1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: %%%%%%%%%%%%% MANUSCRIPT %%%%%%%%%%%%%%%%%%%%%
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: %format latex
5: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
6: \tolerance = 10000
7: %\documentstyle[aps,preprint,epsf]{revtex}
8: %\documentstyle[preprint,aps,psfig]{revtex}
9: \documentclass[prb,twocolumn]{revtex4}
10: %\documentclass[prb]{revtex4}
11: %\documentclass[preprint,prb,aps]{revtex4}
12: %\documentclass[prb]{revtex4}
13: %\documentstyle[aps,prl,multicol,epsf]{revtex}
14: %\documentclass[preprint,aps,prl]{revtex4}
15: %\documentclass[prl,twocolumn]{revtex4}
16: %\renewcommand{\baselinestretch}{1.7}
17: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
18:
19: \usepackage{graphicx}
20:
21: \begin{document}
22:
23: \title{
24: Magnetoelectric effects in heavy-fermion superconductors without
25: inversion symmetry
26: }
27:
28: \author{Satoshi Fujimoto}
29: \affiliation{
30: Department of Physics,
31: Kyoto University, Kyoto 606-8502, Japan
32: }
33:
34: \date{\today}
35:
36: \begin{abstract}
37: We investigate effects of strong electron correlation on
38: magnetoelectric transport phenomena
39: in noncentrosymmetric superconductors
40: with particular emphasis on its application to
41: the recently discovered heavy-fermion superconductor CePt$_3$Si.
42: Taking into account electron correlation effects in a formally exact way,
43: we obtain the expression of the magnetoelectric coefficient
44: for the Zeeman-field-induced paramagnetic supercurrent, of which
45: the existence was predicted more than a decade ago.
46: It is found that in contrast to the usual Meissner current, which is
47: much reduced by the mass renormalization factor
48: in the heavy-fermion state, the paramagnetic supercurrent is not affected by
49: the Fermi liquid effect.
50: This result implies that the experimental observation of
51: the magnetoelectric effect is more feasible in heavy-fermion systems
52: than that in conventional metals with moderate effective mass.
53: \end{abstract}
54:
55: \pacs{PACS number: 74.20.-z, 74.25.Fy, 74.70.Tx}
56:
57: \maketitle
58: %%%%%
59: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
60: \section{Introduction}
61:
62: It has been discussed for decades that metallic
63: systems with noncentrosymmetric crystal structure
64: may exhibit non-trivial magnetoelectric
65: effects.~\cite{lev,ede0,ede1,ede2,yip}
66: The existence of an asymmetric potential gradient $\vec{\nabla}V$ due to
67: the noncentrosymmetric structure gives rise to
68: the spin-orbit interaction
69: $(\vec{p}\times \vec{\nabla}V)\cdot\vec{\sigma}$,
70: which breaks the individual inversion and spin rotation symmetry.
71: As a result, the charge or energy current operator
72: may couple to the spin density operator.
73: In this context, current flows induced by an applied magnetic field, and
74: current-flow-driven magnetization
75: have been investigated extensively
76: both in normal metals~\cite{lev,ede0} and
77: superconductors.~\cite{ede1,ede2,yip}
78: In particular, Edelstein predicted the remarkable magnetoelectric effect
79: in superconducting states; i.e.
80: in noncentrosymmetric superconductors,
81: the Zeeman field induces a supercurrent, and vice versa,
82: the supercurrent flow induces a magnetization.~\cite{ede1,ede2}
83: Later, the former effect is elegantly re-formulated by Yip in terms of
84: the ``van Vleck'' contribution which stems
85: from the inversion-symmetry-breaking spin-orbit interaction.~\cite{yip}
86: Since a static magnetic field can not induce dissipative current flows,
87: the Zeeman-energy-induced current should vanish in the normal state.
88: However, in the superconducting state,
89: the existence of the paramagnetic supercurrent
90: is not forbidden in the absence of the inversion symmetry.
