1: %\documentclass[twocolumn,superscriptaddress,showpacs]{revtex4}
2: \documentclass[twocolumn,showpacs]{revtex4}
3: \usepackage{times,xspace}
4: \usepackage{amsbsy,amssymb,amsmath,bm}
5: \usepackage{graphicx,color,epsfig,rotate}
6: \usepackage{fancyhdr}
7: \def\one{{\mathchoice {\rm 1\mskip-4mu l} {\rm 1\mskip-4mu l} {\rm
8: 1\mskip-4.5mu l} {\rm 1\mskip-5mu l}}}
9: \def\bbbc{{\mathchoice {\setbox0=\hbox{$\displaystyle\rm C$}\hbox{\hbox
10: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}
11: {\setbox0=\hbox{$\textstyle\rm C$}\hbox{\hbox
12: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}
13: {\setbox0=\hbox{$\scriptstyle\rm C$}\hbox{\hbox
14: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}
15: {\setbox0=\hbox{$\scriptscriptstyle\rm C$}\hbox{\hbox
16: to0pt{\kern0.4\wd0\vrule height0.9\ht0\hss}\box0}}}}
17: \newcommand{\Naturals}{\ensuremath{\mathbb{N}}\xspace}
18: \newcommand{\Integers}{\ensuremath{\mathbb{Z}}\xspace}
19: \newcommand{\Rationals}{\ensuremath{\mathbb{Q}}\xspace}
20: \newcommand{\Reals}{\ensuremath{\mathbb{R}}\xspace}
21: \newcommand{\Complex}{\ensuremath{\mathbb{C}}\xspace}
22: \newcommand{\Zero}{\ensuremath{\mathbb{O}}\xspace}
23: \newcommand{\Punitary}{\ensuremath{\mathbb{I}}\xspace}
24: \newcommand{\UPunitary}{\ensuremath{\mathbb{U}}\xspace}
25: \newcommand{\VPunitary}{\ensuremath{\mathbb{V}}\xspace}
26: \newcommand{\ket}[1]{|{#1}\rangle}
27: \newcommand{\bra}[1]{\langle{#1}|}
28: \newcommand{\kets}[2]{|{#1}\rangle_{{}_{\!\!{#2}}}}
29: \newcommand{\bras}[2]{{}_{{}_{{#2}\!\!}}\langle{#1}|}
30: \newcommand{\slb}[2]{{{#1}^{({#2})}}}
31: \newcommand{\trace}{\mbox{tr}}
32: \renewcommand{\tensor}{\otimes}
33: \newcommand{\vspan}{\mbox{span}}
34: \newcommand{\lowertr}{\mbox{lt}}
35: \newcommand{\diag}{\mbox{dg}}
36: \newcommand{\id}{I}
37:
38: \newcommand{\ignore}[1]{}
39: \newcommand{\mComment}[1]{}
40: \newcommand{\gComment}[1]{}
41: \newcommand{\jComment}[1]{}
42: \newcommand{\rComment}[1]{}
43: \newcommand{\lComment}[1]{}
44:
45: \renewcommand{\mComment}[1]{\textcolor{blue}{Manny: #1}}
46: \renewcommand{\gComment}[1]{\textcolor{red}{Gerardo: #1}}
47: \renewcommand{\jComment}[1]{\textcolor{green}{Jim: #1}}
48: \renewcommand{\rComment}[1]{\textcolor{magenta}{Ray: #1}}
49: \renewcommand{\lComment}[1]{\textcolor{purple}{Rolando: #1}}
50:
51: \pagestyle{fancy}
52: \pagestyle{fancyplain}
53: %\footrulewidth 0.4pt
54: %\plainheadrulewidth 0.4pt
55: %\plainfootrulewidth 0.4pt
56: \lhead{\large LA-UR-04-6501}
57: \chead{ \today}
58: \rhead{\sl submitted to Physical Review B}
59: \cfoot{\sc\thepage}
60: \lfoot{}
61: \rfoot{}
62:
63:
64: \begin{document}
65:
66:
67:
68: \title{Spin and Spin-Wave Dynamics in Josephson Junctions}
69: \author{Zohar Nussinov$^{1,2}$, Alexander Shnirman$^{3}$,
70: Daniel P. Arovas$^{4}$,
71: Alexander V. Balatsky$^{1}$, and Jian Xin Zhu$^{1}$}
72: \address{$^{1}$Theoretical Division,
73: Los Alamos National Laboratory, Los Alamos, NM 87545, USA}
74: \address{$^{2}$ Department of Physics, Washington University, St. Louis,
75: MO 63160-4899, USA}
76: \address{$^{3}$ Institut f\"ur Theoretische Festk\"orperphysik,
77: Universit\"at Karlsruhe,
78: D-76128 Karlsruhe, Germany}
79: \address{$^{4}$ Department of Physics, University of California,
80: San-Diego, La Jolla, CA 92093, USA}
81:
82: \date{Received \today }
83:
84:
85:
86: \begin{abstract}
87:
88: We extend the Keldysh formulation
89: to quantum spin systems and derive exact equations
90: of motion. This allows us to explore the dynamics of
91: single spins and of ferromagnets when
92: these are inserted between superconducting
93: leads. Several new effects are reported.
94: Chief amongst these are nutations
95: of single S=1/2 spins in Josephson junctions.
96: These nutations are triggered by the superconducting pairing
97: correlations in the leads. Similarly, we find that
98: on rather universal grounds, magnets
99: display unconventional spin wave dynamics
100: when placed in Josephson junctions. These lead
101: to modifications in the tunneling current.
102:
103:
104: \end{abstract}
105:
106: \pacs{71.27.+a, 71.28.+d, 77.80.-e}
107:
108: \maketitle
109:
110:
111: \section{Introduction}
112:
113: There is a growing interest in a number of techniques that allow
114: detection and manipulation of a single spin. A
115: partial list includes optical detection of electron spin resonance
116: (ESR) in a single molecule~\cite{Koehler93}, tunneling through a
117: quantum dot~\cite{Engel01}, and, more recently, the ESR-scanning
118: tunneling microscopy (ESR-STM) technique~\cite{Mana89,Durkan02}.
119: Interest in ESR-STM lies in the potential of detection and
120: manipulation of a single spin~\cite{Manoharan02,Balatsky02}- an ability
121: which is crucial to spintronics and quantum information processing.
122: Much work also addressed
123: coupling, feedback effects, and decoherence in a coupled
124: electronic-vibrational systems, such as nanomechanical oscillators
125: and local vibrational modes~\cite{Mozy02b}.
126: In particular, spintronic and quantum computing applications
127: greatly intensified interest in Josephson junctions.
128: In a previous publication \cite{nut03}, four of us studied
129: the effect of the supercurrent on a macroscopic spin cluster
130: (of spin $S \gg 1$)
131: precessing in the presence of a magnetic field when placed
132: in a Josephson junction to find new spin dynamics.
133: In \cite{bulaevskii}, these systems were examined anew
134: wherein variations in the DC current
135: were predicted for all systems harboring a spin of any finite size
136: $S$. In the current article, we complement
137: \cite{nut03} by studying, for the first time, the explicit dynamics of
138: {\em single quantum $S=1/2$ spins} in Josephson junctions
139: to find new intriguing dynamical effects
140: for which we provide quantitative expressions.
141: The single spin ($S=1/2$) dynamics which we
142: study here differs significantly from the
143: the large magnetic cluster ($S \gg 1$) dynamics
144: studied in \cite{nut03}. In the current
145: article, we further examine spin wave dynamics in ferromagnets
146: when placed in Josephson junctions.
147:
148: The analysis of spins embedded in Josephson junctions has a
149: long and rich history.
150: Early on, Kulik~\cite{Kulik_JETP66} argued that spin flip processes in
151: tunnel barriers reduce the critical Josephson current
152: as compared to the Ambegaokar-Baratoff
153: limit~\cite{Ambegaokar_Baratoff}. More than a decade later,
154: Bulaevskii et al.~\cite{Bulaevskii_pi_junction} conjectured that
155: $\pi$-junctions may form if spin flip processes dominate.
156: The competition between the Kondo effect and the superconductivity was
157: elucidated in \cite{Glazman_Matveev_SC_Kondo}. A nice review of
158: experimental works on certain aspects of
159: magnetic nanoparticles in Josephson junctions
160: is found in \cite{wolfgang}. Transport
161: properties formed the central core of many
162: pioneering works, while spin
163: dynamics was relegated a relatively trivial secondary role.
164: In the current article, we report on exact new non-stationary spin dynamics
165: and illustrate how a quantum $S=1/2$ spin is affected
166: by the Josephson current. As a consequence of
167: the Josephson current, spins exhibit non-planar
168: precessions while subject to the external magnetic field.
169: As well known, a single quantum spin in a magnetic field
170: exhibits circular Larmor precession
171: about the direction of the field.
172: As we report here, when the spin
173: is further embedded between two superconducting leads, quantum
174: pairing correlations lead to new out-of-plane
175: longitudinal motion, much alike that displayed by a classical mechanical
176: top will arise.
177: We term this effect the {\em Josephson nutation}. Similar effects occur
178: when a ferromagnetic slab is placed between two superconducting leads.
179: We outline how transport is, in turn, modulated by this rather
180: unusual spin dynamics.
181: The coupling of the spin to the supercurrent leads to an effective non-local
182: in time interaction of the single spin with itself.
183: Keldysh contour calculations illustrate
184: that a non-local
185: in time single fermion action is also found in situations wherein
186: the single spin is replaced by an Anderson
187: impurity~\cite{avishai}. As well known, in the limit of small
188: hopping amplitudes to and from an Anderson impurity, the impurity
189: attains a Kondo like character much like that of the single spin
190: which is the focus of our attention. Here we consider the origin
191: of this rather generic non-locality in time present in the
192: dynamics of a Josephson junction. En route to deriving this new spin
193: dynamics we illustrate that even in the presence of
194: non-local in time interactions, certain variants of
195: the classical equations of motion become trivially exact by virtue of
196: compactness of the spin variables. An elaborate extension of
197: these ideas will be detailed elsewhere \cite{mns}.
198:
199:
200:
201: \section{Outline of the article}
202: The main goal of the current publication
203: is to report on the spin and spin wave dynamics
204: (of single spins and of magnetic systems, respectively) in
205: Josephson junctions.
206:
207: To achieve this aim, we will initially
208: (in Sections(\ref{exact},\ref{keldrot}))
209: extend the non-equilibrium Keldysh formalism to address
210: these problems. In Section(\ref{exact}),
211: we illustrate that even in the presence
212: of effective non-local of time interactions of
213: a spin with itself (such as those borne
214: by the interaction of a single
215: spin with a Josephson current),
216: the equations of motion undergo
217: a trivial modification. In Section(\ref{keldrot}),
218: we rewrite these equations within the standard
219: Keldysh basis best suited for non-equilibrium
220: problems. Sections(\ref{exact},
221: \ref{keldrot}) may be seen as
222: independent extensions of basic facets
223: of the non-equilibrium Keldysh
224: formalism for a spin
225: system.
226:
227: In Section(\ref{single_spin}), we apply
228: the rather general formalism developed
229: in Sections(\ref{exact},\ref{keldrot})
230: to the specific problem of a single $S=1/2$
231: spin in a Josephson junction (with
232: a time independent potential difference
233: between the two superconducting leads).
234: We start, in subsection(\ref{system}),
235: in writing down the relevant Hamiltonian
236: of such a Josephson junction harboring
237: a single spin. In subsection(\ref{scales}),
238: we briefly highlight the natural time
239: scales in the problem- which will indeed
240: come to the fore in the detailed solution which
241: we will later expose. In the all-important
242: subsection(\ref{eff_action}), we highlight
243: the origin of the effective non-local
244: in time interactions of the
245: spin with itself. Here, we integrate
246: out the lead electrons to
247: find the effective spin only
248: action harboring such
249: non-local in time interactions.
250: These non-trivial interactions
251: are the reason that we needed
252: to develop and extend
253: (Sections (\ref{exact},\ref{keldrot}))
254: the Keldysh formalism
255: to a very general spin system
256: with such interactions.
257: In subsection(\ref{eom.}),
258: we invoke the results of
259: Sections(\ref{exact},\ref{keldrot})
260: to the resultant effective
261: spin-only action of Section(\ref{eff_action})
262: to write down the equations of motion
263: for the spin. In subsection(\ref{exact_soln.}),
264: we solve these equations of motion
265: to lowest order in the spin-dependent tunneling
266: amplitude. Detailed technical aspects of the
267: solution on which subsection(\ref{exact_soln.})
268: dutifully relies on have been relegated to
269: appendices B and C. The perturbative
270: solution to the equations of motion-
271: the final equations of
272: subsection(\ref{exact_soln.})- form one of
273: the main core results of the current publication.
274: In subsection(\ref{consequences}), we
275: examine the physical meaning
276: of this solution of the single
277: spin problem to unearth several
278: new predictions for this $S=1/2$ system.
279: In this subsection, we aim
280: to further arm the reader with an
281: intuitive understanding for the physical
282: origin of these new effects.
283: Some of these predicted effects
284: (and our prediction of nutation in particular)
285: are highlighted in Fig.(\ref{FIG:RIGID}).
286: In subsection(\ref{retS}),
287: we examine the behavior of the
288: system for a single spin of
289: magnitude $S > 1/2$. In the
290: large $S \gg 1$ limit, we
291: recover our very different
292: semi-classical spin ($S \to \infty$)
293: results of \cite{nut03}.
294:
295:
296: Next, in Section(\ref{AC_effects.}), we discuss
297: a variation of the single spin problem
298: wherein an AC voltage bias is applied
299: across the Josephson junction. Our
300: main result are the predictions
301: of specific time dependent
302: spin dynamics displaying
303: an infinite number of
304: harmonics and new DC
305: lock-in effects. The
306: predicted supercurrent
307: in this system is also
308: discussed.
309:
310:
311: In Section(\ref{ferroS}), we
312: examine the problem of
313: a ferromagnet in a Josephson
314: junction. In the spin-wave
315: approximation, we find that
316: each spin-wave mode displays some of the unusual
317: effects predicted in subsection(\ref{exact_soln.},
318: \ref{consequences}) for the single spin problem.
319: The predicted spin wave dynamics and
320: associated transport (current), are
321: furnished. In Section(\ref{Geometry.}), we
322: discuss simple extensions of our
323: results to other systems
324: generated by a trivial
325: change of geometry wherein
326: at least one of the superconductors
327: forming the Josephson junction
328: is replaced by a planar superconductor.
329: In Section(\ref{Large_S.}), we
330: write down the $S \gg 1$ equations
331: of motion for general magnets
332: and antiferromagnetic chains.
333: The non-uniform temporal evolution
334: of each of the spin-waves is
335: highlighted in the resultant
336: solution. We conclude the main text,
337: in Section(\ref{conc.}),
338: by highlighting our conclusions.
339:
340: In Appendix A, we briefly discuss
341: several experimental manifestations
342: of our effect and highlight a
343: proposed experiment which
344: may verify our predictions.
345:
346:
347:
348:
349:
350:
351:
352: \section{Exact Spin-1/2 Equations of Motion on Keldysh contours}
353: \label{exact}
354:
355:
356: \begin{figure}
357: \centerline{\psfig{figure=keldysh1.eps,height=4.5cm,width=5cm,angle=0}}
358: \caption{The standard Keldysh contour.
359: The times $T$ and $T^{\prime}$ are taken to
360: be $-\infty$ and $\infty$ respectively. The form of this
361: contour will be heavily employed in our work when time
362: ordering various spin products.} \label{FIG:keldysh1}
363: \end{figure}
364:
365:
366: We start by deriving the equations of motion
367: for a very general spin-1/2 system having
368: two (or more) local and non-local spin-spin
369: interactions at different times. In this work
370: we employ the non-equilibrium Keldysh technique.
371: Within this framework, the spin operators
372: on both up and down portions of the (Keldysh) contour
373: of Fig. 1 are normalized
374: and satisfy $[\vec{S}_{u}(t), \vec{S}_{d}(t')] =0$.
375: In what briefly follows, we will dispense with
376: operator formulations and employ a
377: path integral representation. Towards
378: this end, our working horses will be
379: the $CP_{1}$ spin coherent
380: variables ($z$) \cite{WZW,perel}
381: wherein the spins are represented by
382: \begin{eqnarray}
383: \vec{S} = S z^{*}_{a} \vec{\sigma}_{ab} z_{b}
384: \label{CP1.}
385: \end{eqnarray}
386: (with $S$ the spin magnitude). Here and throughout, we set $\hbar=1$.
387: In Eq.(\ref{CP1.}), $a,b = \uparrow, \downarrow$ and we assume
388: an implied summation over repeated indices. The vectors
389: $\vec{\sigma}_{ab}$ are the $ab$ components
390: of the three Pauli matrices.
