$,
$A=\Phi_{\rm ext}/L$ is the vector potential with the circumference of loop $L$,
and $\Delta(z)$ is the pair potential as a function of spatial coordinate $z$.
For a superconducting loop {\it without} junction threaded by an
Aharonov-Bohm flux, the quasiparticle wave functions $u(z)$ and
$v(z)$ contain extra factor due to gauge invariance such as
$u(z)=C e^{i(\lambda_{e}+\pi f/L)z}$ and $v(z)=C e^{i(\lambda_{h}-\pi f/L)z}$,
where $f\equiv \Phi_{\rm ext}/\Phi_0$.
Uniform flow of persistent current is derived by the BdG equation with the pair potential,
$\Delta (z)=\Delta e^{i(\lambda_{e}-\lambda_{h}+\pi f/L)z}$ \cite{Gennes}.
If the superconducting loop is interrupted by a normal sector as shown in Fig. \ref{fig:loop},
the wave function $\Psi=(u(z)~~ v(z))^T$ of a pair of electron and hole in normal sector $(0<z<d)$
and of a pair of quasiparticle and quasihole in the intermediate region of superconducting sector
$(d\lesssim z \, \, \mbox{and} \, \, z\lesssim L)$ is given by
\begin{eqnarray}
\label{wfns}
\left( \matrix{u(z) \cr v(z)}\right)=\left\{
\matrix{
\left( \matrix{A e^{i(k_0+\frac{\pi}L f)z} \cr B e^{i(k_1-\frac{\pi}L f)z}
}\right)~~~~~(0<z<d)~ \cr
\left( \matrix{C e^{i(\lambda_{+e}+\frac{\pi}L f)z+i\eta} \cr
C e^{i(\lambda_{+h}-\frac{\pi}L f)z-i\eta}}\right)~~~~~ (d \lesssim z)~ \cr
\left( \matrix{D e^{i(\lambda_{-e}+\frac{\pi}L f)z-i\eta} \cr
D e^{i(\lambda_{-h}-\frac{\pi}L f)z+i\eta}}\right)~~~~~ (z \lesssim L),
} \right.
\end{eqnarray}
where $\eta$ is the phase shift due to Andreev reflection at the NS interface
\cite{Buttiker,Mtb,KulikComm} and, when a pair of quasiparticles passes through the NS interface,
each quasiparticle acquires the additional phase equal to $\eta$.
Since we do not make any assumption about the sign of the wave vectors $k_0$ and $k_1$,
this wave function can describe both the excitations moving clockwise and counterclockwise.
%
Here a notable point is that the wave vectors of the quasiparticle and the quasihole
in the previous work \cite{Kulik,Buttiker,Mtb} are set to be the same.
It is, however, natural to discriminate the wave vectors like those of particles
$k_0$ and $k_1$ in the normal sector in order to satisfy the current conservation
condition. Therefore, we introduce different wave vectors, $\lambda_e$
and $\lambda_h$, for the quasiparticle and the quasihole in the intermediate region of
superconducting sector.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Fig. 1
\begin{figure}[t]
%\vspace*{4.8cm}\hspace{3cm}
%\special{psfile=SNLoop.eps vscale=40 hscale=40 hoffset=60 voffset=40 angle=0}
\vspace*{-0.3cm}
\onefigure[scale=0.5]{SNLoop.eps}
\vspace*{0.3cm}
\caption{A superconductor/normal/superconductor loop with length $L$. $k$ and $\lambda$
denote the wave vector of a electron in normal sector and a quasiparticle
in superconducting sector, respectively.}
\label{fig:loop}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
This equation can be solved easily for the normal sector in which $\Delta(z)=0$,
and yields the relations for $\xi$ and $E$ such that
\begin{eqnarray}
\label{normalU}
\xi=\frac{\hbar^2}{4m_e}(k^2_0+k^2_1),\\
\label{normalE}
E=\frac{\hbar^2}{4m_e}(k^2_0-k^2_1).
\end{eqnarray}
%Here $\xi$ is the average energy of a pair of electrons in normal sector and
%$E$ is the energy splitting. %due to the persistent current.
%The external flux $f$ splits the degenerate levels at $f=0$ into two levels,
%$\hbar^2k^2_0/2m_e=\xi+E$ and $\hbar^2k^2_1/2m_e=\xi-E$.
%
In the intermediate region of superconducting sector, however,
the BdG equation must be solved with the pair potential
\begin{eqnarray}
\Delta(z)=\left\{\matrix{
\Delta e^{i(\lambda_{+e}-\lambda_{+h}+2\pi f/L)z}~(z\gtrsim d)~ \cr
\Delta e^{i(\lambda_{-e}-\lambda_{-h}+2\pi f/L)z}~(z\lesssim L).
}\right.
\end{eqnarray}
%$\Delta(z)=\Delta e^{i(\lambda_{+e}-\lambda_{+h})z+2\pi f/L}$.