91: The recent discovery of superconducting materials without inversion
92: symmetry such as CePt$_3$Si, UIr, and Cd$_2$Re$_2$O$_7$
93: stimulates the renewed interest
94: in this issue.~\cite{bau,uir,cd,take,hari,met,yogi}
95: Under an applied magnetic field, the Meissner diamagnetic supercurrent
96: in addition to the Zeeman-field-induced paramagnetic supercurrent
97: should exist. Thus, it is important for the experimental observation
98: of this effect to discriminate between these two supercurrents.
99:
100: In the present paper, we would like to investigate
101: the Fermi liquid corrections to this Edelstein's magnetoelectric effect,
102: which may be important for the application to heavy-fermion superconductors
103: such as CePt$_3$Si and UIr.
104: We obtain the formula for the magnetoelectric effect coefficient
105: taking into account Fermi liquid corrections exactly.
106: The most important finding is that
107: the Zeeman-energy-induced paramagnetic supercurrent is not at all
108: affected by electron correlation effects provided that
109: ferromagnetic spin fluctuation is not developed, in contrast with
110: the diamagnetic Meissner current of which the magnitude is
111: much reduced by the mass renormalization effect.
112: This result implies that the experimental detection of
113: the paramagnetic supercurrent in heavy-fermion superconductors may be
114: more feasible than that in weakly correlated metals.
115:
116: The organization of this paper is as follows.
117: In Sec.II, we present the basic formulation of the Fermi liquid theory
118: for a model system without inversion symmetry.
119: We would like to make a brief comment on the superconducting state realized in
120: CePt$_3$Si in Sec.III.
121: In Sec.IV, the exact formula of the magnetoelectric coefficient
122: is obtained. In Sec.V, the implication for
123: the experimental observation of this effect is discussed.
124: Summary and discussion are given in the last section.
125:
126: \section{Model and analysis based on the Fermi liquid theory}
127:
128: As a simplest model which realizes the broken inversion symmetry,
129: we consider an interacting electron system with the Rashba spin-orbit
130: interaction.~\cite{ras}
131: The Hamiltonian is given by,
132: \begin{eqnarray}
133: \mathcal{H}&=&\sum_{p,\sigma} \varepsilon_p c^{\dagger}_{\sigma p}c_{\sigma p}
134: +\alpha\sum_{p,\sigma\sigma'}
135: (\vec{p}\times \vec{n})\cdot\vec{\sigma}_{\sigma\sigma'}
136: c^{\dagger}_{\sigma p}c_{\sigma' p} \nonumber \\
137: %+\sum_{p,\sigma\sigma'} [\Delta_{\sigma\sigma'}
138: %c_{\sigma' -p}c_{\sigma p}+h.c.]
139: &&+U\sum_i n_{\uparrow i}n_{\downarrow i}, \label{ham}
140: \end{eqnarray}
141: where $c^{\dagger}_{\sigma p}$ ($c_{\sigma p}$)
142: is the creation (annihilation) operator for an electron with spin $\sigma$
143: and momentum $p$. The number density operator at
144: the site $i$, $n_{\sigma i}=c^{\dagger}_{\sigma i}c_{\sigma i}$.
145: The second term of (\ref{ham}) is the Rashba spin-orbit interaction
146: which incorporates the broken inversion symmetry.
147: Here,
148: the unit vector parallel to the asymmetric potential gradient is given by
149: $\vec{n}=(0,0,1)$.
150: This system is considered to be a model of CePt$_3$Si, with which we are
151: mainly concerned in this paper.
152: The $f$-electron of CePt$_3$Si is in the $\Gamma_7$ Kramers
153: doublet state.~\cite{met}
154: Expanding the $\Gamma_7$ doublet in terms of the $s_z=1/2$ and $-1/2$ basis,
155: we found that the Rashba spin-orbit interaction term expressed
156: in term of the $\Gamma_7$ basis has the matrix structure
157: given above up to a constant factor which can be absorbed into
158: the re-definition of the coupling constant $\alpha$.
159: Thus, in the case of CePt$_3$Si, the spin index $\sigma$ in (\ref{ham})
160: represents the $\Gamma_7$ Kramers doublet.