391: The components $z_{a= \uparrow, \downarrow}$ code
392: for a two component
393: complex spinor subject to the normalization constraint,
394: $|z_{\uparrow}|^{2} + |z_{\downarrow}|^{2}=1$.
395: By glancing at Eq.(\ref{CP1.}), we note that
396: a knowledge of $\vec{S}$ specifies the two component spinor
397: $z$ only up to a global multiplicative phase.
398:
399: As well appreciated, in a spin coherent
400: basis, the Berry phases associated with the spin coherent states
401: are the net area of the spherical triangle spanned
402: by the spin as it moves on the Bloch sphere.
403: The latter may be expressed in the $CP_{1}$
404: basis as $S_{Berry} = i \int dt \sum_{a} z_{a}^{*} \partial_{t}
405: z_{a}$ \cite{WZW,perel}. For the benefit of readers
406: unfamiliar with this formalism, we provide in \cite{explain_CP1}
407: a quick derivation for this form of the Berry phase.
408:
409: We now assume the action contains
410: the single spin term $- \eta_{a} \int dt \vec{S}_{a} \cdot \vec{h}$
411: describing a single spin in an external magnetic field set by $\vec{h}$.
412: The parity $\eta_{a} = \pm 1$ is fixed by the direction of the
413: contour- $\eta_{up} = 1, \eta_{down} = -1$.
414: We further include a non-local in time spin interaction
415: $ \eta_{a} \eta_{b} \int dt \int dt^{\prime}
416: ~\overline{K}_{ab}(t,t^{\prime}) \vec{S}_{a}(t) \cdot
417: \vec{S}_{b}(t^{\prime})$. The kernels $\overline{K}_{ab}$ encapsulate
418: non-local temporal dependence. The generalization to higher order
419: terms is straightforward and leads to no qualitative change.
420: With the Berry phase included, the general action
421: \begin{eqnarray}
422: S = 2 iS \eta_{a} \int dt z_{a}^{*} \partial_{t} z_{a}
423: - S \int dt \eta_{a} \vec{h}_{a} \cdot z_{a}^{*}
424: \vec{\sigma} z_{a} \nonumber
425: \\ +S^{2} \eta_{a} \eta_{b} \int dt \int dt^{\prime}
426: \overline{K}_{ab}(t,t^{\prime}) z_{a}^{*}(t) \vec{\sigma} z_{a}(t)
427: \cdot z_{b}^{*}(t^{\prime}) \vec{\sigma} z_{b}(t^{\prime}).
428: \label{Sg}
429: \end{eqnarray}
430: Varying the action,
431: \begin{eqnarray}
432: \frac{\delta S}{\delta z^{*}_{a \nu}(t)} =
433: S \eta_{a} \Big(2i \partial_{t} z_{a \nu}(t) - \vec{h} \cdot
434: \vec{\sigma}_{\nu \gamma} z_{a \gamma}(t) \nonumber
435: \\ + \eta_{b} \int dt^{\prime} K_{ab}(t,t^{\prime})
436: z_{b \gamma}^{*}(t^{\prime}) \vec{\sigma}_{\gamma \delta}
437: z_{b \delta}(t^{\prime})
438: \cdot \vec{\sigma}_{\nu \gamma} z_{a \gamma}(t) \Big) \nonumber
439: \\ \equiv S \eta_{a} \Big( 2i \partial_{t} z_{a \nu}(t) - \vec{H}(t)
440: \cdot \vec{\sigma}_{\nu \gamma} z_{a \gamma}(t) \Big).
441: \label{fd}
442: \end{eqnarray}
443: Here,
444: \begin{eqnarray}
445: K_{ab}(t,t^{\prime}) \equiv \overline{K}_{ab}(t,t^{\prime})
446: + \overline{K}_{ba}(t^{\prime}, t) \nonumber
447: \\
448: \vec{H}(t) \equiv \vec{h} + S \eta_{a} \int dt^{\prime}
449: K_{ab}(t,t^{\prime}) z_{b \gamma}^{*}(t^{\prime})
450: \vec{\sigma}_{\gamma \delta} z_{b \delta}(t^{\prime}).
451: \label{KH}
452: \end{eqnarray}
453:
454: Next, we briefly generalize Ehrenfest's theorem
455: to situations such as the one of relevance
456: here where a non-local in time action is present.
457: A full discussion of this theorem
458: for general systems will be presented elsewhere
459: \cite{mns}. In what follows, the expectation value of any quantity
460: ${\cal{A}}$ evaluated with the action $S$ is denoted by
461: \begin{eqnarray}
462: \langle {\cal{A}} \rangle_{S} \equiv \frac{1}{Z} \int Dz Dz^{*} \delta(|z|^{2}-1)
463: {\cal{A}} e^{iS},
464: \end{eqnarray}
465: with $Z = \int Dz Dz^{*} \delta(|z|^{2}-1) e^{iS}$
466: the associated partition function. Similar definitions
467: apply, with a trivial replacement of
468: the measure when the action is a functional of one of more
469: real fields $\{x_{\alpha}(t)\}$. In the current context, $x_{\alpha}$ code
470: for the real or imaginary parts of the complex spinor
471: components $z$. Next, we note that for any cyclic coordinate $x$,
472: the expectation value of the variational derivative,
473: \begin{eqnarray}
474: \langle \frac{\delta S}{\delta x} \rangle_{S} =
475: \frac{-i}{Z} [e^{iS}]_{x_{i}(t)}^{x_{f}(t)}=0.
476: \label{Ehrenfest}
477: \end{eqnarray}
478: In the above, by the compactness of $x$,
479: in integrating all possible trajectories $x(t)$, the initial
480: and final trajectories are equal $x_{i}(t) = x_{f}(t)$.
481: This in turn lead to the vanishing expectation value
482: given in Eq.(\ref{Ehrenfest})
483: for all non-singular actions.
484: Analogously, this result follows by noting that for compact
485: coordinates, the transformation
486: [$x(t) \to x(t) + \delta x(t)$], with any $\delta x(t)$ leads to
487: no change to the value of $Z$- the range of integration in
488: $Z= \int Dx~ e^{iS}$ is unchanged. This, in turn, mandates
489: that $\langle \frac{\delta S}{\delta x} \rangle_{S} =0$ \cite{mns}.
490: Next, we consider ${\cal{A}}^{i} \equiv z^{*} \sigma^{i}
491: \frac{\delta S}{\delta z^{*}}$ and explicitly illustrate that its expectation
492: value vanishes, $\langle {\cal{A}}^{i} \rangle = 0$.
493: To this end, we write the spinors, longhand, in terms of
494: real and imaginary components,
495: $z^{*} = (z^{1}_{Re} - i z^{1}_{Im}~ z^{2}_{Re} - i z_{Im}^{2})$
496: and the measure $DzDz^{*} \delta(|z|^{2}-1) = Dz^{1}_{Re} Dz^{1}_{Im}
497: Dz^{2}_{Re} Dz^{2}_{Im} \delta(|z^{1}_{Re}|^{2} + |z^{1}_{Im}|^{2} +
498: |z^{2}_{Re}|^{2} + |z^{2}_{Im}|^{2} -1)$. Here and in what briefly follows
499: we suppress a uniform Keldysh contour index. The expectation value
500: $\langle A^{i} \rangle$ for each value of the spin index, $i=x,y,z$, is an
501: integral over bilinears in $z$ and hence amounts
502: to a sum of integrals of the type
503: \begin{eqnarray}
504: I_{\alpha \beta} \equiv
505: \int Dz^{1}_{Re} Dz^{1}_{Im}
506: Dz^{2}_{Re} Dz^{2}_{Im} \nonumber
507: \\ \delta(|z^{1}_{Re}|^{2} + |z^{1}_{Im}|^{2} +
508: |z^{2}_{Re}|^{2} + |z^{2}_{Im}|^{2} -1)
509: z_{\alpha} \frac{\delta S}{\delta z_{\beta}} e^{iS}.
510: \end{eqnarray}
511: Here, the indices $\alpha$ and $\beta$ span the four possible values
512: $(1 ~Re, 1 ~Im, 2 ~Re, 2~Im)$. An immediate consequence of the vanishing
513: of the expectation value $\langle \frac{\delta S}{\delta x} \rangle$
514: for any cyclic coordinate $x$ is that all integrals
515: of the form $I_{\alpha \neq \beta}$ vanish. An inspection
516: of $\langle {\cal{A}}^{i} \rangle$ reveals
517: that the contributions of all integrals of the type
518: $I_{\alpha \alpha}$ cancel identically
519: when $i=z$ (the only place where integrals
520: of the type $I_{\alpha = \beta}$
521: appear). Similarly, $\langle {\cal{A}}^{i \dagger} \rangle
522: = \langle \frac{\delta S}{\delta z(t)}
523: \sigma^{i} z \rangle = 0$. The
524: vanishing $\langle {\cal{A}}^{i} \rangle =
525: \langle {\cal{A}}^{i \dagger} \rangle =0$
526: imply that their difference,
527: \begin{eqnarray}
528: 0= \langle [\frac{\delta S}{\delta z_{a \sigma}(t)}
529: \sigma^{j}_{\sigma \sigma^{\prime}} z^{a \sigma^{\prime}}
530: - z^{*}_{a \sigma} \sigma^{j}_{\sigma \sigma^{\prime}}
531: \frac{\delta S}{\delta z^{*}_{a \sigma^{\prime}}(t)}] \rangle_{S},
532: \label{identity}
533: \end{eqnarray}
534: where the Keldysh contour index ($a$) is reinstated.
535:
536: Next, we explicitly insert Eq.(\ref{fd}) into Eq.(\ref{identity}).
537: As a consequence of the SU(2) algebra
538: of the Pauli matrices, we find that for each Keldysh contour index
539: $a=$ top/bottom,
540: \begin{eqnarray}
541: \langle \frac{\partial \vec{S}_{a}}{\partial t} \rangle_{S} = - \langle
542: \vec{H} \times \vec{S}_{a} \rangle_{S}.
543: \label{ud}
544: \end{eqnarray}
545: Eq.(\ref{ud})
546: is none other than the equation of motion for precession
547: of the spin $\vec{S}$ in the instantaneous
548: field given by $\vec{H}$ of Eq.(\ref{KH}). We find that such
549: classical equations of motion for a nonlocal in time
550: action are exact in the quantum arena.
551: [For affectionados of parafermion methods, we briefly note as an aside
552: that although throughout we employed the bosonic
553: spin coherent path integral representation,
554: a similar result follows if the spinors $z$ were Grassmann variables
555: (a net even number of permutations of the spinor coordinates
556: are involved in proving Eq.(\ref{ud})).]
557: The bulk of the paper will be devoted to a solution of
558: Eq.(\ref{ud}) for different realizations
559: of a Josephson junction system.
560:
561: We will momentarily dispense with the Keldysh
562: contour indices. Due to the commutation relations
563: $\vec{S} \times \vec{S} = i \vec{S}$,
564: although the field $\vec{H}$ contains a
565: piece which is linear in $\vec{S}$, the planar components of
566: Eq.(\ref{ud}), may be reduced for certain
567: problems to a linear
568: equation in planar spin components
569: $\langle S_{i} \rangle$ ($i=x,y$)
570: which then must have the solution
571: \begin{eqnarray}
572: \langle S_{i}(t) \rangle = U_{ij}(t)
573: \langle S_{j}(0) \rangle.
574: \label{SF}
575: \end{eqnarray}
576: We now invoke symmetry constraints.
577: An external magnetic field $\vec{h}$ in the action
578: (Eq.(\ref{Sg})) lifts the $SU(2)$ spin rotational symmetry of the
579: free spin leading in turn to a lower $U(1)$ symmetry
580: of rotations about the external magnetic field
581: axis. Such a symmetry is trivially encapsulated by
582: the operator $R^{z}(\theta)$ rotating $\langle \vec{S} \rangle$
583: by an angle $\theta$ about the z (or magnetic field) axis.
584: As a consequence,
585: the evolution operator $U(t)$ of Eq.(\ref{SF}) must commute with
586: $R^{z}(\theta)$.
587: This, in turn, dictates that if the solution is in the
588: form of Eq.(\ref{SF}), then the time evolution
589: operator $U(t)$ must have the form
590: \begin{eqnarray}
591: U(t) = \left(
592: \begin{array}{cc}
593: p(t) & q(t)\\
594: -q(t) & p(t) \\
595: \label{Uxy}
596: \end{array}
597: \right).
598: \end{eqnarray}
599: Similarly, due the azimuthal rotational symmetry encapsulated
600: by $R^{z}(\theta)$, the expectation value
601: $\langle S_{z}(t) \rangle$
602: must be independent of $\langle S_{x}(0) \rangle$ and $ \langle
603: S_{y}(0) \rangle$.
604: This form will indeed be borne out for our
605: full Keldysh problem.
606:
607:
608:
609:
610: \section{The Keldysh basis equations of motion}
611: \label{keldrot}
612:
613: Within the non-equilibrium Keldysh formalism it is often advantageous
614: to apply a simple linear transformation from the basis
615: of up and down contour fields to the symmetric and
616: antisymmetric linear combination of these
617: fields. E.g., for the spin
618: \begin{eqnarray}
619: \vec{S}_{cl} \equiv
620: \frac{1}{2}(\vec{S}_{up} + \vec{S}_{down}), \nonumber
621: \\ \vec{S}_{qu} \equiv (\vec{S}_{up} - \vec{S}_{down}).
622: \label{rotation}
623: \end{eqnarray}
624: The utility of this basis has its roots in the natural form
625: for the various correlation functions- all simply related to
626: the advanced, retarded, and ``Keldysh'' correlators.
627: The subscripts ``cl'' and ``qu'' of Eq.(\ref{rotation}) coding for
628: ``classical'' and ``quantum'' suggest
629: an intimate relation to classical and quantum
630: Langevin like dynamics. We refer the uninitiated reader
631: to excellent texts such as \cite{das}, \cite{alex}
632: where the origin of this link is explored in depth.
633: In Eq.(\ref{rotation}) we trivially generalize this
634: change of basis to quantum spin systems.
635: In this basis, when taken as operators
636: in Eq.(\ref{rotation}) [prior to a passage
637: to a path integral representation],
638: the spins no longer obey canonical
639: commutation relations the spins no longer obey canonical commutations
640: relations (e.g., $[\vec{S}_{qu},\vec{S}_{cl}] \neq 0$)
641: and are no longer normalized $(\vec{S}_{up} \pm \vec{S}_{down}$
642: may correspond to
643: a spin-triplet, $S=1$, or to a spin singlet, $S=0$).
644: Thus, we may not directly employ the $CP_{1}$ representation in this basis.
645: For the current purposes, the equations of motion
646: in this basis may be derived from Eq.(\ref{ud})
647: for the up and down contour spins,
648: \begin{eqnarray}
649: 0 = \langle \frac{d}{dt} S_{cl}^{k} + (\vec{h} \times \vec{S}_{cl})_{k}
650: + \int dt_{2} \epsilon_{ijk}
651: (S^{j}_{cl}(t) S^{j}_{qu}(t))
652: \nonumber
653: \\
654: \left(
655: \begin{array}{cc}
656: \frac{K_{uu}+K_{ud}- K_{du} - K_{dd}}{2} &
657: \frac{K_{uu} - K_{ud} - K_{du} + K_{dd}}{4} \\
658: \frac{K_{uu}+ K_{ud} + K_{du} + K_{dd}}{4} &
659: \frac{K_{uu} - K_{ud} + K_{du} - K_{dd}}{8}
660: \end{array}
661: \right) \nonumber
662: \\
663: \left(
664: \begin{array}{c}
665: S_{cl}^{i}(t_{2}) \\
666: S_{qu}^{i}(t_{2})
667: \end{array}
668: \right) \rangle_{S},
669: \label{exactcl}
670: \end{eqnarray}
671: and
672: \begin{eqnarray}
673: 0 = \langle \frac{d}{dt} S_{qu}^{k} + (\vec{h} \times \vec{S}_{qu})_{k}
674: + \int dt_{2} \epsilon_{ijk}
675: (S^{j}_{cl}(t) S^{j}_{qu}(t)) \nonumber
676: \\
677: \left(
678: \begin{array}{cc}
679: K_{uu} + K_{ud} + K_{du} + K_{dd} &
680: \frac{K_{uu} - K_{ud} + K_{du} - K_{dd}}{2} \\
681: \frac{K_{uu} + K_{ud} - K_{du} - K_{dd}}{2} &
682: \frac{K_{uu} - K_{ud} - K_{du} + K_{dd}}{4}
683: \end{array}
684: \right) \nonumber
685: \\
686: \left(
687: \begin{array}{c}
688: S_{cl}^{i}(t_{2}) \\
689: S_{qu}^{i}(t_{2})
690: \end{array}
691: \right) \rangle_{S}.