Then the BdG equation for $z\gtrsim d$ becomes
$\hbar^2\lambda^2_{+e}/2m_e-\xi+\Delta e^{-2i\eta}=E$ and
$\Delta e^{2i\eta}-\hbar^2\lambda^2_{+h}/2m_e+\xi=E.$
%\begin{eqnarray}
%\label{BdG1}
%\left[\frac{\hbar^2}{2m_e}\lambda^2_{+e}-\xi\right]+\Delta e^{-2i\eta}=E,\\
%\label{BdG2}
%\Delta e^{2i\eta}-\left[\frac{\hbar^2}{2m_e}\lambda^2_{+h}-\xi\right]=E.\\
%\end{eqnarray}
Representing $\lambda_{+e}$ and $\lambda_{+h}$
such as $\lambda_{+e}=\lambda_0+i\lambda'_0$ and
$\lambda_{+h}=\lambda_1+i\lambda'_1$ we get an expression for $\xi$ and $E$,
\begin{eqnarray}
\label{U}
\xi&=&\frac{\hbar^2}{4m_e} (\lambda^2_0+\lambda^2_1)(1-\alpha)\\
\label{E}
E&=&\frac{\hbar^2}{4m_e}
(\lambda^2_0-\lambda^2_1)(1+\alpha)+\Delta\cos2\eta
\end{eqnarray}
%By eliminating $E$ in Eqs. (\ref{normalE}) and (\ref{scE}),
%we obtain the equation
%\begin{eqnarray}
%\label{E}
%\frac{\hbar^2}{4m_e}(k^2_0-k^2_1)&=&\frac{\hbar^2}{4m_e}
%(\lambda^2_0-\lambda^2_1)(1+\alpha)+\Delta\cos2\eta \nonumber\\
%\end{eqnarray}
with $\alpha \equiv(m_e\Delta \sin 2\eta/\hbar^2\lambda_0\lambda_1)^2$,
$\lambda'_0=m_e\Delta \sin 2\eta/\hbar^2\lambda_0$ and
$\lambda'_1=m_e\Delta \sin 2\eta/\hbar^2\lambda_1$.
For $z\lesssim L$, we can also solve the BdG equation and find that
$\lambda_{-e}=\lambda_0-i\lambda'_0$ and
$\lambda_{-h}=\lambda_1-i\lambda'_1$.
%For ${\rm SN_{long}S}$ junction, the relation between the phase shift $2\eta$
%and the energy levels $E$ can be represented as $2\eta=\arccos(E/\Delta)$,
%where the wave vectors of the quasiparticles are set to be equal. In the present case of NS
%loop, we set the wave vectors different from each other and thus
%have the relation of Eq. (\ref{scE}) which can be
%reduced to that of the ${\rm SN_{long}S}$ junction case when $\lambda_0=\lambda_1$.
The phase matching conditions for the wave function
of Eq. (\ref{wfns}) at $z=d$ and $z=L$ are given by
${\hat A} e^{ik_0 d-i\lambda_0 d-i\eta}={\hat B} e^{ik_1 d-i\lambda_1 d+i\eta}={\hat C}$ and
${\hat A} e^{-i(\lambda_0+\pi f/L)L+i\eta}={\hat B} e^{-i(\lambda_1-\pi f/L)L-i\eta}= {\hat D}$,
%\begin{eqnarray}
%{\hat A} e^{ik_0 d-i\lambda_0 d-i\eta}={\hat B} e^{ik_1 d-i\lambda_1 d+i\eta}={\hat C},~~~~~~~~~~\\
%{\hat A} e^{-i(\lambda_0+\pi f/L)L+i\eta}={\hat B} e^{-i(\lambda_1-\pi f/L)L-i\eta}= {\hat D},
%\end{eqnarray}
where ${\hat A}, {\hat B}, {\hat C}$ and ${\hat D}$ represent the
phase part of the coefficients, $A,B,C$ and $D$, respectively. The
condition for the existence of a solution leads to the boundary
condition
%$(k_0-k_1)d+\chi-4\eta=2\pi n$ with $\chi=\chi_1-\chi_2$.
\begin{eqnarray}
\label{contns}
\left(k_0-k_1+\frac{2\pi f}L \right)d+\left(\lambda_0-\lambda_1+\frac{2\pi f}L \right)(L-d)
-4\eta =2\pi n.
\end{eqnarray}
%
%This condition corresponds to a periodic boundary condition for the wave function in Eq. (\ref{wfns}).
%The former part of Eq. (\ref{contns}) represents the phase evolution of a pair of
%electron and hole through the normal sector with thickness $d$ and the latter part
%that of a Cooper pair through the superconducting sector with thickness $(L-d)$.
When an electron becomes Andreev-reflected at the superconductor/nomal interface,
the transmitted quasiparticle pair obtains the additional phase $2\eta$.
The phase $4\eta$ in the boundary condition in Eq. (\ref{contns}) is the sum of
two phase changes $2\eta$ due to the Andreev reflections at each interface.
The condition of current conservation at NS interface will be given
by using the representation of the flux
$j=-(1/2m)[\phi(-i\hbar\partial/\partial z-eA/c)\phi^*
-\phi^*(-i\hbar\partial/\partial z-eA/c)\phi ]$.