161:
162: In the following, we do not specify the pairing mechanism
163: of superconductivity, but assume that the superconducting state is realized
164: by an effective pairing interaction with an angular momentum $l\geq 1$
165: which may stem from the on-site Coulomb interaction in (\ref{ham}) or may have
166: any other origin not included in the Hamiltonian (\ref{ham}).
167: Then, we can analyze electron correlation effects on this superconducting state
168: in a formally exact way by using the superconducting
169: Fermi liquid theory.~\cite{leg,mig,fuji}
170:
171: In the conventional Nambu representation,\cite{sch}
172: the inverse of the single-particle
173: Green's function is defined as,
174: \begin{eqnarray}
175: \hat{\mathcal{G}}^{-1}(p)=
176: \left(
177: \begin{array}{cc}
178: i\varepsilon_n-\hat{H}(p) & -\hat{\Delta}(p) \\
179: -\hat{\Delta}^{\dagger}(p) & i\varepsilon_n+\hat{H}^{t}(-p)
180: \end{array}
181: \right), \label{gin}
182: \end{eqnarray}
183: where $p=(\vec{p},i\varepsilon)$, and,
184: \begin{eqnarray}
185: \hat{H}(p)=\hat{H}_0(p)+\hat{\Sigma}(p), \label{h1}
186: \end{eqnarray}
187: \begin{eqnarray}
188: \hat{H}_0=\varepsilon_p-\mu+\alpha(\vec{p}\times\vec{n})\cdot\vec{\sigma}
189: -\mu_{\rm B}\sigma_xH_x, \label{h0}
190: \end{eqnarray}
191: with $\mu$ chemical potential.
192: Here, to discuss the magnetoelectric effect,
193: we take into account the Zeeman magnetic field
194: in the $x$ direction, $H_x$.
195: For simplicity, we assume that the $g$-value is equal to 2.
196: The self-energy matrix $\hat{\Sigma}$ consists of
197: both diagonal and off-diagonal components,
198: \begin{eqnarray}
199: \hat{\Sigma}=
200: \left(
201: \begin{array}{cc}
202: \Sigma_{\uparrow\uparrow}(p) & \Sigma_{\uparrow\downarrow}(p) \\
203: \Sigma_{\downarrow\uparrow}(p) & \Sigma_{\downarrow\downarrow}(p)
204: \end{array}
205: \right). \label{self}
206: \end{eqnarray}
207: We can easily see from the symmetry argument that
208: under the applied in-plane magnetic field,
209: $\Sigma_{\uparrow\uparrow}(p)=\Sigma_{\downarrow\downarrow}(p)
210: \equiv\Sigma(p)$, and
211: $\Sigma_{\downarrow\uparrow}(\vec{p},i\varepsilon)=
212: \Sigma_{\uparrow\downarrow}^{*}(\vec{p},-i\varepsilon)$.
213: The superconducting gap function is
214: $\{\hat{\Delta}(p)\}_{\alpha\beta}=\Delta_{\alpha\beta}(p)$
215: ($\alpha,\beta=\uparrow,\downarrow$).