692: \end{eqnarray}
693: An average over $\exp[iS]$ is implicit in $\langle
694: ~ \rangle_{S}$. As emphasized
695: earlier, these are not merely saddle point
696: equations but are rather exact.
697: In the above, although the time arguments were
698: not explicitly written down, $K_{\alpha \beta}$
699: serves as a shorthand for $K_{\alpha \beta}(t,t_{2})$.
700:
701:
702: \section{Single Spin Dynamics in a Josephson Junction}
703: \label{single_spin}
704:
705: \subsection{The system}
706: \label{system}
707:
708: Our system is sketched in
709: Fig.~\ref{FIG:SETUP}. It consists of two identical
710: ideal superconducting
711: leads coupled each to a single spin; the
712: entire system is further subjected to a weak external
713: magnetic field. In Fig.(\ref{FIG:SETUP}), $\mu_{L,R}$ denote
714: the chemical potentials of the left and right leads,
715: ${\vec{B}}$ is a weak external magnetic field
716: along the z-axis,
717: and ${\vec {S}} = (S_{x}, S_{y},S_{z})$ is
718: the operator of the localized spin.
719: \begin{figure}
720: \centerline{\includegraphics[width=0.55\columnwidth]{jfig1.eps}}
721: \caption{Magnetic spin coupled to two superconducting leads.}
722: \label{FIG:SETUP}
723: \end{figure}
724: The wave-functions of our
725: system are superpositions of
726: the direct product of states of the left contact, the
727: impurity spin, and the right contact,
728: \begin{eqnarray}
729: | \psi \rangle = \sum f_{LSR} (|\psi_{L} \rangle \otimes
730: |\psi_{S} \rangle \otimes |\psi_{R} \rangle).
731: \label{entangled_eq}
732: \end{eqnarray}
733: A tunneling matrix
734: couples these different states.
735: The Hamiltonian of this system
736: reads
737: \begin{eqnarray}
738: {\cal H}={\cal H}_0+{\cal H}_T, \nonumber
739: \\
740: {\cal H}_0={\cal H}_{L}+{\cal H}_{R}-\mu B_{z}S_z, \nonumber
741: \\
742: {\cal H}_T=\sum_{\vec{k},\vec{p},\alpha,\alpha'} e^{i\phi/2}\;
743: c_{R\vec{k}\alpha}^{\dagger}\left[T_0\delta_
744: {\alpha\alpha'}+T_1\,\mbox{$\vec{\sigma}$}_{\alpha\alpha'}\cdot{\vec
745: S}\,\right]\,c_{L\vec{p}\alpha'}\nonumber
746: \\ \!\!\!\!\!\!\! \! \! \! + h.c.
747: \label{0-T}
748: \end{eqnarray}
749: Here, ${\cal H}_{L}$ and ${\cal H}_{R}$ are the Hamiltonians in
750: the left and right superconducting leads, while $c^{\dagger}_{ik\alpha}$
751: ($c_{ik\alpha}$) creates (annihilates) an electron in the lead a
752: in the state $\vec{k}$ with spin $\alpha$ in the right/left
753: lead for $i=L/R$ respectively.
754: ${\cal H}_{L(R)}=\sum_{k(p);\sigma}\epsilon_{k(p)}
755: c_{k(p),\sigma}^{\dagger}c_{k(p),\sigma}
756: +\frac{1}{2}\sum_{k(p);\sigma,\sigma^{\prime}}
757: [\Delta_{\sigma\sigma^{\prime}}(k(p)) c_{k(p),\sigma}^{\dagger}
758: c_{-k(-p),\sigma^{\prime}}^{\dagger} +\mbox{h.c.}]\;,$ where we
759: denote the electron creation (annihilation) operators in the
760: left (L) lead by $c_{k\sigma}^{\dagger}$ ($c_{k\sigma}$) while
761: those in the right (R) lead by $c_{p\sigma}^{\dagger}$
762: ($c_{p\sigma}$). The quantities $k$ ($p$) are momenta, $\sigma$
763: the spin index, while $\epsilon_{k(p),\sigma}$ and
764: $\Delta_{\sigma\sigma^{\prime}}(k(p))$ are, respectively, the
765: single particle energies of conduction electrons, and the pair
766: potential in the leads. In Eq.(\ref{0-T}), the
767: components $\mbox{$\vec{\sigma}$}_{\alpha\alpha'}$
768: are entries of the three Pauli matrices
769: $(\sigma^{x}_{\alpha\alpha'}, \sigma^{y}_{\alpha\alpha'},
770: \sigma^{z}_{\alpha\alpha'})$.
771: In the current publication we consider
772: s-wave symmetry pairing in the superconducting leads.Here,
773: $\mu$ is the magnetic moment of the spin. With the spin embedded
774: in the tunneling barrier, the conduction electron tunneling matrix becomes,
775: spin-dependent \cite{Balatsky02}
776: $\hat{T}=[T_0 \hat 1 + T_1 \vec{S} \cdot
777: \vec{\sigma}_{c}].$
778: %\end{equation}
779: Here $T_0$ is a spin-independent tunneling matrix element and
780: $T_1$ is a spin-dependent matrix element originating from the
781: direct exchange coupling $J$ of the conduction electron spin
782: $\vec{\sigma}_{c}$ to the
783: localized spin ${\vec{S}}$. Henceforth,
784: we will omit the $c$ subscript. We take both tunneling matrix elements
785: ($T_{0}$ and $T_{1}$) to be momentum independent.
786: This is not a crucial assumption and is merely introduced to
787: simplify notations. Typically, from the expansion of the work
788: function for tunneling, $\frac{T_1}{T_0} \sim J/U$,
789: where $U$ is the height of
790: a spin-independent tunneling barrier~\cite{Zhu_Balatsky}.
791: A weak external magnetic field $B_{z} \sim 100$ Gauss will not influence
792: the superconductors and we may
793: ignore its effect on the leads.
794: The operator $e^{i\phi/2}$ is the (single electron) number
795: operator. When the junction is linked to an external environment, the
796: coupling between the junction and the environment induces fluctuations
797: of the superconducting phase difference
798: across the junction $(\phi(t))$.
799:
800: \subsection{Physical Time Scales}
801: \label{scales}
802:
803: The Josephson junction with
804: the
805: spin has two time scales: (i) The Larmor
806: precession frequency of the spin
807: $\omega_L = g \mu_B B \equiv h$, where $g, \mu_B$ are the gyromagnetic
808: ratio and
809: Bohr magneton of the conduction electron, respectively. (ii) The
810: frequency $\omega_{J} = 2 eV$, with $e$ the electronic charge,
811: characterizes the Josephson effect when an external
812: voltage $V$ is applied across the junction.
813:
814:
815: \subsection{The Effective Action}
816: \label{eff_action}
817:
818: Josephson junctions are necessarily
819: embedded into external electrical circuits.
820: This mandates that the dynamics
821: explicitly depends on the superconducting phase difference
822: $\phi(t)$ across the junction.
823: The evolution operator is given by the real-time path integral
824: \begin{equation}
825: Z = \int D\phi D{\vec{S}} \; \exp\left[i S \right] \ .
826: \label{ZS}
827: \end{equation}
828:
829: The net action of Eq.(\ref{ZS}) is given by
830: $S= [\mathcal{S_{\rm
831: circuit}}(\phi)+ \mathcal{S_{\rm spin}}({\vec{S}})+
832: \mathcal{S_{\rm tunnel}(\phi,{\vec{S}})}]$.
833: The effective action $\mathcal{S_{\rm tunnel}}$ contribution
834: describes the junction itself. If
835: all external fields are the same on both forward and backward
836: branches of the Keldysh contour ($K$) then $\mathcal{Z} =
837: \mbox{Tr}\, T_K\,\exp [-i \oint_K dt H_{T}(t)] = 1$, where the
838: trace is over both the electron and the spin degrees of freedom
839: and $T_{K}$ denotes time ordering along the Keldysh contour.
840: The label $\oint_K$ denotes integration
841: along the Keldysh contour as shown in Fig.(\ref{FIG:keldysh1}).
842: We first take a partial trace in $\mathcal{Z}$ over the lead
843: fermions (the bath) to obtain an effective spin action. The
844: Josephson contribution to the resulting spin action reads
845: $- \frac{1}{2} \oint_K dt \oint_K dt' \langle
846: T_K {\cal H}_{T}({\mathbf S(t)},t) {\cal H}_{T}({\mathbf
847: S(t^{\prime})},t^{\prime}) \rangle$, much in the spirit of
848: Refs.~\cite{AES_82,Larkin_Ovchinnikov_83,ESA_84}. For brevity,
849: we set $A_{\sigma,\sigma^{\prime}} \equiv \sum_{k,p}
850: c_{k\sigma}^{\dagger} c_{p\sigma^{\prime}}$. The tunneling
851: Hamiltonian of a phase (voltage) biased junction
852: \begin{eqnarray}
853: {\cal H}_T = [T_0 \delta_{\sigma \sigma'} + T_1 \mathbf{S} \cdot
854: \bf{\sigma}_{\sigma \sigma'}] \nonumber
855: \\ \times \big(A_{\sigma
856: \sigma^{\prime}} \exp(i\phi/2) + A_{\sigma \sigma^{\prime}}^\dag
857: \exp (-i \phi/2)\big)\;. \label{EQ: A1}
858: \end{eqnarray}
859: In the presence of a dc voltage bias, $\phi = 2 eVt$. If $\phi$ is
860: treated classically (i.e. $\phi$ is the same on the upper and the
861: lower branches of the Keldysh contour), the contribution $\propto
862: T_0^2$ to $\delta \mathcal{S}$ vanishes. The mixed contribution
863: $\propto T_0 T_1$ vanishes due to the singlet spin structure of
864: the s-wave superconductor. The only surviving contribution reads
865: \begin{eqnarray}
866: -\frac{T_1^2}{2} \oint_K dt\oint_K
867: dt' \left[{\bf S}(t)\cdot \bf{\sigma}_{\alpha \beta}\right] \,
868: \left[{\bf S}(t') \cdot \bf{\sigma}_{\delta \gamma}\right] \nonumber
869: \\ \times \left(\langle T_K A_{\alpha \beta}(t) A_{\delta \gamma}(t')
870: \rangle e^{ i\frac{\phi(t) +\phi(t')}{2}} + (A,\phi\rightarrow
871: A^{\dag},-\phi)\right) &
872: \label{EQ:Seff1}
873: \end{eqnarray}
874: where we keep only the Josephson (off-diagonal) terms. The spin
875: structure simplifies for the s-wave case: \begin{eqnarray}
876: && T_1^2 \oint_K dt\oint_K dt' \left[{\bf
877: S}(t)\cdot {\bf S}(t')\right] [iD(t,t')] \ , \label{EQ:Seff3}
878: \end{eqnarray}
879: where the kernel $iD(t,t')$ is dictated
880: by $\langle T_K A_{\uparrow\uparrow}(t)
881: A_{\downarrow\downarrow}(t')\rangle e^{ i\frac{\phi(t)
882: +\phi(t')}{2}} + (A,\phi\rightarrow A^{\dag},-\phi)$. The operators
883: $A$ are bilinears in Fermi operators and thus the correlator
884: $\langle T_K A_{\uparrow\uparrow}(t)
885: A_{\downarrow\downarrow}(t')\rangle$
886: will amount to a sum of a product of two terms:
887: a product of two normal Green's function $G$ and
888: a product of two pair correlators $F$. Thus, generalizing
889: the known effective tunneling action for a spin-less
890: junction~\cite{AES_82,Larkin_Ovchinnikov_83,ESA_84} to the
891: new spin-dependent arena, we obtain
892: \begin{eqnarray}
893: \label{EQ:S_EFF} \mathcal{S_{\rm tunnel}} = - 2 \oint_K dt\oint_K
894: dt'\, \alpha(t,t')\,\left[T_0^2 + T_1^2 {\bf S}(t)\cdot{\vec
895: S}(t')\right]\, \nonumber
896: \\ \cos\left[{\frac{\phi(t) -\phi(t')}{2}}\right]
897: \nonumber \\ - 2 \oint_K dt\oint_K dt'\,
898: \beta(t,t')\,\left[T_0^2 - T_1^2 {\vec S}(t)\cdot{\vec
899: S}(t')\right]\, \nonumber
900: \\ \cos\left[{\frac{\phi(t) +\phi(t')}{2}}\right] \ ,
901: \end{eqnarray}
902: where $i\alpha(t,t')\equiv G(t,t')G(t',t)$ and $i\beta(t,t')\equiv
903: F(t,t')F^{\dag}(t,t')$. Here, the Green functions
904: \begin{eqnarray}
905: &&G(t,t')\equiv -i\sum_\mathbf{k}\,\langle T_{K}
906: c_{\mathbf{k}\sigma}^{\phantom{\dagger}}(t)
907: c_{\mathbf{k}\sigma}^{\dagger}(t') \rangle,\\
908: &&F(t,t')\equiv
909: -i\sum_\mathbf{k}\,\langle T_{K}
910: c_{\mathbf{k}\uparrow}(t)c_{\mathbf{-k}\downarrow}(t') \rangle, \\
911: &&F^{\dag}(t,t') \equiv -i \sum_\mathbf{k} \langle T_{K}\,
912: c^{\dag}_{\mathbf{k}\uparrow}(t)\, c^{\dag}_{-\mathbf{k}
913: \downarrow}(t') \rangle
914: \ .
915: \end{eqnarray}
916:
917:
918: We now express the spin action on Keldysh
919: contour in the basis of coherent states
920: \begin{equation}
921: \mathcal{S}_{\rm spin} = - \oint_{K} dt ~ {\vec{h}} \cdot {\vec{S}} +
922: \mathcal{S}_{WZNW} \ .
923: \label{wzwn}
924: \end{equation}
925: The second,
926: Wess-Zumino-Novikov-Witten (WZNW), term in Eq.(\ref{wzwn})
927: depicts the Berry phase
928: accumulated by the spin which we discussed earlier in the coherent
929: spin representation wherein it amounts to a kinetic bilinear-
930: the first term of Eq.(\ref{Sg})).
931: In the calculations that follow we replace the spin measure $DS$ by the
932: coherent spin state measure $Dz Dz^{*}$ and rely on our derived exact
933: equations
934: of motion. We now perform the Keldysh rotation of Eq.(\ref{rotation}),
935: defining
936: the values of the spin and the phase variables.
937: For the superconducting phase, we
938: introduce (with notations following Refs.~\cite{AES_82,ESA_84})
939: \begin{equation}
940: \phi \equiv \frac{1}{2} (\phi_{up} + \phi_{down}) \ \ ,
941: \ \ \chi \equiv \phi_{up} -
942: \phi_{down} \ .
943: \end{equation}
944: Within the Keldysh framework, the Josephson current is given by
945: \begin{eqnarray}
946: \langle I(t) \rangle = \frac{2 \pi}{\Phi_{0}} \langle
947: \frac{\delta S}{\delta \chi(t)} \rangle,
948: \label{i(t)}
949: \end{eqnarray}
950: with $\Phi_{0}$ the unit fluxon (with full units
951: restored, $\Phi_{0} = hc/e$ with $c$ the speed of light).
952: With these definitions in hand, the tunneling part
953: of the action reads
954: \begin{equation}
955: \label{EQ:S_tunnel} \mathcal{S}_{\rm tunnel} =
956: \mathcal{S}_{\alpha} + \mathcal{S}_{\beta} \ ,
957: \end{equation}
958: where the normal (quasi-particle) tunneling part
959: $\mathcal{S}_{\alpha}$ is expressed via the Green functions
960: $\alpha^{R}\equiv \theta(t-t')(\alpha^{>} - \alpha^{<})$ and
961: $\alpha^{K}(\omega) \equiv \alpha^{>} + \alpha^{<}$, where
962: $i\alpha^{>}(t,t') \equiv G^{>}(t,t')G^{<}(t',t)$ and
963: $i\alpha^{<}(t,t') \equiv G^{<}(t,t')G^{>}(t',t)$. Similarly the
964: Josephson-tunneling part $\mathcal{S}_{\beta}$ is expressed via
965: the off-diagonal Green's functions $\beta^{R}\equiv
966: \theta(t-t')(\beta^{>} - \beta^{<})$ and $\beta^{K}(\omega) \equiv
967: \beta^{>} + \beta^{<}$, where $i\beta^{>}(t,t') \equiv
968: F^{>}(t,t')F^{\dag >}(t,t')$ and $i\beta^{<}(t,t') \equiv
969: F^{<}(t,t')F^{\dag <}(t,t')$. The pair correlators
970: $F^{<}(t,t')$ are derived from $F^{>}(t,t^{\prime})$
971: by the interchange of $t$ with $t^{\prime}$.
972: In the current article, we focus on
973: the interaction between the supercurrent and the
974: spin.