In the wave function $\Psi=(u(z) ~~ v(z))^T$ in Eq. (\ref{wfns}), $u(z)$
is the electronlike wave function and $v(z)$ is the holelike wave
function which is the complex conjugate of the electronlike wave function.
The current representation for a pair of electrons thus should be obtained
with the wave function $\phi=(u(z) ~~ v^*(z))^T$ and we can get the relation
\begin{eqnarray}
\label{current}
k_0-k_1=\lambda_0-\lambda_1 \equiv 2q.
\end{eqnarray}
%Here we have one more variable in $\lambda$'s and one more equation, Eq. (\ref{current}).
%
The Cooper pairs in superconducting sector then should
carry the current $I\equiv n_c 2e\hbar 2q/m_c$
with the density of Cooper pairs $n_c$ and $m_c=2m_e$.
%and the wave function of Cooper pairs can be represented as $e^{2iqx}$.
\section{Persistent currents of a superconductor/normal/superconductor loop}
When we calculate the current through the ${\rm SN_{long}S}$ junction,
we may be able to set the average energy $\xi$ of the particle and the hole in Eq. (\ref{normalU})
equal to the chemical potential. But, in the ${\rm SN_{long}S}$ loop,
$\xi$ in Eq. (\ref{normalU}) need not be the constant chemical potential.
For example, we consider a simple normal loop with a threading flux.
The average energy of two particles $\xi$ at Fermi level is different from the chemical potential $\mu$
such that $\xi=(1/2)(\hbar^2/2m_e) (k_0^2 +k_1^2)=(1/2)(\hbar^2/2m_e) ((k_F-\pi f/L)^2 +(k_F+\pi f/L)^2)
\neq (\hbar^2/2m_e)k^2_F=\mu$.
Since $k_0$ and $k_1$ depend on the dynamic variable $\eta$ as well as $f$ in the ${\rm SN_{long}S}$ loop,
$\xi$ cannot be set as a constant chemical potential but should also be a dynamic variable to be determined.
%Eliminating $\xi$ from Eqs. (\ref{normalU}) and (\ref{U}) we can obtain the equation,
%\begin{eqnarray}
%\label{Uns}
%\frac12(k^2_0+k^2_1)&=&\frac12 (\lambda^2_0+\lambda^2_1)(1-\alpha).
%\end{eqnarray}
Since an extra variable $\xi$ is introduced, we need one more independent relation.
That is given by the requirement of free energy minimum \cite{Kim}.
%By minimizing the free energy we can obtain another condition to determine
%the five dynamic variables, $k_0,k_1,\lambda_0,\lambda_1,$ and $\eta$ \cite{Kim}.
Since the intermediate region of superconducting sector is so thin that
we neglect the energy of this region in the free energy expression and consider only the
energy of Cooper pairs.
In superconducting sector the Cooper pairs carry the persistent current corresponding
to the Cooper pair wave vector $2q=\lambda_0-\lambda_1$ and the energy of a Cooper pair
can be written as $(\hbar^2/2m_c)(2q)^2$ with $m_c=2m_e$.
Therefore the total free energy per particle $U_{\rm tot}$ can be written as
\begin{eqnarray}
2U_{\rm tot}=\frac{\hbar^2}{2m_e} (k^2_0+k^2_1)\frac{d}{L}
+\frac{\hbar^2}{2m_c}(\lambda_0-\lambda_1)^2 \left(1-\frac{d}{L}\right).
\end{eqnarray}
%where the constant $U_0$ is the minimum total energy.
Using Eqs. (\ref{normalU}) and (\ref{U}), the total free energy can be represented in $\lambda_0$
and $\lambda_1$.
From the condition $dU_{\rm tot}/d\eta=0$, we obtained the equation,
\begin{eqnarray}
\label{CE}
&&\left[(\lambda_0+\lambda_1)+(\lambda_0-\lambda_1)\frac{L}d+\frac{(2m_e\Delta\sin
2\eta)^2}{2\lambda^3_0 \hbar^4}\right]\frac{d\lambda_0}{d\eta}\nonumber\\
&+&\left[(\lambda_0+\lambda_1)-(\lambda_0-\lambda_1)\frac{L}d+\frac{(2m_e\Delta\sin
2\eta)^2}{2\lambda^3_1 \hbar^4}\right]\frac{d\lambda_1}{d\eta}\nonumber\\
&=&\frac{\lambda^2_0+\lambda^2_1}{2\lambda^2_0\lambda^2_1}
\left(\frac{2m_e\Delta}{\hbar^2}\right)^2\sin 4\eta.