216:
217: $i\varepsilon_n-\hat{H}(p)$ and $i\varepsilon_n+\hat{H}^t(-p)$ in
218: $\hat{\mathcal{G}}^{-1}(p)$
219: are diagonalized by the unitary transformation
220: $\hat{\mathcal{A}}(p)\hat{\mathcal{G}}^{-1}(p)\hat{\mathcal{A}}^{\dagger}(p)$
221: with,
222: \begin{eqnarray}
223: \hat{\mathcal{A}}(p)=
224: \left(
225: \begin{array}{cc}
226: \hat{U}(\vec{p}) & 0 \\
227: 0 & \hat{U}^{t\dagger}(-\vec{p})
228: \end{array}
229: \right),
230: \end{eqnarray}
231: \begin{eqnarray}
232: \hat{U}(\vec{p})=
233: \frac{1}{\sqrt{2}}\left(
234: \begin{array}{cc}
235: 1 & i\hat{t}_{-} \\
236: i\hat{t}_{+} & 1
237: \end{array}
238: \right)
239: \end{eqnarray}
240: where $\hat{t}_{\pm}=\hat{t}_x\pm i\hat{t}_y$, and
241: $\hat{t}_x$, $\hat{t}_y$ are, respectively, the $x$ and $y$ components
242: of the unit vector $\vec{\hat{t}}_p\equiv\vec{t}_p/|\vec{t}_p|$ with
243: \begin{eqnarray}
244: \vec{t}_p(i\varepsilon)=
245: (p_x+\frac{1}{\alpha}{\rm Im}\Sigma_{\uparrow\downarrow},
246: p_y+\frac{1}{\alpha}{\rm Re}\Sigma_{\uparrow\downarrow}
247: -\frac{\mu_{\rm B}H_x}{\alpha},0). \label{tp}
248: \end{eqnarray}
249: As seen from eqs.(\ref{h1}),(\ref{h0}), and (\ref{self}),
250: the main effect of the off-diagonal self-energy
251: $\Sigma_{\uparrow\downarrow}$ is to renormalize
252: the Rashba interaction term, replacing the momentum $\vec{p}$
253: in the Rashba term with the vector $\vec{t}_p$.
254: Since the on-site Coulomb interaction does not change
255: the symmetry of the system, the off-diagonal self-energy
256: should satisfy the following condition in the absence of magnetic fields;
257: \begin{eqnarray}
258: {\rm Re}\Sigma_{\uparrow\downarrow}(p_x,-p_y)=
259: -{\rm Re}\Sigma_{\uparrow\downarrow}(p_x,p_y),
260: \end{eqnarray}
261: \begin{eqnarray}
262: {\rm Im}\Sigma_{\uparrow\downarrow}(-p_x,p_y)=
263: -{\rm Im}\Sigma_{\uparrow\downarrow}(p_x,p_y),
264: \end{eqnarray}
265: and ${\rm Re}\Sigma_{\uparrow\downarrow}$
266: (${\rm Im}\Sigma_{\uparrow\downarrow}$) is
267: an even function of $p_x$ ($p_y$).
268: %From eq.(\ref{tp}),
269: %we see that the energy dependence of $\Sigma_{\uparrow\downarrow}$
270: %is negligible as long as the relevant energy scale is
271: %lower than the Rashba spin-orbit splitting.
272: %Thus, we ignore it in the following.
273:
274: In the normal state,
275: the single-particle excitation energy $\varepsilon_{p\tau}^{*}$
276: for the quasi-particle with the helicity $\tau=\pm 1$
277: is given by the solution of the equation $z-\hat{H}(\vec{p},z)=0$,
278: which is, in the diagonalized representation,
279: \begin{equation}
280: \varepsilon_{p\tau}^{*}+\mu-\varepsilon_{p}
281: -\tau\alpha|\vec{t}_p(\varepsilon_{p\tau}^{*})|
282: -{\rm Re}\Sigma(\vec{p},\varepsilon_{p\tau}^{*})=0.
283: \end{equation}
284: %with
285: %\begin{equation}
286: %\varepsilon_{p\tau}=\varepsilon_p +\tau\alpha|\vec{t}_p|.
287: %\end{equation}
288:
289: The gap functions $\hat{\Delta}(p)$ and $\hat{\Delta}^{\dagger}(p)$ in
290: $\hat{\mathcal{G}}^{-1}(p)$ are also diagonalized by the
291: unitary transformation
292: $\hat{\mathcal{A}}(p)\hat{\mathcal{G}}^{-1}(p)\hat{\mathcal{A}}^{\dagger}(p)$
293: provided that the gap function has the following structure,
294: \begin{equation}
295: \hat{\Delta}(p)=\Delta_s(p)i\sigma_y
296: +\Delta_t(p)(\vec{\hat{t}}_p\times \vec{n})\cdot\vec{\sigma}i\sigma_y.
297: \label{gap}
298: \end{equation}
299: Here $\Delta_s(p)$ and $\Delta_t(p)$ are even functions of momentum $\vec{p}$.