975:
976: In Eq.~(\ref{EQ:S_EFF}),
977: the normal-tunneling part $\mathcal{S}_{\alpha}$ is obtained from
978: $\mathcal{S}_{\beta}$ by the following substitution:
979: $\beta^{R/K}(t,t') \rightarrow \alpha^{R/K}(t,t')$, $\phi(t')
980: \rightarrow -\phi(t')$, and $\chi(t') \rightarrow -\chi(t')$. The
981: Keldysh terms (those including $\beta^{K}$ and $\alpha^{K}$),
982: which normally give rise to random Langevin terms (see, e.g.,
983: Ref.~\cite{ESA_84}) are, in our case, suppressed at temperatures
984: much lower than the superconducting gap ($T \ll \Delta$), due to
985: the exponential suppression of the correlators
986: $\beta^{K}(\omega)$ and $\alpha^{K}(\omega)$ at $\omega <
987: \Delta$.
988:
989: To obtain $\beta^{R}$ we start from the Gor'kov Green functions
990: \begin{eqnarray}
991: F^{>}(t,t') = -i \sum_k \frac{\Delta}{2 E_k} e^{-iE_k (t-t')}, \nonumber
992: \\ F^{>\dag}(t,t') = i \sum_k \frac{\Delta}{2 E_k} e^{-iE_k
993: (t-t')} \ ,
994: \end{eqnarray}
995: where the quasi-particle energy
996: $E_k\equiv \sqrt{\Delta^2 + \epsilon_k^2}$, with $\epsilon_k$
997: the free-conduction-electron dispersion in the leads. Putting
998: all of the pieces together, we find that
999: \begin{equation}
1000: \beta^{R}(t-t')= - \theta(t-t') \sum_{k,p} \frac{\Delta^2}{2 E_k E_p}
1001: \sin\left[(E_k+E_p)(t-t')\right]\ .
1002: \label{BetaR}
1003: \end{equation}
1004: The kernel $\beta^{R}(t-t')$ decays on (short) time scales
1005: of order ${\cal{O}}(\hbar/\Delta)$.
1006: Similarly,
1007: \begin{eqnarray}
1008: \beta^{K}(t-t^{\prime}) = - i \sum_{k,p}
1009: \frac{\Delta^2}{2 E_k E_p}
1010: \cos \left[(E_k+E_p)(t-t')\right]\ .
1011: \label{BetaK}
1012: \end{eqnarray}
1013: Henceforth, we will often employ the shorthand $\beta^{R/K}(t,t^{\prime})
1014: \equiv \beta^{R/K}(t-t^{\prime})$.
1015: Looking at Eq.(\ref{BetaK}), we see that the
1016: Fourier transform $\beta^{K}(\omega)$ vanishes for
1017: frequencies $\omega < \Delta$. This
1018: is not so for the retarded correlator
1019: $\beta^{R}$ due to the presence of the theta function.
1020: For now, we ignore the fluctuations in the superconducting
1021: phase and set $\phi_{up}(t) = \phi_{down}(t) = \phi(t) = \omega_{J}t$
1022: with $\omega_{J} = 2eV$ (and thus $\chi=0$).
1023: In this, ``classical'', limit
1024: \begin{eqnarray}
1025: S_{tunnel} \simeq 4 \int dt \int dt^{\prime} \beta^{R}(t,t^{\prime})
1026: T_{1}^{2} \vec{S}_{qu}(t) \cdot \vec{S}_{cl}(t^{\prime}) j(t,t^{\prime})
1027: \nonumber
1028: \\
1029: + \int dt \int dt^{\prime} T_{1}^{2} \beta^{K}(t,t^{\prime})
1030: \vec{S}_{qu}(t)
1031: \cdot \vec{S}_{qu}(t^{\prime}) j(t,t^{\prime}),
1032: \label{Stunnel*}
1033: \end{eqnarray}
1034: with
1035: $j(t,t^{\prime}) \equiv \cos \frac{\phi(t)
1036: + \phi(t^{\prime})}{2}$.
1037:
1038:
1039: \subsection{The Equations of Motion}
1040: \label{eom.}
1041:
1042: With the action at our disposal, we now write down the
1043: exact equations of motions and give a
1044: solution, exact to order ${\cal{O}}(T_{1}^{2})$.
1045: Extracting, in the up-down contour basis,
1046: the coefficients, $\overline{K}_{ab}(t,t^{\prime})$
1047: of the $\vec{S}_{a}(t) \cdot \vec{S}_{b}(t^{\prime})$ terms
1048: in Eq.(\ref{Stunnel*}), constructing $K_{ab}(t,t^{\prime})$ from
1049: Eq.(\ref{KH}), and invoking Eq.(\ref{exactcl}),
1050: we find
1051: \begin{eqnarray}
1052: 0 = \langle \frac{d}{dt} \vec{S}_{cl} + \vec{h} \times \vec{S}_{cl} \nonumber
1053: \\ + 4 |T_{1}|^{2}
1054: \int dt^{\prime} j(t,t^{\prime})
1055: \beta^{R}(t,t^{\prime}) \vec{S}_{cl}(t^{\prime})
1056: \times \vec{S}_{cl}(t) \nonumber
1057: \\ + 2 |T_{1}|^{2} \int dt^{\prime} j(t,t^{\prime})
1058: \beta^{K}(t,t^{\prime}) \vec{S}_{qu}(t^{\prime})
1059: \times \vec{S}_{cl}(t) \nonumber
1060: \\ + |T_{1}|^{2} \int dt^{\prime} j(t,t^{\prime})
1061: \beta^{R}(t,t^{\prime}) \vec{S}_{qu}(t^{\prime}) \times \vec{S}_{qu}(t)
1062: \rangle_{S} \nonumber
1063: \\ \equiv \langle \frac{d}{dt} \vec{S}_{cl} + \vec{h} \times \vec{S}_{cl}
1064: + \vec{I}_{cl-cl}+ \vec{I}_{qu-cl} + \vec{I}_{qu-qu}
1065: \rangle_{S}.
1066: \label{class_final}
1067: \end{eqnarray}
1068: The final subscript $S$ serves to remind us that this
1069: is the path integral average computed with the action $S$.
1070: The various subscripts of the integrals $I$ denote
1071: the terms that they originate from
1072: (e.g. $I_{cl-cl} = 4 |T_{1}|^{2}
1073: \int dt^{\prime} j(t,t^{\prime})
1074: \beta^{R}(t^{\prime},t) \vec{S}_{cl}(t^{\prime})
1075: \times \vec{S}_{cl}(t)$).
1076: In Appendices B and C we outline, in detail,
1077: the evaluation of the various terms in Eq.(\ref{class_final}).
1078: We will now solve Eq.(\ref{class_final})
1079: to order ${\cal{O}}(T_{1}^{2})$.
1080:
1081:
1082:
1083:
1084:
1085:
1086:
1087:
1088:
1089:
1090:
1091:
1092:
1093:
1094:
1095:
1096:
1097:
1098:
1099:
1100:
1101: \subsection{Spin Dynamics in a Josephson Junction: An Exact Solution
1102: to ${\cal{O}}(T_{1}^{2})$}
1103: \label{exact_soln.}
1104:
1105:
1106: With all of the ingredients in place, we may now
1107: solve Eq.(\ref{class_final}) to
1108: determine the spin dynamics to ${\cal{O}}(T_{1}^{2})$.
1109: Henceforth, we will examine throughout the observable ``classical''
1110: component of the spin $\vec{S}_{cl}$. To make the expressions more
1111: appealing we will dispense with the classical ``cl'' subscript.
1112: Similarly, the action $S$ subscript in all expectation values
1113: will be omitted as no time ordering subtleties appear below.
1114: We expand the spin as
1115: \begin{eqnarray}
1116: \langle \vec{S}(t) \rangle = \langle \vec{S}_{0}(t) \rangle + \langle
1117: \delta \vec{S}(t) \rangle.
1118: \label{deltacorrection}
1119: \end{eqnarray}
1120: Here, $\vec{S}_{0}$ is the solution to the (Larmor) problem
1121: of a single free spin in an external magnetic
1122: field. We computed the integrals borne
1123: by these zeroth order Larmor components in subsection(\ref{integrals}).
1124: Similarly, $\delta S(t)$ are the
1125: contributions borne by the retarded and Keldysh
1126: correlations. These corrections will lead to
1127: higher order contributions in $\langle \vec{I} \rangle$ which
1128: are irrelevant to our ${\cal{O}}(T_{1}^{2})$
1129: solution. We insert Eq.(\ref{deltacorrection})
1130: into the equations of motion
1131: (Eqs.(\ref{class_final}))
1132: and retain all terms to order
1133: ${\cal{O}}(|T_{1}|^{2})$.
1134: This trivially leads to
1135: \begin{eqnarray}
1136: \frac{d}{dt} \langle \delta S_{x} \rangle - \omega_{L} \langle \delta S_{y}
1137: \rangle + \langle I_{x} \rangle =0, \nonumber
1138: \\ \frac{d}{dt} \langle \delta S_{y} \rangle + \omega_{L} \langle
1139: \delta S_{x} \rangle + \langle I_{y} \rangle = 0, \nonumber
1140: \\ \frac{d}{dt} \langle \delta S_{z} \rangle + \langle I_{z} \rangle = 0.
1141: \label{setxyz}
1142: \end{eqnarray}
1143:
1144: Here, $I_{\alpha=x,y,z}$ is the $\alpha$ direction component
1145: of $\langle \vec{I}_{cl-cl}+ \vec{I}_{qu-cl} \rangle$ which was computed
1146: in the previous subsection to order ${\cal{O}}(|T_{1}|^{2})$.
1147: We see that the integrals $\vec{I}$ play the role
1148: of a driving force. Integrating, we find that
1149: \begin{eqnarray}
1150: \langle \delta S_{z}(t) \rangle = |T_{1}|^{2} (1- \cos \omega_{J} t)
1151: [\sum_{k,p} \frac{\Delta^{2}
1152: \omega_{L}}{E_{k} E_{p}(E_{k}+ E_{p})^{3}} \nonumber
1153: \\ + \langle S_{z}(0) \rangle \sum_{k,p} \frac{\Delta^{2}}{E_{k}E_{p}
1154: (E_{k}+E_{p})^{2}}].
1155: \label{deltaSz}
1156: \end{eqnarray}
1157:
1158: Differentiating the equation of motion for $\langle \delta S_{x,y}
1159: \rangle$ in Eq.(\ref{setxyz}) and inserting the equation of motion
1160: for $\langle \delta S_{y,x} \rangle$ we immediately
1161: obtain the equation of motion of a driven harmonic oscillator.
1162: A simple solution yields
1163: \begin{eqnarray}
1164: \langle \delta S_{x}(t) \rangle
1165: = c_{1} \cos \omega_{L} t + c_{2} \sin \omega_{L} t \nonumber
1166: \\ + \sum_{\omega_{n} } (\frac{A_{n}}{\omega_{L}^{2} - \omega_{n}^{2}}
1167: \cos \omega_{n} t
1168: + \frac{B_{n}}{\omega_{L}^{2} - \omega_{n}^{2}} \sin \omega_{n} t)
1169: \label{sxsoln}
1170: \end{eqnarray}
1171: with
1172: \begin{eqnarray}
1173: A_{\omega_{L} + \omega_{LJ}} = |T_{1}|^{2} \sum_{k,p}
1174: \frac{\Delta^{2} \langle S_{x}(0) \rangle
1175: (2 \omega_{L}^{2} + \omega_{J}^{2}
1176: +3 \omega_{L} \omega_{J} )}{2E_{k} E_{p} (E_{k}+E_{p})^{2}}
1177: \nonumber
1178: \\ A_{\omega_{L} - \omega_{J} } = |T_{1}|^{2} \sum_{k,p}
1179: \frac{\Delta^{2} \langle S_{x}(0) \rangle
1180: (2 \omega_{L}^{2} + \omega_{J}^{2}-3 \omega_{L}
1181: \omega_{J}) }{2E_{k} E_{p} (E_{k}+E_{p})^{2}}
1182: \nonumber
1183: \\ B_{\omega_{L}+ \omega_{J}} =
1184: |T_{1}|^{2} \sum_{k,p}
1185: \frac{\Delta^{2}
1186: \langle S_{y}(0) \rangle (
1187: 2 \omega_{L}^{2} + \omega_{J}^{2}+ 3 \omega_{L} \omega_{J})}
1188: {2E_{k} E_{p} (E_{k}+E_{p})^{2}} , \nonumber
1189: \\ B_{\omega_{L}- \omega_{J}} =
1190: |T_{1}|^{2} \sum_{k,p}
1191: \frac{\Delta^{2}
1192: \langle S_{y}(0) \rangle (2 \omega_{L}^{2} + \omega_{J}^{2}
1193: - 3 \omega_{L} \omega_{J})}{2E_{k} E_{p} (E_{k}+E_{p})^{2}} .
1194: \label{AB}
1195: \end{eqnarray}
1196:
1197: All in all, to ${\cal{O}}(T_{1}^{2})$, the evolution of
1198: the planar spin components can be expressed in the format
1199: of Eqs.(\ref{SF},\ref{Uxy}) with
1200: \begin{eqnarray}
1201: p(t) = \cos \omega_{L} t + |T_{1}|^{2} \sum_{k,p} \frac{\Delta^{2}}
1202: {2 E_{k} E_{p} (E_{k}+ E_{p})^{2}} \nonumber
1203: \\ \times \
1204: \Big( \frac{(2 \omega_{L}^{2} + \omega_{J}^{2} + 3 \omega_{L} \omega_{J})
1205: \cos(\omega_{L} + \omega_{J}) t}
1206: {\omega_{L}^{2} - (\omega_{L} + \omega_{J})^{2}} \nonumber
1207: \\ + \frac{(2 \omega_{L}^{2} + \omega_{J}^{2} - 3 \omega_{L} \omega_{J})
1208: \cos(\omega_{L} - \omega_{J}) t}{\omega_{L}^{2} - (\omega_{L} - \omega_{J})^{2}}
1209: \Big),
1210: \label{pform}
1211: \end{eqnarray}
1212: and
1213: \begin{eqnarray}
1214: q(t) = \sin \omega_{L} t + |T_{1}|^{2} \sum_{k,p} \frac{\Delta^{2}}
1215: {2 E_{k} E_{p} (E_{k}+ E_{p})^{2}} \nonumber
1216: \\ \times \Big( \frac{(2 \omega_{L}^{2} + \omega_{J}^{2} + 3 \omega_{L}
1217: \omega_{J})
1218: \sin(\omega_{L} + \omega_{J}) t}
1219: {\omega_{L}^{2} - (\omega_{L} + \omega_{J})^{2}} \nonumber
1220: \\
1221: + \frac{(2 \omega_{L}^{2} + \omega_{J}^{2} - 3 \omega_{L} \omega_{J})
1222: \sin(\omega_{L} - \omega_{J}) t}{\omega_{L}^{2} - (\omega_{L} - \omega_{J})^{2}}
1223: \Big).
1224: \label{qform}
1225: \end{eqnarray}
1226:
1227: This concludes our solution
1228: for the dynamics of a spin in a Josephson
1229: junction. Our analysis throughout
1230: centered on Josephson junctions composed of
1231: s-wave superconductors (see our starting point
1232: Eq.(\ref{EQ:Seff3})). Slightly different quantitative
1233: results appear for other pairing symmetries
1234: (allowing, in theory, a determination of the pairing
1235: symmetry from observations of the spin/spin-wave dynamics
1236: and associated effects). The deviations from
1237: simple Larmor precessions are far stronger for
1238: triplet (i.e. odd angular momenta) superconductors.
1239:
1240: \subsection{Physical Consequences:
1241: Josephson Nutations and Other Dynamical Effects}
1242: \label{consequences}
1243:
1244: We now discuss the physics behind our exact
1245: (to ${\cal{O}}(T_{1}^{2})$) solution.
1246: Our solution provides testimony (and to new {\em quantitative}
1247: predictions) for several, inter-related, intriguing dynamical effects.
1248: We outline these below.
1249:
1250: $\bullet$ {\underline{{\em Josephson Nutations:}}
1251:
1252: In any system harboring a continuous rotational
1253: $U(1)$ symmetry about a certain axis (z),
1254: the orbital angular momentum $L_{z}$
1255: is a constant of motion. Needless to say, the same
1256: trivially holds true for any spin
1257: system in which $[H, S_{z}]=0$
1258: with $H$ the system Hamiltonian.
1259: In the presence of an external magnetic
1260: field along (or defining) the z-axis, as in the Larmor problem,
1261: the Hamiltonian $H = - h S_{z}$ commutes with $S_{z}$
1262: and the longitudinal magnetization $\langle S_{z}(t) \rangle$ is a constant
1263: of motion. In our system with non
1264: local in time interactions triggered by superconducting pair
1265: correlations, such a conservation law no longer holds.