%~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
\end{eqnarray}
Here $d\lambda_0/d\eta$ and $d\lambda_1/d\eta$ can be obtained by differentiating
Eqs. (\ref{normalU}), (\ref{normalE}), (\ref{U}), (\ref{E}), (\ref{contns}) and (\ref{current}) such as
$d\lambda_0/d\eta = [{\mathcal F}+\{8\lambda_1(\alpha k_0+k_1)+4\lambda_0\Gamma\sin 2\eta\}/L]/{\mathcal D}$ and
$d\lambda_1/d\eta = [{\mathcal F}+\{8\lambda_0(\alpha k_0+k_1)-4\lambda_1\Gamma\sin 2\eta\}/L]/{\mathcal D}$,
%\begin{eqnarray}
%\frac{d\lambda_0}{d\eta} = \frac{1}{{\mathcal D}}\left[{\mathcal F}
%+\frac1L\left\{8\lambda_1(\alpha k_0+k_1)+4\lambda_0\Gamma\sin 2\eta\right\}\right]~\\
%\frac{d\lambda_1}{d\eta} = \frac{1}{{\mathcal D}}\left[{\mathcal F}
%+\frac1L\left\{8\lambda_0(\alpha k_0+k_1)-4\lambda_1\Gamma\sin 2\eta\right\}\right],
%\end{eqnarray}
where ${\mathcal F}\equiv 2\lambda_0\lambda_1\Gamma\cos 2\eta
-2\tilde{\Delta}(k_0+k_1)\sin 2\eta-8k_0k_1/L$
and ${\mathcal D} \equiv -2(\lambda_0-\lambda_1)(\alpha k_0+k_1)
+(\lambda_0+\lambda_1)\Gamma \sin 2\eta$
with $\Gamma\equiv (k_0\lambda^2_0-k_1\lambda^2_1)\tilde{\Delta}^2\sin 2\eta
/2\lambda^3_0\lambda^3_1$ and $\tilde{\Delta}\equiv 2m_e\Delta/\hbar^2$.
Solving the coupled equations (\ref{normalU}), (\ref{normalE}), (\ref{U}), (\ref{E}), (\ref{contns}),
(\ref{current}) and (\ref{CE}) numerically, we obtain the energy levels corresponding to the two solutions
of the BdG equation in Eq. (\ref{BdG})
%which are degenerate at $f=\pm 0.25$
as a function of external flux $\varphi\equiv 2\pi f=2\pi\Phi_{\rm ext}/\Phi_0$
as shown in Fig. \ref{fig:EnDiagram}(a), where the solid and the
dashed lines denote the lower levels and the other lines the
higher levels. The ground state corresponds to the solid (dashed)
line for $\varphi <0 ~(\varphi >0)$. The persistent currents in Fig.
\ref{fig:EnDiagram}(b) are represented by the same lines as those
of the corresponding states in Fig. \ref{fig:EnDiagram}(a). The
persistent current, $I\equiv n_c 2e\hbar 2q/m_c$, can be written
by $I=I_0 2q/\sqrt{\tilde{\Delta}_0}$, where $I_0\equiv n_c
2e\hbar\sqrt{\tilde{\Delta}_0}/m_c$ and $\tilde{\Delta}_0\equiv
2m_e\Delta_0/\hbar^2$ is a gap potential chosen arbitrary.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Fig. 2
\begin{figure}[t]
%\vspace*{6.5cm}\hspace{3cm}
%\special{psfile=EnPc.eps vscale=90 hscale=100 hoffset=10 voffset=0 angle=0}
\vspace*{-6.5cm}
\onefigure[scale=0.9]{EnPc.eps}
\caption{(a) Excitation spectrum of the ${\rm SN_{long}S}$ loop. The solid (dashed)
line for $\varphi<0$ ($\varphi>0$) shows the ground state energy, where $\varphi\equiv 2\pi f$.
%with $\tilde{\Delta}=1$ and $L=10$.
(b) Persistent currents corresponding to the energy levels in (a).
Here we set $L\sqrt{\tilde{\Delta}_0}=10$ and $\Delta/\Delta_0=1.$}
\label{fig:EnDiagram}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Since the Cooper pairs in the superconductor are in the coherent condensate state,
each Cooper pair carries the same superconducting current.
A Cooper pair changes into a pair of normal electrons in the normal sector via
a pair of quasiparticles in the intermediate region.
Since the current in the loop should be conserved, the current equal to the macroscopic persistent
current in the superconducting sector flows in the normal sector.
%The current conservation condition in Eq. (\ref{current}) implies that the current of a pair of electrons
%in the normal sector must be the same as that of a Cooper pair in the superconducting sector.
%Hence each pair of electrons in the normal sector carries the same current,
%which makes the macroscopic persistent current flowing through the normal
%sector and the superconducting sector in the ${\rm SN_{long}S}$ loop as long as the phase coherence is maintained.
%Using Eq. (\ref{current}) the boundary condition in Eq. (\ref{contns}) can be reexpressed as
%$2qL-4\eta=2\pi(n-f)$, which indicates that the persistent current in ${\rm SN_{long}S}$ loop does not depend on
%the thickness of the normal sector $d$ but on the Andreev reflection phase shift $\eta$ and the
%total magnetic flux $f$.
%
In Fig. \ref{fig:EnDiagram}(b), we can see several persistent currents,
of which the persistent current at ground state corresponds to the solid line for $\varphi<0$
and the dashed line for $\varphi>0$, the saw-tooth type current.
Recently a single-junction interferometer experiment was done on HTS junction \cite{IlichevNS}
(${\rm YBa_2Cu_3O_{7-x}}$ /${\rm PrBa_2Cu_3O_7}$/${\rm YBa_2Cu_3O_{7-x}}$)
which is incorporated into a superconducting loop with penetrating magnetic flux.