300: This means that the spin singlet and triplet component is mixed
301: in the diagonalized basis labeled by $\tau=\pm 1$, and
302: the $\vec{d}$ vector of the triplet component is
303: $\vec{\hat{t}}_p\times \vec{n}$.~\cite{ede1,gor}
304: In the case that $\hat{\Delta}(p)$ is not expressed as eq.(\ref{gap}),
305: $\hat{H}(p)$ and $\hat{\Delta}(p)$ can not be diagonalized simultaneously,
306: and the non-zero off-diagonal components of $\hat{\Delta}(p)$ which
307: correspond to the Cooper pairing between the different Fermi surfaces
308: induce pair-breaking effects, resulting in the decrease of the transition
309: temperature $T_c$.
310: Thus, the highest transition temperature is achieved by
311: the gap function given by eq.(\ref{gap}).~\cite{fri}
312: The realization of the gap function (\ref{gap})
313: in the case with no inversion center is
314: also elucidated by the group theoretical argument.~\cite{ser}
315:
316: Taking the inverse of eq.(\ref{gin}), we have,
317: \begin{eqnarray}
318: \hat{\mathcal{G}}(p)=
319: \left(
320: \begin{array}{cc}
321: \hat{G}(p) & \hat{F}(p) \\
322: \hat{F}^{\dagger}(p) & -\hat{G}^{t}(-p)
323: \end{array}
324: \right), \label{g1}
325: \end{eqnarray}
326: where
327: \begin{eqnarray}
328: \hat{G}(p)=\sum_{\tau=\pm 1}\frac{1+\tau(\vec{\hat{t}}_p\times\vec{n})
329: \cdot\vec{\sigma}}{2}G_{\tau}(p),
330: \end{eqnarray}
331: \begin{eqnarray}
332: \hat{F}(p)=\sum_{\tau=\pm 1}\frac{1+\tau(\vec{\hat{t}}_p\times\vec{n})
333: \cdot\vec{\sigma}}{2}i\sigma_yF_{\tau}(p),
334: \end{eqnarray}
335: and,
336: \begin{eqnarray}
337: G_{\tau}(p)=\frac{z_{p\tau}(i\varepsilon+\varepsilon_{p\tau}^{*})}
338: {(i\varepsilon+i\gamma {\rm sgn}\varepsilon)^2-E^{2}_{p\tau}},
339: \end{eqnarray}
340: \begin{eqnarray}
341: F_{\tau}(p)=\frac{z_{p\tau}\Delta_{\tau}(p)}
342: {(i\varepsilon+i\gamma {\rm sgn}\varepsilon)^2-E^{2}_{p\tau}}.
343: \end{eqnarray}
344: Here the mass renormalization factor is,
345: \begin{eqnarray}
346: z_{p\tau}&=&\biggl[1-\frac{\partial {\rm Re}\Sigma(p)}
347: {\partial (i\varepsilon)} \nonumber \\
348: &+&\tau\left(
349: \hat{t}_x\frac{\partial {\rm Im}\Sigma_{\uparrow\downarrow}(p)}
350: {\partial (i\varepsilon)}
351: +\hat{t}_y\frac{\partial {\rm Re}\Sigma_{\uparrow\downarrow}(p)}
352: {\partial (i\varepsilon)}
353: \right)
354: \biggr]^{-1}\biggr|_{i\varepsilon=E_{p\tau}},
355: \end{eqnarray}
356: and $\gamma$ is the quasi-particle damping.
357: The single-particle excitation energy is
358: $E_{p\tau}=\sqrt{\varepsilon_{p\tau}^{* 2}+\Delta_{\tau}^2(p)}$
359: with $\Delta_{p\tau}=z_{p\tau}(\Delta_s(p)+\tau\Delta_t(p))$.
360:
361: The superconducting gap function $\Delta_{\alpha\beta}$ and
362: the transition temperature are determined by
363: the self-consistent gap equation,
364: \begin{eqnarray}
365: \Delta_{\alpha\beta}=T\sum_{n,p}{\rm Tr}[\hat{\Gamma}^{\alpha\beta}(p,p')
366: \hat{F}(p')],
367: \end{eqnarray}
368: where we have introduced the four-point vertex function matrix
369: $\{\hat{\Gamma}^{\alpha\beta}(p,p')\}_{\gamma\delta}$ which is diagrammatically
370: expressed as shown in FIG.1.