1266: Perusing Eq.(\ref{deltaSz}), we find that
1267: the spin {\em nutates} above its average value.
1268: This occurrence for the $S=1/2$ is similar
1269: to that reported in \cite{nut03} for macroscopic
1270: spin clusters $S \gg 1$. Here, however,
1271: the quantum fluctuations are profound for the $S=1/2$ case and
1272: lead to strong deformations of the nutation shape.
1273: The physical engine behind the nutations
1274: is the small time separation between
1275: the two tunneling electrons forming
1276: the Cooper pair. As the ``first''
1277: electron tunnels through, it exerts
1278: a torque on the spin. A certain time
1279: later (of order $\hbar/ \Delta$ with dimensions restored)
1280: after the spin $\vec{S}$ has already
1281: revolved a small amount, the opposite
1282: spin member of the Cooper pair tunnels
1283: through and exerts a torque of an opposite
1284: sign on the spin $\vec{S}$. Due to the small
1285: time lag between the two tunneling
1286: electrons, the two opposite sign
1287: torques exerted on $\vec{S}$ by the two
1288: opposite sign spins of the tunneling
1289: singlet do not cancel and lead to a net
1290: effect. This origin is made
1291: evident in the retarded correlations
1292: $\beta^{R}$ which further spark a non-vanishing
1293: driving force $\langle I_{cl-cl} \rangle$ along
1294: the z-axis. Mathematically, all of this
1295: results as the tunneling portion of
1296: the action contains terms which trivially
1297: do not conserve $S_{z}$. In the aftermath,
1298: this led to an effective
1299: time dependent force acting on $S_{z}$.
1300: Its form may be seen by examining
1301: the integral $\langle I_{z} \rangle$
1302: appearing in Eq.(\ref{setxyz}).
1303: The latter is the z-component of the
1304: integrals $\langle \vec{I}_{cl-cl} \rangle$
1305: and $\langle \vec{I}_{qu-cl} \rangle$
1306: appearing in Eqs.(\ref{Icl-cl*}, \ref{Iqu-cl*}).
1307: (Needless to say, if both members of the Cooper
1308: pair share the same polarization (as in triplet
1309: superconductors) then a far greater effect results.)
1310:
1311:
1312: A manifestation of the resulting dynamical effect as
1313: a consequence of these effective external forces in conventional
1314: (s-wave) Josephson Junctions is vividly seen in Eq.(\ref{deltaSz}).
1315: We have derived similar expressions via an
1316: independent density matrix approach \cite{NS}.
1317: An exaggerated schematic of this effect
1318: is provided in Fig.(\ref{FIG:RIGID})
1319: which, qualitatively, is none other than the standard
1320: illustration for classical rigid
1321: body nutations. We find that such motions
1322: now appear in the quantum arena for a
1323: single $S=1/2$ particle! The precise shape of our
1324: trajectories, however, is markedly different from that
1325: exhibited by classical rigid rotors.
1326:
1327:
1328: \begin{figure}
1329: \centerline{\includegraphics[width=0.5\columnwidth]{jfig4.eps}}
1330: \caption{The resulting spin motion on the unit sphere in the
1331: general case. As in the motion of classical spinning top, the spin
1332: exhibits undulations along the polar direction. As a consequence of
1333: entanglement with the tunneling electrons, the magnitude
1334: of the spin is not constant- the spin further ``breaths''
1335: in and out as it nutates.}
1336: \label{FIG:RIGID}
1337: \end{figure}
1338:
1339:
1340:
1341:
1342:
1343: $\bullet$ {\underline{\em{Spin Contractions and
1344: Effective Longitudinal Fields:}}
1345:
1346: Glancing at Eq.(\ref{Iqu-cl*}), the reader
1347: will see that the effective $\langle \vec{I}_{qu-cl} \rangle$
1348: can be seen to dilate the spin (the uniform contribution
1349: proportional to $\langle \vec{S}(t) \rangle$
1350: in the second equality of Eq.(\ref{Iqu-cl*}))
1351: and in unison to effectively
1352: emulate a time dependent
1353: magnetic field $\vec{\delta h}_{eff} \propto \hat{e}_{z} \cos \phi(t)$
1354: along the z-axis in the spin equation of motion,
1355: $d \langle \vec{S} \rangle /dt = ... + \langle \vec{S} \rangle
1356: \times \vec{\delta h}_{eff}$. Both of these effects were
1357: noted in \cite{bulaevskii}. In Eq.(\ref{Iqu-cl*}),
1358: we explicitly see their origin. The uniform contraction
1359: is triggered by an entanglement of the tunneling electrons
1360: with our $S=1/2$ particle. We now very briefly elaborate
1361: on the physics of this statement for the benefit of
1362: general readers. The expectation values $\langle \vec{S} \rangle$
1363: amount to weighted sums over all possible states $| \psi \rangle$
1364: (see Eq.(\ref{entangled_eq})).
1365: In any pure (i.e. unentangled) state
1366: of a spin-1/2 problem, the sum
1367: $[\langle S_{x} \rangle^{2} + \langle S_{y} \rangle^{2}
1368: + \langle S_{z} \rangle^{2}] = 1/4$- the spin expectation
1369: values lie on the Bloch sphere. Entanglement
1370: in a spin-1/2 problem such as ours is marked
1371: by a contraction, $[\langle S_{x} \rangle^{2} + \langle S_{y} \rangle^{2}
1372: + \langle S_{z} \rangle^{2}] < S^{2} = 1/4$. Any single
1373: spin expectation value within the Bloch sphere,
1374: $|\langle \psi | \vec{S} | \psi \rangle| < S$,
1375: denotes an expectation value computed in a multi-particle state
1376: $|\psi \rangle$
1377: which is entangled. In the case here $|\psi \rangle$
1378: spans the single spin and the tunneling electrons.
1379: Such a time dependent contraction in the norm
1380: of $\langle \vec{S} \rangle$
1381: relative to the Bloch radius is evident in our exact solution
1382: of Eqs.(\ref{SF},\ref{Uxy}, \ref{pform}, \ref{qform},
1383: \ref{deltacorrection}, \ref{deltaSz}).
1384:
1385:
1386: $\bullet$ {\underline{\em Nonlinear planar precession:}}
1387:
1388: A notable facet of the dynamics given by
1389: the effects discussed above
1390: are non-uniform planar precessions. We find that
1391: within the plane transverse to
1392: the applied field direction, the azimuthal
1393: angle describing the spin orientation,
1394: $\varphi(t) = \tan^{-1}(\langle S_{y}(t) \rangle/\langle S_{x}(t) \rangle)$
1395: is no longer a linear in time. This effect
1396: bears, once again, strong semblance to nutations in classical
1397: rigid body dynamics. In the Larmor problem
1398: of a free spin in a magnetic field,
1399: $\varphi(t) = \omega_{L} t$. In our case, the precession
1400: about the applied field direction is no
1401: longer uniform. Its form is encapsulated
1402: in Eqs.(\ref{sxsoln},\ref{AB})
1403: or, alternatively, by Eqs.(\ref{SF}, \ref{Uxy},\ref{pform}.\ref{qform}).
1404: Once again, mathematically, the origins of this
1405: effect are rooted in the effective planar (xy) components
1406: of the effective force $\langle \vec{I} \rangle$ appearing
1407: in Eq.(\ref{setxyz}). The explicit form of this
1408: effective force is given by
1409: the sum of the two integrals evaluated
1410: in Eqs.(\ref{Icl-cl*}, \ref{Iqu-cl*})
1411: and whose origin explicitly lies, once again,
1412: in the same non-local in time correlations
1413: borne by the superconducting correlations.
1414: %Unlike the driving force $\langle I_{z} \rangle$
1415: %which led to spin nutations, as
1416: %seen from Eqs.(\ref{class_final},\ref{largeSform})
1417: %contained terms originating from $S^{2}$ expressions
1418: %in $\langle \vec{I}_{cl-cl} \rangle$, the planar
1419: %driving force $\langle \vec{I}_{\perp} \rangle$ contains
1420: %terms originating only from $\langle \vec{I}_{qu-cl} \rangle$
1421: %which, as seen from Eq.(\ref{Iqu-cl*}) scales as $S$
1422: %in this spin-1/2 problem. Although, we do not elaborate
1423: %on expressions beyond $S=1/2$ here, a detailed analysis
1424: %indeed reveals that several effects stemming from
1425: %the planar driving force $\langle \vec{I}_{\perp} \rangle$
1426: %vanish in the large $S$ limit. Especially noteworthy is
1427: %the work of \cite{bulaevskii} where current
1428: %variations which were shown to exist in
1429: %the $S=1/2$ problem are absent in the
1430: %$S \gg 1$ system \cite{nut03}.
1431:
1432: In summary, all of the above qualitative findings
1433: for the problem a single $S=1/2$ spin inserted in
1434: a Josephson junction are made vivid
1435: in our ${\cal{O}}(T_{1}^{2})$ exact solution.
1436: From Eqs.(\ref{SF},\ref{Uxy},\ref{pform}, \ref{qform}) for
1437: the planar spin components and from Eqs.(\ref{deltacorrection},
1438: \ref{deltaSz}) for the longitudinal spin we clearly see how
1439: all of these effects come into play.
1440:
1441:
1442: \section{Retarded Correlations in General Spin $S$
1443: Dynamics in a Josephson Junction}
1444: \label{retS}
1445:
1446:
1447: The equation of motion, Eq.(\ref{ud}),
1448: is valid for all spins $S$. Much
1449: of our formalism follows with
1450: no change. We now examine the
1451: integrals $\langle \vec{I} \rangle$
1452: in the general spin $S$ problem.
1453:
1454: We find that for a spin of size $S$,
1455: the integral $\langle \vec{I}_{qu-cl} \rangle$ undergoes
1456: no change relative to its $S=1/2$ form- Eq.(\ref{Iqu-cl*}) remains the same.
1457: The associated physics fleshed out in the second equality
1458: of Eq.(\ref{Iqu-cl*}) which was described in the previous section
1459: (spin contractions and the presence of an effective longitudinal field)
1460: undergoes no change for the general spin $S$ case.
1461:
1462: Next, we evaluate $\langle \vec{I}_{cl-cl} \rangle$.
1463: For large spins, $S \gg 1$, the product $\langle \vec{S} (t) \times
1464: \vec{S}(t^{\prime}) \rangle$ is well approximated by the
1465: vector product of averages $\langle \vec{S} (t)\rangle \times
1466: \langle \vec{S}(t^{\prime}) \rangle$. Then, the approach of
1467: Ref.~\cite{nut03} is well justified and we can obtain the {\em
1468: Josephson nutations} of a big spin. For any $S$ and $t'>t$ we
1469: obtain
1470: \begin{eqnarray}
1471: &&\langle \vec{S}_{cl}(t^{\prime}) \times \vec{S}_{cl}(t)
1472: \rangle_{S}
1473: \nonumber \\
1474: &=& \frac{1}{2} \Big \langle \big[\{S_{y}(t),S_{z}(t)\}_{+}\,
1475: [\cos\omega_L(t'-t)-1]
1476: \nonumber \\
1477: &-&\{S_{x}(t),S_{z}(t)\}_+\,\sin\omega_L(t'-t)\big]\,
1478: \hat{e}_{x} \nonumber \\
1479: &+&\big[\{S_{x}(t),S_{z}(t)\}_{+}\,[1-\cos\omega_L(t'-t)]
1480: \nonumber \\
1481: &-&\{S_{y}(t),S_{z}(t)\}_+\,\sin\omega_L(t'-t)\big]\,
1482: \hat{e}_{y} \nonumber \\
1483: &+&(2S_{x}^2(t)+2S_{y}^2(t))\,\sin\omega_L(t'-t)\,\hat{e}_{z}\
1484: \Big \rangle,
1485: \label{SIcl-cl}
1486: \end{eqnarray}
1487: where $\{...\}_+$ denotes an anticommutator. In the $S=1/2$ case, the
1488: integral stemming from these vectorial product
1489: correlations is parallel to the z-axis,
1490: $\langle \vec{I}_{cl-cl} \rangle || \hat{e}_{z}$.
1491: For spins of size $S>1/2$, however, as we see from Eq.(\ref{SIcl-cl}),
1492: the planar ($x,y$) components also come to the fore
1493: and lead to retarding correlation $(\beta^{R})$ effects
1494: in Eq.(\ref{class_final}). Furthermore, the magnitude of
1495: the driving force $\langle I_{cl-cl;z} \rangle$ along
1496: the z-axis, much unlike the $S=1/2$ case is
1497: time dependent.
1498:
1499: For general spins of size $S > 1/2$, both retarded ($\beta^{R}$)
1500: and Keldysh ($\beta^{K}$) are non-zero along any spin
1501: direction. All of the effects discussed in subsection(\ref{consequences})
1502: are present.
1503:
1504: It is noteworthy to discuss the scaling of all terms
1505: with the spin size $S$. As evident from Eq.(\ref{SIcl-cl}),
1506: the integral $\langle \vec{I}_{cl-cl} \rangle$ spawned
1507: by retarded correlations scales as $S^{2}$. Similarly,
1508: as seen from Eq.(\ref{Iqu-cl*}), whose form holds
1509: for arbitrary $S$, the effective driving force
1510: $\langle \vec{I}_{qu-cl} \rangle$ generated by
1511: Keldysh correlations $(\beta^{K}$) scales
1512: as $S$, i.e. $\langle \vec{I}_{qu-cl} \rangle ~ ~
1513: \alpha ~ ~S$. Thus, for large spins $S \gg 1$,
1514: the retarded contributions overwhelm stochastic Keldysh
1515: contributions. In the classical limit, $S \to \infty$,
1516: only the retarded contributions remain \cite{nut03}.
1517:
1518: \section{Spin triggered ac effects}
1519: \label{AC_effects.}
1520:
1521: Thus far our discussion centered on a Josephson
1522: junction for a time independent potential difference $V$
1523: (dc voltage bias) between the two superconducting leads for which $\phi(t) =
1524: \omega_{J} t$ with $\omega_{J} = 2eV$.
1525:
1526: We now briefly sketch matters for an ac voltage bias
1527: wherein the potential drop is oscillatory in time
1528: and the corresponding
1529: phase difference is $\phi(t) = A \sin \Omega t$.
1530: To make the physics more transparent, we
1531: omit any dc contributions to the voltage
1532: (and thus linear in time contributions to the phase).
1533: This serves as a caricature of rf driven Josephson junctions
1534: known to exhibit the famous Shapiro steps \cite{shapiro}.
1535:
1536: The setup is given by Fig.~\ref{FIG:SETUP}
1537: for a spin $S=1/2$ particle yet now with
1538: an ac voltage applied across the junction.
1539: In the sections that follow, we will resume our central
1540: focus on the constant voltage drop case, $\phi(t) =
1541: \omega_{J} t$. Only in this short section
1542: do we analyze an applied ac voltage bias.
1543:
1544: The calculations for the ac voltage bias case parallel the analysis
1545: of the previous sections. First, we express all
1546: terms by pure harmonics. This is readily achieved by
1547: relying on the identity
1548: \begin{eqnarray}
1549: e^{iC \sin x} = \sum_{n} J_{n}(C) e^{inx},
1550: \end{eqnarray}
1551: with $\{J_{n}(C)\}$ Bessel functions.
1552: The factor $j(t,t^{\prime})$ of Eq.(\ref{Stunnel*}) and thereafter
1553: now becomes
1554: \begin{eqnarray}
1555: j(t,t^{\prime}) = \sum_{n,m} J_{n}(\frac{A}{2}) J_{m}(\frac{A}{2})
1556: \cos[\Omega(nt+mt^{\prime})].
1557: \end{eqnarray}
1558: The analog of Eq.(\ref{Icl-cl*})
1559: for the ac voltage bias case is
1560: \begin{eqnarray}
1561: \langle \vec{I}_{cl-cl} \rangle_{S} =
1562: - |T_{1}|^{2} \hat{e}_{z} \sum_{k,p} \frac{\Delta^{2}}{E_{k} E_{p}}
1563: \sum_{n,m} J_{n}(\frac{A}{2}) J_{m}(\frac{A}{2}) \nonumber
1564: \\ \times \frac{2m \Omega \omega_{L} \sin \Omega (n+m)t}{(E_{k}+ E_{p})^{3}}.
1565: \end{eqnarray}
1566: Further resonant (delta function)
1567: terms make an appearance for $m \ggg 1$.
1568:
1569: Similarly, the analog of Eq.(\ref{Iqu-cl*})
1570: reads
1571: \begin{eqnarray}
1572: \! \! \! \!\! \! \! \!\! \! \! \!\! \! \! \!\! \! \! \!\! \! \! \!\! \! \! \!\! \! \! \!