Since there is no misorientation angle between the two d-wave superconductors
across the interlayer in this experiment, the phase difference across the interlayer
can be brought about only by the threading magnetic flux.
They obtained the current-phase relation at finite temperature,
where the slope for zero external flux is larger than that for the external flux $\varphi=\pi$.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Fig. 3
\begin{figure}[t]
%\vspace*{6.3cm}\hspace{3cm}
%\special{psfile=PcTemp.eps vscale=80 hscale=80 hoffset=0 voffset=-5 angle=0}
\vspace*{-1cm}
\onefigure[scale=0.8]{PcTemp.eps}
\caption{Current-phase relation at finite temperatures. As the temperature goes down,
the maximum approaches the point $\varphi=0$}
\label{fig:PcSin}
\end{figure}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
In this experiment, the thickness of the PBCO interlayer of the junction is as long as hundreds ${\rm \AA}$
which is much larger than the high-$T_c$ superconducting correlation length by
an order of magnitude and thus the Cooper pairs cannot directly tunnel the junction.
These long range proximity effects have been observed in many experiments on HTS junctions
\cite{Barner,Bozovic,Polturak,Umezawa} and considered as a characteristic of ${\rm SN_{long}S}$ junctions.
%for ${\rm La_{1.85}Sr_{0.15}CuO_4/La_2CuO_4/La_{1.85}Sr_{0.15}CuO_4}$ junction \cite{Bozovic}
%and in experiments for YBCO/PBCO/YBCO junctions.
Thus the experimental results may be explained by calculating the currents
of the ${\rm SN_{long}S}$ loop as a function of the threading external flux.
Here we can raise a question that whether it is a real proximity effect, i.e., there are
filaments of pinholes or microshots so that the Cooper pairs can be transported by the resonant
tunneling through localized states in filaments.
Recently an experiment on the trilayer HTS junction \cite{Bozovic} has been reported,
where they synthesized atomically smooth films of HTS and uniform trilayer junction
so that they conclude that the long-range proximity effect do not originate from the resonant
tunneling through the energy-aligned states in the microshots, but it is an intrinsic
property of the interlayer.
%Therefore, the supercurrents are expected to flow via metallic channels in the interlayer.
In fact, the supercurrents are known to flow via metallic ${\rm CuO}$ chains in the PBCO interlayer
\cite{Fehrenbacher,Suzuki}.
%It is well-known that the ${\rm CuO}$ chains in PBCO compound form metallic channels
%whereas the ${\rm CuO_2}$ planes show insulating behaviors at low temperature.
%\cite{Takenaka,Mizokawa,Fehrenbacher,Suzuki}
%Optical studies showed that
%holes are doped into the ${\rm CuO}$ chains while the ${\rm CuO_2}$ is undoped \cite{Takenaka}
%and thus the ${\rm CuO}$ chains have a dispersive band structure. \cite{Mizokawa}
%The model for hole depletion in the ${\rm CuO_2}$ plane was proposed \cite{Fehrenbacher}
%and it was experimentally demonstrated that, while the ${\rm CuO_2}$ plane
%shows the Mott variable range hopping and the insulation behavior, the ${\rm CuO}$
%chains highly conducting behavior at low temperature. \cite{Suzuki}
%Hence the long-range proximity effect and the tunneling current in YBCO/PBCO/YBCO junction
%of Ref. 7 come into being through the metallic ${\rm CuO}$ chains in PBCO interlayer,
%which can be explained by the present model of ${\rm SN_{long}S}$ junction in superconducting loop.
%The theory for the proximity effect in ${\rm SN_{long}S}$ junction \cite{deGennes} requires that
%the critical current of the junction take the form, $I_c \propto (T-T_c)^2 e^{-Kd}$,
%where $K^{-1}$ is the decay length in normal segment. In the experiments showing the long-range proximity
%effect, we can see the dependence of the critical current on temperature,\cite{Bozovic,Polturak}
%$I_c \propto (T-T_c)^2$, and on the barrier thickness,\cite{Polturak,Umezawa} $I_c \propto e^{-Kd}$,
%which can also be the grounds for the PBCO interlayer behaving as a normal metal.
From the currents in Fig. \ref{fig:EnDiagram}(b) corresponding to the states in Fig. \ref{fig:EnDiagram}(a)
we can obtain the persistent currents of the ${\rm SN_{long}S}$ loop at finite temperatures.
%The persistent currents corresponding to the states in Fig. \ref{fig:EnDiagram}(a) are shown
%in Fig. \ref{fig:EnDiagram}(b), where the ground state current takes the saw-tooth type form.
%All the Cooper pairs in the superconducting sector of the ${\rm SN_{long}S}$ loop carry the same macroscopic current
%which is also carried by the electrons in the normal sector due to the current conservation condition.
At finite temperature, since the current state with energy $\epsilon$ has the probability proportional to
$e^{-\epsilon/k_B T}$, we can obtain the thermally averaged persistent currents
from those current states in Fig. \ref{fig:EnDiagram} as shown in Fig. \ref{fig:PcSin}.
%
We can observe highly non-sinusoidal currents in Fig. \ref{fig:PcSin},
where the amplitude maxima approach the point $\varphi=0$ and the slope of current
at $\varphi=0$ becomes larger than that at $\varphi=\pi$ as temperature goes down.
The current takes the sinusoidal form only after the temperature goes up such that
$k_B T/\Delta \approx 0.4$ which corresponds to the temperature near $T_c$.
%
Actually the similar behavior can also be seen in the loop interrupted by a grain boundary Josephson junction
not by an ${\rm SN_{long}S}$ junction, which however comes from the phase difference across
the grain boundary Josephson junction due to the misorientation angle between two d-wave superconductors \cite{IlichevJJ}.
In ${\rm SN_{long}S}$ junctions the wave functions acquire the Andreev reflection phase shift $\eta$
which appears in the boundary condition of Eq. (\ref{contns}). The Andreev phase shift takes the place of the
misorientation angle of the grain boundary Josephson junction and thus results in the non-sinusoidal current-phase relation as
shown in the manuscript.
%
In Ref. \cite{IlichevNS} the authors explained their own experimental
results by assuming thermal fluctuations which induced the highly
non-sinusoidal current-phase relation at low temperatures. At
higher temperatures near the critical temperature they can recover
the sinusoidal current-phase relations. In the present study,
however, we can show that the non-sinusoidal current-phase
relation emerges naturally at low temperature without assuming
thermal fluctuations.
%
%In the previous studies \cite{Buttiker,Mtb} the periodicity $\Phi_0/2$ can also be seen
%when the thickness of the normal sector is much smaller than that of the superconducting sector.
%
%The periodicity of $\Phi_0/2$ of ${\rm SN_{long}S}$ loop can be seen in the modified Rifkin-Deaver experiments,
%\cite{Ebel,Rifkin} where the junction to be investigated is incorporated into a superconducting loop.
%Even though these experiments show the current-phase relation of ${\rm SN_{long}S}$ junctions, \cite{Ebel}
%the $2\pi$ periodicity in the phase difference across the junction can be interpreted
%as the $\Phi_0/2$ periodicity in the total magnetic flux threading the superconducting loop
%interrupted by a normal sector. \cite{Rifkin}
%
%Since the normal sector is usually very thin in the modified Rifkin-Deaver experiment,
%the experiment for the ${\rm SN_{long}S}$ loop with thick normal sector is invoked.
%
%This behavior is consistent with the experiments of the Andreev interferometer, \cite{Reulet}
%where the oscillation period is exactly the superconducting unit flux quantum
%regardless of the system parameters such as the thickness of the normal sector.
%
%Furthermore the periodicity of unit superconducting flux quantum has been also expected in recent
%theoretical studies. \cite{Samuelsson,Sharov}
%In Fig. \ref{fig:EnDiagram}(c) we show the phase shif $4\eta$ for the lower levels
%denoted as solid and dashed lines in Fig. \ref{fig:EnDiagram}(a), where
%the value of the ground state persistent current becomes zero when $4\eta=\pi$.
%Thus, by fitting the numerical data of the persistent current of this work we found
%a current phase relation different from the Josephson type formula as follows,
%\begin{eqnarray}
%\label{CR}
%I=C_0I_0\sqrt{\frac{\Delta}{\Delta_0}}\sin{4\eta(L,\Delta,f)},
%\end{eqnarray}
%where $C_0$ is a constant.
%and the dependence of amplitude on $L$ is included through $\eta(L,\Delta,f)$.
%The persistent currents expressed by the formula of Eq. (\ref{CR}) are indicated by
%various marks in Fig. \ref{fig:PcSin}. One can see that the
%fitting of formula (\ref{CR}) works very well especially when $1/L\sqrt{\tilde{\Delta}_0} \ll
%\Delta/\Delta_0$, in other words, for low energy excitation, $E \ll
%\Delta$, since $E \propto 1/L^2$.
%
%By comparing the numerical results with the relation in Eq. (\ref{CR})
%we show that the values of $C_0$ converge such that $C_0 \approx 0.3535 \approx 1/2\sqrt{2}$
%in the limit $E \ll \Delta$ as shown in Fig. \ref{fig:C0}.
%
%The main difference in the current relation between the NS junction and the Josephson tunnel junction
%is the dependence on gap function; $I/\sin{4\eta} \propto \sqrt{\Delta}$ for NS junction
%while $I/\sin\phi \propto\Delta$ for Josephson tunnel junctions. \cite{Tinkham}
%
%The $\sqrt{\Delta}$ dependence of the current in Eq. (\ref{CR}) can be obtained at
%the free energy minimum where the dynamical variable is determined.
%Recently quantum computing with the persistent current qubit using the two current states
%of the loop with Josephson junctions operating at $f\approx0.5$ has been actively studied \cite{Wal}.
%But, due to the severe decoherence problem, the two qubit coherent oscillation has not yet been achieved
%experimentally. One of the main causes for decoherence is the noise due to
%the external magnetic field source.
%In Fig. \ref{fig:EnDiagram} (a) we can see the degenerate ground states at $f=0$.