1573: \langle \vec{I}_{qu-cl} \rangle_{S} = - |T_{1}|^{2}
1574: \sum_{n,m,k,p} \frac{\ J_{n}(\frac{A}{2}) J_{m}(\frac{A}{2})
1575: \Delta^{2}}{E_{k} E_{p}(E_{k}+ E_{p})^{2}} \nonumber
1576: \\ \times [ (2 m \Omega \langle S_{x}(t) \rangle \sin (n+m) \Omega t
1577: - \langle S_{y}(t) \rangle \omega_{L}
1578: \cos (n+m) \Omega t) \hat{e}_{x} \nonumber
1579: \\ + ( 2 m \Omega \langle S_{y}(t) \rangle \sin (n+m) \Omega t
1580: + \langle S_{x}(t) \rangle \omega_{L} \cos(n+m)
1581: \Omega t) \hat{e}_{y} \nonumber
1582: \\ \! \! \! \!\! \! \! \!\! \! \! \!\! \! \! \!
1583: + 2 m \Omega \langle S_{z}(0) \rangle \sin (n+m) \Omega t
1584: \hat{e}_{z}]. \nonumber
1585: \end{eqnarray}
1586: To ${\cal{O}}(|T_{1}|^{2})$, the nutations are given by
1587: \begin{eqnarray}
1588: \langle \delta S_{z}(t) \rangle
1589: = |T_{1}|^{2} \sum_{k,p} \frac{\Delta^{2}}{E_{k}E_{p}}
1590: \sum_{n+m \neq 0} J_{n}(\frac{A}{2}) J_{m}(\frac{A}{2}) \nonumber
1591: \\ \times \frac{2m \Omega}{(E_{k}+ E_{p})^{2}}
1592: (\frac{\omega_{L}}{E_{k}+ E_{p}} + \langle S_{z}(0) \rangle) \nonumber
1593: \\ \times \frac{1 - \cos \Omega(n+m) t}{n+m}.
1594: \label{deltaSzAC}
1595: \end{eqnarray}
1596: Higher order effects further enhance this response.
1597: Eq.(\ref{deltaSzAC}) is the ac voltage bias analog
1598: of Eq.(\ref{deltaSz}) for the
1599: dc voltage bias case.
1600: The seminal feature of our results is the existence
1601: of frequencies in the spin dynamics
1602: of all integer multiples of the voltage bias driving
1603: ac frequency $\Omega$.
1604: As the spin alters (via back-action effects) the tunneling
1605: supercurrent, the supercurrent will exhibit
1606: oscillations at all frequencies $\omega_{r} = r \Omega$
1607: with $r$ an integer. Extending the results of
1608: \cite{bulaevskii} to this problem,
1609: the supercurrent
1610: \begin{eqnarray}
1611: \langle I(t) \rangle = \sin \phi(t)
1612: \Big[2 \pi^{2} e \rho^{2} \Delta(|T_{0}|^{2}
1613: - \frac{3}{4} |T_{1}|^{2}) \nonumber
1614: \\ + 4e |T_{1}|^{2} \rho^{2} h
1615: \langle S_{z}(t) \rangle \Big],
1616: \label{I_shapiro}
1617: \end{eqnarray}
1618: where $\rho$ is the spin density of states within the leads, with
1619: the spin given by Eqs.(\ref{deltacorrection}, \ref{deltaSzAC})
1620: with the Larmor $\langle S_{z}(t) \rangle_{0} = \langle S_{z}(0) \rangle$.
1621: %Here, we further include the gauge
1622: %invariant
1623: %\begin{eqnarray}
1624: %\tilde{\phi} = \phi - \frac{2e}{\hbar c} \int \vec{A}
1625: %\cdot \vec{dl} = \theta_{L} - \theta_{R} -
1626: %\frac{2e}{\hbar c} \int_{L}^{R} \vec{A}
1627: %\cdot \vec{dl},
1628: %\label{gp}
1629: %\end{eqnarray}
1630: %with $\theta_{L,R}$
1631: %the phases of the left/right superconducting
1632: %condensates, and the vector potential $\vec{A}$
1633: %further includes the faint external magnetic field $\vec{B}$.
1634:
1635:
1636:
1637:
1638:
1639:
1640:
1641:
1642:
1643:
1644:
1645:
1646:
1647: \section{Ferromagnets
1648: In Josephson Junctions}
1649: \label{ferroS}
1650:
1651: We now investigate what transpires when
1652: ferromagnets (instead of a single spin) are immersed
1653: between two s-wave superconductors with a dc bias voltage applied
1654: across the junction (as illustrated in Fig.(\ref{FIG:MAGNET})).
1655: As in the single spin problem, the full problem
1656: involves both the back-action of the spin on the
1657: phase of the superconductors (ignored
1658: here) and the spin dynamics sparked by the tunneling
1659: current (which we focus on below). Further, for extended
1660: junctions, phasons naturally appear.
1661: In what follows, we assume that the phases of the two superconductors
1662: surrounding a single magnetic slab
1663: have a spatially uniform phase difference $\phi(t)$.
1664: The tunneling action amounts to
1665: a sum over individual tunneling actions
1666: through each of the individual spins labeled
1667: by their sites $\vec{r}$,
1668: \begin{eqnarray}
1669: S_{tunnel} \simeq \nonumber
1670: \\ 4 \sum_{\vec{r}}
1671: \int dt \int dt^{\prime} \beta^{R}(t,t^{\prime})
1672: T_{1}^{2} \vec{S}_{qu}(\vec{r},t) \cdot
1673: \vec{S}_{cl}(\vec{r},t^{\prime}) j(t,t^{\prime}) \nonumber
1674: \\
1675: + \int dt \int dt^{\prime} T_{1}^{2} \beta^{K}(t,t^{\prime})
1676: \vec{S}_{qu}(\vec{r},t)
1677: \cdot \vec{S}_{qu}(\vec{r},t^{\prime}) j(t,t^{\prime}).
1678: \nonumber
1679: \end{eqnarray}
1680: For ferromagnetic spin chains/planes with
1681: arbitrary exchange constants $J(\vec{r},\vec{r}^{\prime})$,
1682: and scaled external magnetic field $\vec{h}$,
1683: the exact equation of motion reads
1684: \begin{eqnarray}
1685: 0 = \langle \frac{d\vec{S}_{cl}(\vec{r})}{dt} +
1686: \vec{h} \times \vec{S}_{cl}(\vec{r}) + \vec{I}_{cl-cl;r} + \vec{I}_{qu-cl;r}
1687: \nonumber
1688: \\ + \sum_{\vec{r}^{\prime}} J(\vec{r},\vec{r}^{\prime})
1689: \vec{S}_{cl}(\vec{r}^{\prime},t) \times
1690: \vec{S}_{cl}(\vec{r},t) \rangle_{S}.
1691: \label{class_final1}
1692: \end{eqnarray}
1693:
1694: \begin{figure}
1695: \centerline{\includegraphics[width=0.8\columnwidth]{setup-slab.eps}}
1696: \caption{A ferromagnetic slab inserted between two superconducting leads.
1697: The entire system is subjected to a weak external magnetic field $B$.
1698: A schematic of the precessing spins is shown.}
1699: \label{FIG:MAGNET}
1700: \end{figure}
1701:
1702: It is hard not to notice a resemblance between
1703: the single spin problem (Eq.(\ref{class_final}))
1704: and the problem of the ferromagnet (Eq.(\ref{class_final1})).
1705: Indeed, as we will shortly demonstrate the spin wave dynamics
1706: in the ferromagnet within a Josephson junction bears
1707: much in common with the single spin problem
1708: with the proviso that the various ferromagnetic
1709: spin waves feel an effective momentum dependent
1710: magnetic field of strength $h_{eff} = h +S[J(\vec{k})-J(0)]$
1711: with $J(\vec{k})$ the Fourier transform of the
1712: two spin interaction $J(\vec{r}, \vec{r}^{\prime})$.
1713:
1714: The solution proceeds much the same as for the single spin
1715: problem. Henceforth, we discuss
1716: the qualitative physics
1717: expected. Unlike the precise
1718: solutions presented till now,
1719: what follows is a quick
1720: qualitative sketch by way
1721: of an analogy. An exact solution
1722: will be detailed elsewhere \cite{long}.
1723:
1724: Transforming from spin variables to bosonic operators ($b(\vec{r})$)
1725: at all lattice sites $\vec{r}$,
1726: \cite{DM}
1727: \begin{eqnarray}
1728: S^{+}(\vec{r}) = b^{\dagger}(\vec{r})\sqrt{2S}, \nonumber
1729: \\ S^{-}(\vec{r}) = [b(\vec{r})
1730: - \frac{1}{2S} b^{\dagger}(\vec{r})
1731: b(\vec{r}) b(\vec{r})]\sqrt{2S}, \nonumber
1732: \\ S_{z} = - S + b^{\dagger}(\vec{r}) b(\vec{r}).
1733: \label{dm}
1734: \end{eqnarray}
1735: Sans the ${\cal{O}}(|T_{1}|^{2}$)
1736: tunneling part of the action, the action is
1737: quadratic in the bosonic operators and is readily
1738: diagonalized in $\vec{q}$ space. We find that
1739: the free part of the action
1740: \begin{eqnarray}
1741: S_{0} = - \int dt \int \frac{d^{d}q}{(2 \pi)^{d}} \{S[J(\vec{q})- J(0)]
1742: + h \} \nonumber
1743: \\ \times b^{*}(\vec{q}) b(\vec{q}),
1744: \end{eqnarray}
1745: with $d$ the dimension of the inserted magnet.
1746: (As the problem is ferromagnetic, $J(0) = \min_{\vec{q}} \{ J(\vec{q})\}$).
1747: Comparing this action to the one appearing in the single
1748: spin problem, we find that to Gaussian order the spin-wave
1749: problem is identical to the dynamics of a single spin
1750: with the replacement
1751: \begin{eqnarray}
1752: h \to h_{eff}(\vec{q}) \equiv \{S[J(\vec{q})- J(0)]
1753: + h\}.
1754: \label{heff}
1755: \end{eqnarray}
1756: The quadratic contribution
1757: of the ${\cal{O}}(|T_{1}|^{2})$ portion of the action
1758: involving non-local in time correlations
1759: has precisely the same form for both the single
1760: spin problem and for each mode $\vec{q}$
1761: of the spin-wave problem.
1762: Thus, the quadratic in $b$, ${\cal{O}}(|T_{1}|^{2})$ corrections
1763: to the spin dynamics are given by Eqs.(\ref{deltaSz},\ref{sxsoln})
1764: with the replacement of Eq.(\ref{heff}).
1765:
1766: For instance, the above analogy suggests that
1767: that the net ferromagnetic moment variation
1768: in $S=1/2$ ferromagnets is
1769: \begin{eqnarray}
1770: \frac{\delta M}{V} \nonumber
1771: \\ = |T_{1}|^{2} (1- \cos \omega_{J} t)
1772: [\sum_{k,p} \frac{\Delta^{2}
1773: \omega_{L}}{E_{k} E_{p}(E_{k}+ E_{p})^{3}} \nonumber
1774: \\ + \frac{M}{V} \sum_{k,p} \frac{\Delta^{2}}{E_{k}E_{p}
1775: (E_{k}+E_{p})^{2}}],
1776: \label{deltaSzq}
1777: \end{eqnarray}
1778: with $V$ the volume of the magnet
1779: and $M$ its magnetization sans the
1780: supercurrent.
1781: Alternatively, the analysis may
1782: parallel the derivation of the previous
1783: sections word for word while taking
1784: the unperturbed solution
1785: (the analogue of the Larmor
1786: solution of Eq.(\ref{Larmoreq}))
1787: to be a spin wave and computing all
1788: corrections to ${\cal{O}}(T_{1}^{2})$.
1789:
1790: In the continuum limit,
1791: \begin{eqnarray}
1792: h_{eff}(\vec{q}) = \frac{\rho_{s}}{m_{0}} q^{2} + h,
1793: \end{eqnarray}
1794: with $m_{0} \equiv S/v$ (where $v$ is the volume per site),
1795: the magnetization density of the ground state,
1796: and $\rho_{s}$ the spin stiffness.
1797:
1798: This $h \to h_{eff}(\vec{q})$ correspondence
1799: applies to any property inherited by the
1800: single spin dynamics in the Josephson
1801: junction. In particular, in \cite{bulaevskii}
1802: it was beautifully shown how spin dynamics may alter
1803: the super-current in the Junction. The current
1804: may be computed by the likes of Eq.(\ref{i(t)}).
1805: Extending these results to a ferromagnet inserted in
1806: a Josephson junction by the correspondence of
1807: Eq.(\ref{heff}), we find that the new
1808: spin wave dynamics leads to
1809: the supercurrent,
1810: \begin{eqnarray}
1811: \langle I(t) \rangle = \sin \phi(t) \int\frac{d^{d}q}{(2 \pi)^{d}}
1812: \Big[2 \pi^{2} e \rho^{2} \Delta(|T_{0}|^{2}
1813: - \frac{3}{4} |T_{1}|^{2}) \nonumber
1814: \\ + 4e |T_{1}|^{2} \rho^{2} h_{eff}(\vec{q})
1815: \langle S_{z}(\vec{q}) \rangle \Big],
1816: \label{I_quad}
1817: \end{eqnarray}
1818: with $\rho$ the spin density of states within the leads.
1819: A matching of the Josephson and spin frequencies
1820: (such as present here for variations
1821: in the low temperature magnetic
1822: moment (see Eqs.(\ref{deltaSz},\ref{deltaSzq})))
1823: leads to a DC signal; additional
1824: harmonics further appear.
1825: We emphasize that in the above we compared
1826: only the Gaussian portion in the Bose fields.
1827: Higher order (non-Gaussian) terms originating from Eq.(\ref{dm})
1828: as well as phasons alter the natural correspondence of Eq.(\ref{heff}).
1829: A full discussion of these issues will be detailed elsewhere \cite{long}.
1830:
1831: \section{Other Geometries}
1832: \label{Geometry.}
1833:
1834:
1835: If phason
1836: contributions are neglected, then by a trivial change of geometry all
1837: of our results thus far, will apply
1838: for other systems as well. For instance, by replacing one of
1839: the superconducting leads
1840: by a surface, the resulting
1841: system may emulate a superconducting
1842: tip coupled to superconducting surface
1843: through a single spin or a ferromagnet.
1844: Here, all of the results of Sections(\ref{single_spin},\ref{ferroS})
1845: for the spin dynamics and tunneling current hold.
1846:
1847:
1848: Similarly, by replacing both superconducting
1849: leads by surfaces and examining a magnetic
1850: layer inserted in between, the resultant system
1851: looks much alike a layered superconducting/magnetic
1852: system. In this system, the results of
1853: Section(\ref{ferroS}) apply.
1854:
1855:
1856:
1857: \section{Large $S$ Adiabatic Approximations}
1858: \label{Large_S.}
1859:
1860: Thus far we studied the
1861: dynamics of single spins and
1862: of ferromagnets. In \cite{nut03},
1863: the large $S$ limit of the
1864: single spin problem was
1865: studied. In that work,
1866: several approximations
1867: were made:
1868:
1869: (i) The perturbative approach that
1870: we employed in the current article
1871: which allows an exact
1872: evaluation of all pertinent integrals
1873: to low orders was replaced
1874: by an ``adiabatic'' approximation
1875: wherein the slow dynamics
1876: of the spin vis a vis electronic
1877: processes was explicitly incorporated,
1878: $\vec{S}(t^{\prime}) \simeq \vec{S}(t) + (t^{\prime} - t)
1879: (d\vec{S}/dt)$.
1880:
1881: (ii) The (``classical'') large $S$ limit allowed us to
1882: omit many instances of $\vec{S}_{qu}$ in the equations
1883: of motion and only $\beta^{R}$ related contributions
1884: in the tunneling action were consequential. Furthermore,
1885: as briefly alluded to earlier, in this limit,
1886: the average of the vectorial product
1887: $\langle \vec{S}(t^{\prime}) \times \vec{S}(t) \rangle$
1888: is equal to the product of the averages $\langle \vec{S}(t^{\prime}) \rangle
1889: \times \langle \vec{S}(t) \rangle$. Correspondingly, any expectation
1890: value braces may be omitted. Thus,
1891: we may replace any expectation value $\langle {\cal{A}} \rangle$
1892: by ${\cal{A}}$ itself.
1893:
1894: The advantage of this method is that
1895: furnishes an elegant non-perturbative closed
1896: form solution for the spin dynamics.
1897: We will not repeat the results
1898: for the single spin cluster
1899: $(S \gg 1 )$ problem here and rather
1900: refer the reader to \cite{nut03}.
1901: We now briefly comment on applications
1902: of this method to other systems.
1903:
1904:
1905: To ${\cal{O}}(|T_{1}|^{2})$,
1906: the spin wave dynamics
1907: in ferromagnets
1908: may be attained via the
1909: substitution of Eq.(\ref{heff}).