%These states have the persistent currents flowing oppositely and the different Andreev
%reflection phase shifts as shown in Fig. \ref{fig:EnDiagram} (b) and (c).
%The coherent single qubit oscillations can be performed by applying a microwave
%resonating with the energy level shift of two quantum states at $f\approx 0$.
%Since the noise from the external magnetic field source will be very much low at $f\approx 0$,
%the qubit using the two current states in ${\rm SN_{long}S}$ loop may have a much low decoherence.
\section{Summary}
We have studied the current-phase relation of the ${\rm SN_{long}S}$ loop with threading magnetic flux.
The Cooper pairs in the superconducting sector become the electron pairs in the normal sector
via the quasiparticle states in the intermediate region.
The net current in the ${\rm SN_{long}S}$ loop should be preserved and thus we introduce different wave vectors
of the pair of quasiparticles in order to satisfy the current conservation condition.
Furthermore, since the average energy of a pair of particles should be a dynamic variable,
we also consider the free energy minimum condition.
We obtained the persistent currents of the ground and excited states and
found that we can explain the experimentally observed highly non-sinusoidal
current-phase relation with maxima near the zero external flux at low temperature.
%The results show that the persistent current states in ${\rm SN_{long}S}$ loop can be used
%for a qubit with low decoherence.
%See fig.~\ref{f.1}, table~\ref{t.1} and eq.~(\ref{e.1}).
%See also~\cite{b.a,b.b}.
%\begin{equation}
%\label{e.1}
%0\neq1
%\end{equation}
%\begin{figure}
%\onefigure{epl-template.eps}
%\caption{Figure caption.}
%\label{f.1}
%\end{figure}
%\begin{table}
%\caption{Table caption.}
%\label{t.1}
%\begin{center}
%\begin{tabular}{lcr}
%first & table & row\\
%second & table & row
%\end{tabular}
%\end{center}
%\end{table}
\acknowledgments
This work was supported by Korea Research Foundation Grant
No. KRF-2003-070-C00020.
\begin{thebibliography}{0}
%\bibitem{b.a} \Name{Author F., Author S. \and Author T.} \REVIEW{Some Rev. A}{69}{1969}{9691}.
%\bibitem{b.b} \Name{Author F. \and Author S.} \Book{Some Book of Interest} \Editor{A. Editor} \Vol{9}
% \Publ{Publishing house, City} \Year{1939} \Page{666}.
%\bibitem{b.c} \Editor{Editor A.} \Book{Some Book of Interest} \Vol{9} \Publ{Publishing house, City}
% \Year{1939} \Section{A}.
\bibitem{Gennes} \Name{de Gennes P. G.} \Book{Superconductivity of Metals
and Alloys} \Publ{Benjamin, New York} \Year{1966}.
\bibitem{Kulik} \Name{Kulik I. O.} \REVIEW{Zh. Eksp. Teor. Fiz.}{57}{1969}{1745} [\REVIEW{Sov.
Phys. JETP}{30}{1970}{944}].
\bibitem{Ivanov} \Name{Ivanov Z. G., Kupriyanov M. Yu., Likharev K. K., Meriakri S. V. \and Snigirev O. V.}
\REVIEW{Sov. J. Low. Temp. Phys.}{7}{1981}{274}.
%\bibitem{Bardeen} J. Bardeen and J. L. Johnson, Phys. Rev. B {\bf 5}, 72 (1972).
%\bibitem{Blonder} G. E. Blonder, M. Tinkham, and T. M. Klapwijk,
%Phys. Rev. B {\bf 25}, 4515 (1982).
\bibitem{Buttiker} \Name{B{\"u}ttiker M. \and Klapwijk T. M.} \REVIEW{Phys. Rev. B}{33}{1986}{5114}.
\bibitem{Mtb} \Name{Cayssol J., Kontos T. \and Montambaux G.} \REVIEW{Phys. Rev. B}{67}{2003}{184508}.
\bibitem{Andreev} \Name{Andreev A. F.} \REVIEW{Zh. Eksp. Teor. Fiz.}{46}{1964}{1823}
[\REVIEW{Sov. Phys. JETP}{19}{1964}{1228}]; \REVIEW{{\it ibid.}}{49}{1965}{655} [\REVIEW{{\it ibid.}}{22}{1966}{455}].
\bibitem{IlichevNS} \Name{Il'ichev E., Zakosarenko V., Schultze V., Hoenig H. E.,
Meyer H.-G., Subke K. O., Burkhardt H. \and Schilling M.} \REVIEW{Appl. Phys. Lett.}{76}{2000}{100};
\Name{Il'ichev E., Zakosarenko V., Schultze V., Meyer H.-G., Hoenig H. E.,
Glyantsev V. N. \and Golubov A.} \REVIEW{{\it ibid.}}{72}{1998}{731}.
%\bibitem{Tinkham} \Name{M. Tinkham} \Book{Introduction to Superconductivity} \Publ{McGraw-Hill, New York}
%\Year{1996}.
%\bibitem{Ambeg} V. Ambegaokar and A. Baratoff, Phys. Rev. Lett. {\bf 10}, 486 (1963).
%\bibitem{Ebel} M. Ebel, C. Busch, U. Merkt, M. Grajcar, T. Plecenik, and E. Il'ichev, Phys. Rev. B {\bf 71},
%052506 (2005);A. A. Golubov, M. Yu. Kupriyanov, and E. Il'ichev, Rev. Mod. Phys. {\bf 76}, 411 (2004);
%M. Grajcar, M. Ebel, E. Il'ichev, R. K{\" u}rsten, T. Matsuyama, and U. Merkt, Physica C {\bf 372-376}, 27 (2002).
%\bibitem{Rifkin} R. Rifkin and B. S. Deaver, Phys. Rev. B {\bf 13}, 3894 (1976).
%\bibitem{Reulet} \Name{Reulet B., Kozhevnikov A. A., Prober D. E., Belzig W. \and Nazarov Yu. V.}
%\REVIEW{Phys. Rev. Lett.}{90}{2003}{066601}; \Name{Charlat P., Courtois H., Gandit Ph., Mailly D., Volkov A. F.
%\and Pannetier B.} \REVIEW{{\it ibid.}}{77}{1996}{4950}; \Name{Pothier H., Gu{\'e}ron S., Esteve D.
%\and Devoret M. H.} \REVIEW{{\it ibid.}}{73}{1994}{2488}.
\bibitem{KulikComm} In Ref. [2], this phase shift can be defined as $\eta\equiv\phi/2$ with $e^{-i\phi}=\gamma$.
\bibitem{Kim} \Name{Kim M. D., Shin D. \and Hong J.} \REVIEW{Phys. Rev. B}{68}{2003}{134513}.
\bibitem{Barner} \Name{Hashimoto T., Sagoi M., Mizutani Y., Yoshida J. \and Mizushima K.}
\REVIEW{Appl. Phys. Lett.}{60}{1992} {1756}; \Name{Barner J. B., Rogers C. T., Inam A., Ramesh R. \and Bersey S.}
\REVIEW{{\it ibid.}}{59}{1991}{742}.
\bibitem{Bozovic} \Name{Bozovic I., Logvenov G., Verhoeven M. A. J., Caputo P., Goldobin E. \and Beasley M. R.}
\REVIEW{Phys. Rev. Lett.}{95}{2004}{157002}.
%Yu. M. Boguslavskij, J. Gao, A. J. H. M. Rijnders, D. Terpstra, G. J. Gerritsma, and H. Rogalla, Physica C {\bf 194}, 268 (1991).
\bibitem{Polturak} \Name{Polturak E., Koren G., Cohen D., Aharoni E. \and Deutscher G.}
\REVIEW{Phys. Rev. Lett.}{67}{1991} {3038}.
\bibitem{Umezawa} \Name{Umezawa T., Lew D. J., Streiffer S. K. \and Beasley M. R.}
\REVIEW{Appl. Phys. Lett.}{63}{1993}{3221};
\Name{Barner J. B., Hunt B. D., Foote M. C., Pike W. T. \and Vasquez R. P.} \REVIEW{Physica C}{207}{1993}{381}.
%\bibitem{Takenaka} K. Takenaka, Y. Imanaka, K. Tamasaku, T. Ito, and S. Uchida, Phys. Rev. B. {\bf 46}, R5833 (1992).
%\bibitem{Mizokawa} T. Mizokawa, C. Kim, Z.-X. Shen, A. Ino, A. Fujimori, M. Goto, H. Eisaki, S. Uchida, M. Tagami,
%K. Yoshida, A. I. Rykov, Y. Siohara, K. Tomimoto, and S. Tajima, Phys. Rev. B {\bf 60}, 12335 (1999).
\bibitem{Fehrenbacher} \Name{Fehrenbacher R. and Rice T. M.} \REVIEW{Phys. Rev. Lett.}{70}{1993}{3471}.
\bibitem{Suzuki} \Name{Suzuki Y., Triscone J.-M., Eom C. B., Beasley M. R. \and Geballe T. H.}
\REVIEW{Phys. Rev. Lett.}{73} {1994}{328}; \Name{Lee M., Stutzman M. L., Suzuki Y. \and Geballe T. H.}
\REVIEW{Phys. Rev. B}{54}{1996}{R3776}.
\bibitem{IlichevJJ} \Name{Il'ichev E., Zakosarenko V., IJsselsteijn R. P. J., Schultze
V., Meyer H.-G., Hoenig H. E., Hilgenkamp H. \and J. Mannhart} \REVIEW{Phys. Rev. Lett.}{81}{1998}{894}.
%\bibitem{deGennes} P. G. de Gennes, Rev. Mod. Phys. {\bf 36}, 225 (1964).
%\bibitem{Samuelsson} \Name{Samuelsson P., Lantz J., Shumeiko V. S. \and Wendin G.} \REVIEW{Phys. Rev. B}{62}
%{2000}{1319}.
%\bibitem{Sharov} \Name{Sharov S. V. \and Zaikin A. D.} \REVIEW{Phys. Rev. B}{71}{2005}{014518}.
\end{thebibliography}
\end{document}