1910: Equivalently, the spin wave equations
1911: of motion may also be determined directly
1912: when applying the adiabatic approximation
1913: on Eq.(\ref{class_final1}).
1914: We then find
1915: \begin{eqnarray}
1916: \frac{d\vec{S}_{cl}(\vec{r}_i)}{dt} +
1917: \vec{h} \times \vec{S}_{cl}(\vec{r}_{i})
1918: + \sum_{j} J_{ij} \vec{S}_{cl}(\vec{r}_{j},t) \times
1919: \vec{S}_{cl}(\vec{r}_{i},t) \nonumber
1920: \\ + \kappa \vec{S}_{cl}
1921: \times \frac{d \vec{S}_{cl}}{dt}
1922: \sin \omega_{J} t =0
1923: \end{eqnarray}
1924: with $\kappa \equiv \sum_{k,p} \frac{|\Delta|^{2} |T_{1}|^{2}}{E_{k} E_{p}}
1925: [(E_{k}+E_{p}-eV)^{-2}-(E_{k}+E_{p}+eV)^{-2}]$.
1926: The appropriate spin wave equation is
1927: \begin{eqnarray}
1928: \frac{d b(\vec{q})}{dt} = i[h+S\{J(\vec{q}) - J(0) \}]
1929: b(\vec{q}) \nonumber
1930: \\ + \kappa \partial_{t} b(\vec{q},t) \sin \omega_{J} t.
1931: \label{swk}
1932: \end{eqnarray}
1933: The solution to Eq.(\ref{swk}) is
1934: \begin{eqnarray}
1935: b(\vec{q}, t) = b(\vec{q},0) \exp
1936: [ - \frac{2i(S\{J(\vec{q}) - J(0) \}
1937: +h)}{\omega_{J} \sqrt{1 - \kappa^{2}}} \nonumber
1938: \\ \times
1939: \{ \tan^{-1} (\frac{\kappa}{\sqrt{1-\kappa^{2}}}) - \tan^{-1}(\frac{\kappa-
1940: \tan(\omega_{J} t/2)}{\sqrt{1 - \kappa^{2}}})\}],
1941: \label{nonuniform}
1942: \end{eqnarray}
1943: which is quite different from the standard spin-wave evolution
1944: in a magnet outside a Josephson junction.
1945: The key feature is a nonuniform evolution
1946: of each spin-wave. Similar to the azimuthal precession
1947: of a single spin, the planar components $S_{x,y}$
1948: precess as the real and imaginary parts of $\exp[i \varphi(t)]$
1949: with a nonlinear $\varphi(t)$. Thermodynamic
1950: quantities computed via the corrected bosonic spin-wave dispersion
1951: exhibit corrections.
1952:
1953: Similarly, we may examine the adiabatic large $S$ equations of motion
1954: for an antiferromagnetic spin chain oriented along the z-axis
1955: in a Josephson junction (just as in Fig.(\ref{FIG:MAGNET}) yet
1956: now with a single antiferromagnetic
1957: spin chain replacing the ferromagnetic slab
1958: in an $h=0$ background). We then find that that
1959: the staggered spin, $\vec{\tilde{S}}_{i} \equiv (-1)^{i} \vec{S}_{i}$
1960: (with the integer $i$ the spin site location along the chain) satisfies
1961: \begin{eqnarray}
1962: 0= \Box \vec{\tilde{S}}_{cl}(t) + 3 \partial_{t} \vec{\tilde{S}}_{cl}(t)
1963: \times \partial_{z} \vec{\tilde{S}}_{cl}(t) \nonumber
1964: \\ + \kappa
1965: \partial_{t} \vec{\tilde{S}}_{cl}(t) \sin \omega_{J} t,
1966: \label{theta}
1967: \end{eqnarray}
1968: where
1969: $\Box \equiv
1970: \frac{v_{s}}{g} [ \partial_{z}^{2} - \frac{1}{v_{s}^{2}}
1971: \partial_{t}^{2}]$.
1972: Here, $g=2/S$ and the spin wave velocity $v_{s} = 2aJS$,
1973: with $a$ is the lattice constant.
1974: The role of the supercurrent
1975: as an effective driving term is evident in the
1976: last line of Eq.(\ref{theta}).
1977:
1978:
1979:
1980:
1981: \section{Conclusions}
1982: \label{conc.}
1983:
1984:
1985: In conclusion, our work addresses new dynamical
1986: effects exhibited by spins in Josephson Junctions.
1987: En route, many features (general and specific) were
1988: found:
1989: \bigskip{ }
1990:
1991: {\bf(1)} We derived the
1992: {\em exact} equation of motion for
1993: spin systems on Keldysh contours.
1994:
1995: {\bf(2)} The $S=1/2$ spin dynamics
1996: of a single spin in a Josephson junction
1997: was investigated and a perturbative
1998: solution was given. {\em Spin-1/2 Josephson
1999: nutations are predicted.}
2000:
2001: {\bf(3)} Spin dynamically triggered ac effects
2002: are predicted.
2003:
2004: {\bf(4)} The spin wave dynamics {\em of a ferromagnet
2005: in between two superconducting leads} was
2006: investigated. We predict non-trivial
2007: spin wave dynamics as well as
2008: new manifestations of
2009: this dynamics (most
2010: notably in the supercurrent).
2011:
2012: {\bf(5)} Large $S$ expressions were discussed
2013: for ferromagnetic slabs and antiferromagnetic
2014: spin chains in a Josephson junction.
2015:
2016:
2017: \bigskip{ }
2018:
2019:
2020: {\bf Acknowledgments}
2021:
2022:
2023: This work was supported by the US DOE under LDRD X1WX (ZN and AVB).
2024: We gratefully acknowledge discussions with Yu Makhlin.
2025:
2026: \subsection{Appendix A: Detection}
2027:
2028:
2029: The non-trivial spin-wave and associated supercurrent
2030: in Josephson junctions containing ferromagnets
2031: (section \ref{ferroS}) may be seen more readily seen
2032: than those of single spins. The spin dynamics
2033: may be discerned by measuring the
2034: magnetization of the ferromagnetic
2035: slab as a function of time (as
2036: suggested by Eq.(\ref{deltaSzq})) as well
2037: as by monitoring the supercurrent
2038: (given by Eq.(\ref{I_quad})).
2039: Other techniques may involve standard
2040: measurements of microwave radiation from the junction
2041: (and backaction effects). The
2042: magnitudes of these effects will be
2043: studied elsewhere \cite{long}.
2044:
2045:
2046: We now briefly review a detection scheme discussed in \cite{nut03},\cite{snz}
2047: for the Josephson nutations for the $S \gg 1$ limit of
2048: the general spin S results of Section(\ref{retS}).
2049: This corresponds to a single magnetic cluster.
2050:
2051:
2052: As it moves, the spin cluster magnetic moment generates a time-dependent magnetic field, $\delta{\vec B}({\bf
2053: r},t) = \frac{\mu_{0}}{4 \pi} [3 {\vec r} ({\vec r} \cdot {\vec m}(t))
2054: - r^{2} {\vec m}(t)]/r^{5}$. This small field is superimposed against
2055: the constant external field background (${\vec B}$).
2056: In the above, ${\vec r}$ is the position relative to the spin and
2057: ${\vec m}(t)$ is the magnetic moment of the spin.
2058: A ferromagnetic cluster of spin $S = 100$ generates a detectable magnetic field
2059: $\delta B \sim 10^{-10}$~T at a distance of a micron
2060: away from the spin. For a SQUID loop of micron dimensions located
2061: at that position, the ensuing flux variation
2062: $\delta \Phi \sim 10^{-7} \Phi_{0}$ (with
2063: $\Phi_{0} = hc/e$ the flux quantum) are within reach of modern SQUIDs. In
2064: such a setup, with $T_{1}/T_{0} \sim 0.1$, the typical critical
2065: Josephson current
2066: is $J_{S}^{(0)} \sim 10 \;\mu\mbox{A}$,
2067: $\vert\Delta\vert=1\;\mbox{meV}$, and $eV \sim 10^{-3}
2068: |\Delta|$, we find that the relative corrections $ \delta S/S \sim 0.1$. The spin
2069: components orthogonal to ${\vec B}$ vary, to ${\cal{O}}(T_{1}^{2})$,
2070: with Fourier components at frequencies $|\omega_{L} \pm
2071: \omega_{J}|$ ($\omega_L= g \mu_{B} B$), leading to a observable signal in the magnetic
2072: field ${\vec B} + \delta {\vec B}$. For a field $B \sim 200$~G, $\omega_{L} \sim
2073: 560\; \mbox{MHz}$, and a new side band will
2074: appear at $|\omega_{L} - \omega_{J}|$, whose magnitude may be tuned
2075: to 10--100~MHz. This measurable frequency is
2076: easily distinguished from the Larmor frequency $\omega_{L}$.
2077:
2078: The efficiency of this detection scheme may be enhanced by
2079: embedding the spin in one of the Josephson junction arms of the
2080: SQUID itself. Such a setup is illustrated in Fig.~\ref{FIG:DETECTION}.
2081: The Josephson junction harboring the spin is employed
2082: in both triggering the nutations and, along with the second
2083: junction of the SQUID, in the detection of the resulting nutations.
2084: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2085: \begin{figure}
2086: \centerline{\includegraphics[width=0.8 \columnwidth]{detection.eps}}
2087: \caption{A SQUID-based detection scheme. The SQUID monitors the magnetic field
2088: produced by the magnetic cluster in one of the junctions.}
2089: \label{FIG:DETECTION}
2090: \end{figure}
2091: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2092:
2093: \subsection{Appendix B: Time Ordering Along the Keldysh
2094: Contour}
2095:
2096: In averaging within the path integral formalism,
2097: we immediately attain time ordered averages.
2098: In interchanging the order of the spins (if necessary)
2099: in the vectorial product upon time ordering
2100: within the path integral $CP_{1}$
2101: formulation a change of sign is incurred \cite{explain}.
2102: We now go over, in some detail, time ordering
2103: within the Keldysh framework. As the time
2104: ordering is performed along the Keldysh
2105: contour, we will denote it by $T_{K}$
2106: as we have done in deriving the
2107: effective action of subsection(\ref{eff_action}).
2108:
2109: Consider the third term in
2110: Eq.(\ref{class_final}). Upon time ordering,
2111: we find that
2112: \begin{eqnarray}
2113: \langle \vec{S}_{qu}(t^{\prime}) \times \vec{S}_{cl}(t) \rangle_{S} =
2114: \langle T_{K} [\vec{S}_{qu}(t^{\prime}) \times \vec{S}_{cl}(t)]
2115: \rangle \nonumber
2116: \\ = \langle T_{K}[ \frac{1}{2}(\vec{S}_{up}(t^{\prime})
2117: - \vec{S}_{down}(t^{\prime})) \nonumber
2118: \\ \times
2119: (\vec{S}_{up}(t) + \vec{S}_{down}(t))] \rangle.
2120: \label{exampleorder}
2121: \end{eqnarray}
2122: Due to the form of the Keldysh contour
2123: (see Fig.(\ref{FIG:keldysh1})), irrespective
2124: of the values of $t$ and $t^{\prime}$,
2125: $\vec{S}_{up}(t)$ always appears before $\vec{S}_{down}(t^{\prime})$.
2126: Similarly, for $t>t^{\prime}$, $\vec{S}_{up}(t)$ appears after
2127: $\vec{S}_{up}(t^{\prime})$ while $\vec{S}_{down}(t)$ appears before
2128: $\vec{S}_{down}(t^{\prime})$.
2129: With this information at hand,
2130: Eq.(\ref{exampleorder}) leads
2131: to
2132: \begin{eqnarray}
2133: \langle T_{K} [ \vec{S}_{qu}(t^{\prime}) \times
2134: \vec{S}_{cl}(t)] \rangle \nonumber
2135: \\ = - \theta(t- t^{\prime}) [\langle \vec{S}(t^{\prime})
2136: \times \vec{S}(t)
2137: \rangle + \langle \vec{S}(t) \times \vec{S}(t^{\prime}) \rangle].
2138: \end{eqnarray}
2139: The expectation values on the right are the usual operator
2140: expectation values. Here, we disposed of the up/down indices
2141: once we took care
2142: of time ordering. The up/down labels merely
2143: serve as mnemonics for this time ordering
2144: along the Keldysh contour.
2145:
2146: Similarly, we find that
2147: \begin{eqnarray}
2148: \langle T_{K}
2149: [\vec{S}_{cl}(t^{\prime}) \times \vec{S}_{cl}(t)] \rangle = \nonumber
2150: \\ \frac{1}{2} [\theta(t^{\prime}-t) (\langle
2151: \vec{S}(t^\prime) \times \vec{S}(t)
2152: - \vec{S} (t) \times \vec{S} (t^{\prime})) \rangle \nonumber
2153: \\ + \theta(t-t^{\prime}) ( \langle \vec{S} (t)
2154: \times \vec{S}(t^{\prime})
2155: - \vec{S} (t^\prime) \times \vec{S}(t)) \rangle ].
2156: \label{t1}
2157: \end{eqnarray}
2158: By the same token,
2159: $\langle T_{K} [\vec{S}_{qu}(t) \times \vec{S}_{qu}(t^{\prime})] \rangle = 0.$
2160:
2161: As will become clear shortly, in the solution of Eqs.(\ref{class_final}) to
2162: order ${\cal{O}}(T_{1}^{2})$, we will need the spin-spin expectation values
2163: of the usual Larmor problem (i.e. a single spin
2164: in a magnetic field sans any supercurrent).
2165: Here,
2166: \begin{eqnarray}
2167: S_{x}(t) = S_{x}(0) \cos \omega_{L} t + S_{y}(0) \sin \omega_{L} t \nonumber
2168: \\ S_{y}(t) = S_{y}(0) \cos \omega_{L}t
2169: - S_{x}(0) \sin \omega_{L} t, \nonumber
2170: \\ S_{z}(t) = S_{z}(0),
2171: \label{Larmoreq}
2172: \end{eqnarray}
2173: with the external magnetic field
2174: oriented along the positive z axis
2175: and $\omega_{L} = |\vec{h}|$ the Larmor frequency.
2176:
2177: Next, we invoke this solution to compute the various
2178: expectation values within the Larmor problem (i.e.
2179: to order ${\cal{O}}(T_{1}^{0})$). We find that
2180: \begin{eqnarray}
2181: \langle \vec{S}_{qu}(t^{\prime}) \times \vec{S}_{cl}(t) \rangle_{S} \nonumber
2182: \\ =
2183: - \theta(t - t^{\prime})
2184: [i ~ Im \{ \langle \vec{S}(t^{\prime}) \times \vec{S}(t) \rangle \} \nonumber
2185: \\ = -i \theta(t - t^{\prime}) \{[\langle S_{x}(t) \rangle
2186: (1+ \cos \omega_{L}(t^{\prime} -t) ) \nonumber
2187: \\
2188: + \langle S_{y}(t) \rangle \sin \omega_{L}(t^{\prime} -t)]
2189: \hat{e}_{x} \nonumber
2190: \\ + [ \langle S_{y}(t) \rangle (1+ \cos \omega_{L} (t^{\prime} - t))
2191: \nonumber
2192: \\ -
2193: \langle S_{x}(t) \rangle \sin \omega_{L}(t^{\prime} - t)]
2194: \hat{e}_{y} \nonumber
2195: \\ + 2 \langle S_{z}(t) \rangle \cos \omega_{L}(t^{\prime} - t)
2196: \hat{e}_{z}\}.
2197: \label{Ixy}
2198: \end{eqnarray}
2199:
2200: Similarly, for the $S=1/2$ problem,
2201: \begin{eqnarray}
2202: \langle \vec{S}_{cl}(t^{\prime}) \times \vec{S}_{cl}(t) \rangle_{S} \nonumber
2203: \\ =
2204: Re\{ \langle \vec{S}(t^{\prime}) \times \vec{S}(t) \rangle \} \nonumber
2205: \\ =
2206: \frac{1}{2} \sin \omega_{L}(t^{\prime} - t) \hat{e}_{z}.
2207: \label{Iz}
2208: \end{eqnarray}
2209:
2210: In Eqs.(\ref{Ixy},\ref{Iz}), we vividly see that
2211: upon time ordering along the Keldysh contour,
2212: the non-vanishing
2213: spin cross products become simply related to the imaginary and real
2214: parts of $\langle \vec{S}(t^{\prime}) \times \vec{S}(t) \rangle $.
2215:
2216:
2217:
2218: \subsection{Appendic C: Evaluation of Integrals}
2219: \label{integrals}
2220:
2221: We are now ready for the evaluation of the various integrals $I$ that
2222: appear in Eq.(\ref{class_final}) to order ${\cal{O}}(T_{1}^{2})$.
2223:
2224: We start with $\vec{I}_{cl-cl}$. Inserting Eq.(\ref{Iz}) and Eq.(\ref{BetaR})
2225: into Eq.(\ref{class_final}) we find upon invoking
2226: the relation $\phi(t) = \omega_{J} t$
2227: \begin{eqnarray}
2228: \langle \vec{I}_{cl-cl} \rangle_{S} = \nonumber
2229: \\ 4 |T_{1}|^{2} \hat{e}_{z}
2230: \int dt^{\prime} j(t,t^{\prime})
2231: \beta^{R}(t,t^{\prime}) \langle \vec{S}_{cl}(t^{\prime})
2232: \times \vec{S}_{cl}(t) \rangle_{S} \nonumber
2233: \\ = - |T_{1}|^{2} \hat{e}_{z} \sum_{k,p} \frac{\Delta^2}{E_k E_p}
2234: \int dt^{\prime} \theta(t-t') \cos \omega_{J}
2235: \frac{t+t^{\prime}}{2} \nonumber
2236: \\ \times \sin \omega_{L}(t - t^{\prime}) \sin [(E_k+E_p)(t-t')].
2237: \label{intbr}
2238: \end{eqnarray}
2239:
2240: Before evaluating Eq.(\ref{intbr}) exactly, we illustrate
2241: what answer is anticipated.
2242: The underlying observation of this {\em adiabatic} approach
2243: is that, as a consequence of $\omega_{L,J} \ll E_{k,p}$
2244: the spin dynamics is far slower than that of electronic
2245: processes. Thus, in integrals involving both
2246: spin and electronic degrees of freedom, we
2247: may regard the spin as nearly stationary and
2248: approximate $\vec{S}(t^{\prime}) \simeq \vec{S}(t) + (t^{\prime} - t)
2249: (d\vec{S}/dt)$. This physically transparent approximation was
2250: invoked in \cite{nut03}. Employing this approximation
2251: here we anticipate that
2252: $\langle \vec{I}_{cl-cl} \rangle_{S} \simeq C_{1} \hat{e}_{z}
2253: \sin \omega_{L} t$
2254: where $C_{1}$ is, up to trivial prefactors, given by
2255: $\int_{0}^{\infty} d x~ [x^{2} \beta^{R}(x)]$.
2256: Such an anticipation is not far off the mark.
2257:
2258:
2259: Next, we exactly evaluate Eq.(\ref{intbr}) by rewriting products
2260: of trigonometric functions as sums and consequently employing
2261: the identities
2262: \begin{eqnarray}
2263: \int_{0}^{\infty} dx ~\cos a x = \pi \delta(a), \nonumber
2264: \\ \int_{0}^{\infty} dx ~ \sin ax = \pi \delta(a) + \frac{1}{a}.
2265: \label{identityint}
2266: \end{eqnarray}
2267:
2268: In the integrals of interest, $x$ assumes the role of $(t^{\prime}-t)$.
2269: As the applied magnetic field and voltage are far lower
2270: than electronic energy scales, $\omega_{L,J} \ll E_{k,p}$,
2271: we find that all resonances signaled
2272: by the delta functions are physically unaccessible and our
2273: expressions undergo further simplifications. Retaining the leading
2274: order terms in ${\cal{O}}(\omega_{L,J}/(E_{k}+E_{p}))$ we arrive at
2275: \begin{eqnarray}
2276: \langle \vec{I}_{cl-cl} \rangle_{S} = - |T_{1}|^{2} \hat{e}_{z} \sum_{k,p}
2277: \frac{\Delta^{2} \omega_{L} \omega_{J}}{E_{k} E_{p} (E_{k} + E_{p})^{3}}
2278: \sin \omega_{J}t.
2279: \label{Icl-cl*}
2280: \end{eqnarray}
2281: Thus, the form anticipated by the adiabatic approximation
2282: is correct if $C_{1} = - |T_{1}|^{2} \sum_{k,p}
2283: \frac{\Delta^{2} \omega_{L} \omega_{J}}{E_{k} E_{p} (E_{k} + E_{p})^{3}}$.
2284:
2285: Similarly, by inserting Eqs.(\ref{BetaK},\ref{Ixy})
2286: into Eq.(\ref{class_final}) and invoking Eqs.(\ref{identityint}),
2287: we find
2288: \begin{eqnarray}
2289: \langle \vec{I}_{qu-cl} \rangle_{S} = - |T_{1}|^{2}
2290: \sum_{k,p} \frac{|\Delta|^{2}}{E_{k} E_{p} (E_{k}+E_{p})^{2}} \nonumber
2291: \\ \times
2292: [(\langle S_{x}(t) \rangle \omega_{J} \sin \omega_{J} t -
2293: \langle S_{y}(t) \rangle \omega_{L} \cos \omega_{J} t) \hat{e}_{x} \nonumber
2294: \\ + (\langle S_{y}(t) \rangle \omega_{J} \sin \omega_{J} t
2295: + \langle S_{x}(t) \rangle \omega_{L}
2296: \cos \omega_{J} t) \hat{e}_{y} \nonumber
2297: \\ + \langle S_{z}(0) \rangle \omega_{J} \sin \omega_{J}
2298: t \hat{e}_{z}] \nonumber
2299: \\ = C_{2} [\omega_{J} \langle \vec{S}(t) \rangle
2300: \sin \phi(t) + \omega_{L} (\hat{e}_{z} \times
2301: \langle \vec{S}(t) \rangle) \cos \phi(t)],
2302: \label{Iqu-cl*}
2303: \end{eqnarray}
2304: with the constant $C_{2} \equiv - |T_{1}|^{2}
2305: \sum_{k,p} \frac{|\Delta|^{2}}{E_{k} E_{p} (E_{k}+E_{p})^{2}}$
2306: and $\phi(t) = \omega_{J} t$ the superconducting phase
2307: difference across the junction. The last line
2308: of Eq.(\ref{Iqu-cl*}) has a very physically suggestive
2309: meaning regarding spin contractions and
2310: an effective longitudinal magnetic
2311: field- items which we will expand on in
2312: Section(\ref{consequences}).
2313: Our expressions (Eqs.(\ref{Icl-cl*},\ref{Iqu-cl*}))
2314: above are exact to lowest order in $T_{1}$ and the ratios
2315: $(\omega_{L,J}/E_{k,p})$.
2316:
2317:
2318: Finally, the integral $\langle \vec{I}_{qu-qu} \rangle =0$
2319: identically by virtue
2320: of a vanishing
2321: $\langle \vec{S}_{qu}(t^{\prime}) \times \vec{S}_{qu}(t) \rangle_{S} =0$.
2322:
2323: \begin{references}
2324:
\bibitem{Koehler93} J. Koehler {\em et al.}, Nature {\bf 363}, 342
2325: (1993); J. Wrachtrup {\em et al.}, {\em ibid.} {\bf 363}, 244
2326: (1993); Phys. Rev. Lett. {\bf 71}, 3565 (1993).
2327:
2328: \bibitem{Engel01} H.-A. Engel and D. Loss, Phys. Rev. Lett. {\bf
2329: 86}, 4648 (2001); Phys. Rev. B {\bf 65}, 195321 (2002).
2330:
2331: \bibitem{Mana89} Y. Manassen {\em et al.}, Phys. Rev. Lett.
2332: {\bf 62}, 2531 (1989); D. Shachal and Y. Manassen, Phys. Rev. B
2333: {\bf 46}, 4795 (1992); Y. Manassen, J. Magnetic Reson. {\bf 126},
2334: 133 (1997); Y. Manassen, I. Mukhopadhyay, and N. Ramesh Rao, Phys.
2335: Rev. B {\bf 61}, 16223 (2000).
2336:
2337: \bibitem{Durkan02} C. Durkan and M. E.
2338: Welland, Appl. Phys. Lett. {\bf 80}, 458 (2002).
2339:
2340: \bibitem{Manoharan02} H. Manoharan, Nature {\bf 416}, 24 (2002);
2341: H. Manoharan, C. P. Lutz, and D. Eigler, Nature {\bf 403}, 512
2342: (2000).
2343:
2344: \bibitem{Balatsky02} A. V. Balatsky and I. Martin, Quan. Inform.
2345: Process. {\bf 1}, 53 (2002); A. V. Balatsky, Y. Manassen, and R.
2346: Salem, Phys. Rev B {\bf 66}, 195416 (2002).
2347:
2348: %\bibitem{Li98} J. Li {\em et al.}, Phys. Rev. Lett. {\bf 80}, 2893
2349: %(1998).
2350:
2351: \bibitem{Mozy02b} D. Mozyrsky and I. Martin, Phys. Rev. Lett. {\bf
2352: 89}, 018301 (2002).
2353:
2354: \bibitem{nut03} J-X. Zhu, Z. Nussinov, A. Shnirman, A. V. Balatsky,
2355: Phys. Rev. Lett 92, 107001 (2004)
2356:
2357: \bibitem{bulaevskii} L. Bulaevskii, M. Hruska, A. Shnirman, D. Smith,
2358: Yu. Makhlin,
2359: Phys. Rev. Lett. {\bf 92} , 177001 (2004).
2360:
2361: \bibitem{Kulik_JETP66}
2362: I. O. Kulik, Sov. Phys. JETP {\bf 22}, 841 (1966)
2363:
2364:
2365: \bibitem{Ambegaokar_Baratoff} V. Ambegaokar and A. Baratoff,
2366: Phys. Rev. Lett., {\bf 10}, 486 (1963)
2367:
2368: \bibitem{Bulaevskii_pi_junction} L. N. Bulaevskii,
2369: V. V. Kuzii,and A. A. Sobyanin,
2370: JETP Lett. {\bf 25}, 290 (1977)
2371:
2372: \bibitem{Glazman_Matveev_SC_Kondo}
2373: L. I. Glazman and K. A. Matveev, JETP Lett., {\bf 49}, 659 (1989)
2374:
2375: \bibitem{wolfgang} W. Wernsdorfer,
2376: Adv. Chem. Phys., {\bf 118}, 99 (2001)
2377:
2378:
2379: \bibitem{avishai} Y. Avishai, A. Golub, and A. D. Zaikin,
2380: Phys. Rev. B {\bf 63}, 134515 (2001).
2381:
2382: \bibitem{mns} Yu. Makhlin, Z. Nussinov, A. Shnirman, in preparation
2383:
2384: \bibitem{WZW} E. Fradkin, {\em Field Theories of Condensed
2385: Matter Systems} (Addison-Wesley, Redwood City, 1991);
2386:
2387: \bibitem{perel} A. Perelomov, ``Generalized Coherent States and Their
2388: Applications'', Springer-Verlang, NY 1986.
2389:
2390: \bibitem{explain_CP1} For a coherent state oriented along
2391: $\hat{\Omega} = (\sin \theta \cos \phi, \sin \theta \sin \phi,
2392: \cos \theta)$ on the unit sphere, the corresponding
2393: spinor is, up to an arbitrary multiplicative global phase
2394: factor ($\exp[i \chi]$), the spinor
2395: \begin{eqnarray}
2396: z^{*}= (\cos \frac{\theta}{2} e^{-i \phi/2},
2397: \sin \frac{\theta}{2} e^{i \phi/2}).
2398: \label{ZCP1.}
2399: \end{eqnarray}
2400: While inserting complete coherent spin states at all times
2401: steps we invoke
2402: \begin{eqnarray}
2403: \langle \Omega | \Omega^{\prime} \rangle = (\frac{1 + \hat{\Omega} \cdot
2404: \hat{\Omega}^{\prime}}{2})^{1/2} e^{-i S \psi}
2405: \nonumber
2406: \\ \psi = 2 \tan^{-1}\Big(
2407: \tan[\frac{1}{2}(\phi - \phi^{\prime})] \frac{\cos[\frac{1}{2}
2408: (\theta + \theta^{\prime})]}{\cos[\frac{1}{2}(\theta - \theta^{\prime})]}
2409: \Big)
2410: + \chi - \chi^{\prime}
2411: \end{eqnarray}
2412: where $\chi$ and $\chi^{\prime}$ depend on gauge convention,
2413: with $| \Omega \rangle$ and $| \Omega' \rangle$
2414: spin coherent states at two different times.
2415: In the limit of small deviations $(\phi^{\prime} - \phi)$
2416: and $(\theta^{\prime} - \theta)$, the above argument
2417: reduces to $d \psi = [(d \phi/dt) \cos \theta] dt$
2418: which is simply the area elements swept by $\hat{\Omega}(t)$
2419: in an infinitesimal time increment $dt$. Within the path integral
2420: formulation we insert complete sets of coherent states
2421: at different times. We see from the internal product
2422: above that within the spin coherent basis the path integral attains a
2423: Berry phase which is simply the area swept by the spin
2424: its cyclic evolution. (It is cyclic due to the requirement of taking a
2425: trace within the partition function).
2426:
2427: An explicit evaluation with the spinor
2428: of Eq.(\ref{ZCP1.}) reveals that
2429: \begin{eqnarray}
2430: {\cal{L}}_{Berry} = z_{a}^{*} (i \partial_{t}) z_{a}
2431: \end{eqnarray}
2432: is none other than $[(i/2) (d \phi/dt) \cos \theta]$. Upon integration
2433: over time this leads, up to trivial additive constants,
2434: to $(i/2)$ times the area spanned by the spin.
2435:
2436: \bibitem{das} A. Das, ``Finite Temperature Field Theory'',
2437: World Scientific, 1997
2438:
2439: \bibitem{alex} A. Kamenev, cond-mat/0109316
2440:
2441:
2442:
2443: \bibitem{Zhu_Balatsky}
2444: J-X. Zhu, A. V. Balatsky, Phys. Rev. B {\bf 67}, 174505 (2003)
2445:
2446:
2447: \bibitem{AES_82}
2448: V. Ambegaokar,
2449: U. Eckern, G. Sch{\"o}n,
2450: Phys. Rev. Lett. {\bf{48}}, 1745 (1982).
2451:
2452: \bibitem{Larkin_Ovchinnikov_83}
2453: A. I. Larkin, Y. N. Ovchinnikov,
2454: Phys. Rev. B {\bf{28}}, 6281 (1983).
2455:
2456: \bibitem{ESA_84}
2457: U. Eckern, G. Sch{\"o}n, V. Ambegaokar,
2458: Phys. Rev. B {\bf{30}}, 6419 (1984).
2459:
2460:
2461:
2462: \bibitem{NS} Z. Nussinov and A. Shnirman, unpublished
2463:
2464:
2465: \bibitem{shapiro} S. Shapiro, Phys. Rev. Lett. {\bf 11}, 80 (1963)
2466:
2467: \bibitem{DM} F. J. Dyson, Phys. Rev. {\bf 102}, 1230 (1956),
2468: S. Maleev, Zh. Eksp. Teor. Fiz., {\bf 33}, 1010 (1957)
2469:
2470: %\bibitem{MM} T. Matsubara and H. Matsuda, Prog. Theor. Physics, {\bf 36},
2471: %569 (1956)
2472:
2473: \bibitem{long} Z. Nussinov, A. Shnirman, D. P. Arovas, A. V. Balatsky,
2474: J-X. Zhu, In preparation
2475:
2476: \bibitem{snz} A. Shnirman et al.,
2477: Fizika Nizkikh Temperatur {\bf 7/8}, 30 (2004)
2478: \bibitem{explain} The illustration of this
2479: change of sign upon exchanging the order
2480: of times within the
2481: path integral average is straightforward.
2482: \begin{eqnarray}
2483: \int Dz Dz^{*} \delta(|z|^{2}-1) \epsilon_{ijk}
2484: z^{*}_{a}(t_{2}) \sigma^{i}_{ab} z_{b}(t_{2})
2485: z^{*}_{c}(t_{1}) \sigma^{j}_{cd} z_{d}(t_{1}) e^{iS} = \nonumber
2486: \\ - \int Dz Dz^{*} \delta(|z|^{2}-1) \epsilon_{ijk}
2487: z^{*}_{a^{'}}(t_{1}) \sigma^{i}_{a^{'}b^{'}} z_{b^{'}}(t_{1})
2488: z^{*}_{c^{'}}(t_{2}) \sigma^{j}_{c^{'}d^{'}} z_{d^{'}}(t_{2}) e^{iS}. \nonumber
2489: \end{eqnarray}
2490: This can be seen by making an identification
2491: of summation indices, $a^{'} =c,~ b^{'} = d, ~c^{'} = a,~ d^{'} =b$,
2492: and invoking $\epsilon_{ijk} = - \epsilon_{jik}$.
2493:
2494: \end{references}
2495:
2496:
2497: \end{document}
2498: \
2499: Non-planar dynamics (nutations) of classical
2500: spinning rigid bodies are omnipresent.
2501: Amongst many other systems,
2502: nutations are even evident in the Earth's rotation. Here,
2503: we report a similar effect present in
2504: Josephson junctions harboring a single spin.