371: We expand the four-point vertex in the particle-particle channel
372: $\{\hat{\Gamma}^{\alpha\beta}(p,p')\}_{\gamma\delta}$
373: in terms of the basis of the irreducible representations of
374: the point group,
375: and consider a component
376: $\{\hat{\Gamma}^{\alpha\beta}_a(p,p')\}_{\gamma\delta}$ which corresponds to
377: the pairing state giving highest $T_c$.
378: This pairing interaction
379: consists of the spin singlet and triplet channel:
380: \begin{eqnarray}
381: &&\{\hat{\Gamma}^{\alpha\beta}_a(p,p')\}_{\gamma\delta}=\Gamma_s(p,p')
382: i(\sigma_{y})_{\alpha\beta}i(\sigma_{y})_{\gamma\delta} \nonumber \\
383: &&+\Gamma_t(p,p')(\vec{\hat{t}}_p\times\vec{n})\cdot (\vec{\sigma}
384: i\sigma_{y})_{\alpha\beta}(\vec{\hat{t}}_{p'}
385: \times\vec{n})\cdot (\vec{\sigma}i\sigma_{y})_{\gamma\delta}.
386: \end{eqnarray}
387: %If the point group symmetry of $\Gamma_{s}(p,p')$
388: %in the momentum space belongs to an irreducible
389: %representation different from that of $\Gamma_{t}(p,p')$,
390: %the self-consistent gap equations for $\Delta_{s}(p)$ and $\Delta_{t}(p)$
391: %are decoupled, and the mixing of the singlet and triplet states
392: %does not occur.
393: %On the other hand, when
394: The symmetries of $\Gamma_{s}(p,p')$ and $\Gamma_{t}(p,p')$
395: in the momentum space
396: are characterized by the same irreducible
397: representation, and as a result,
398: $\Delta_{s}(p)$ and $\Delta_{t}(p)$ in eq.(\ref{gap}) have the same
399: symmetry in the momentum space.
400: The realized superconducting state is the mixture of the spin singlet
401: and triplet states.~\cite{ede1,gor}
402: In this case, the possible pairing state is $s+p$ or $d+f$ or $g+h$,
403: and so forth.
404: %In the case of CePt$_{3}$Si, it is expected that
405: %the strong on-site coulomb interaction between $f$-electrons suppresses
406: %the pairing in the $s$-channel.
407: %Thus, the $d+f$ wave state is a plausible candidate in this system.
408:
409: \section{A comment on the superconducting state realized in
410: ${\rm CePt_3Si}$ : possibility of an unusual coherence effect}
411:
412: Here, we would like to make a brief remark about
413: the pairing state realized in CePt$_{3}$Si.
414: The NMR measurement carried out by Yogi et al. shows
415: the existence of the coherence peak of $1/T_1T$ just below $T_c$,
416: indicating the full-gap state without nodes.~\cite{yogi}
417: On the other hand, the recent experiment on
418: the thermal transport done by Izawa et al. supports
419: the existence of line nodes of the superconducting gap.~\cite{matsu}
420: A possible resolution of this discrepancy is that
421: the line nodes of the superconducting gap are
422: generated accidentally at the magnetic Zone boundary
423: which emerges as a result of the antiferromagnetic phase
424: transition at $T_N=2.2$ K, and crosses the Fermi surface.
425: For such accidental nodes without the sign change of
426: the superconducting gap function,
427: the coherence factor of $1/T_1T$ does not vanish,
428: resulting in the enhancement of the coherence peak just below $T_c$.
429: In this case, a plausible candidate for the pairing state is
430: the $s+p$ wave state.
431: An important point which we would like to stress here is that
432: even when the superconducting state is dominated by
433: the $p$ wave pairing; i.e. $\Delta_s(p) \ll \Delta_t(p)$,
434: the coherence factor which enters into $1/T_1T$ does not vanish.
435: This contrasts with the case of
436: the usual $p$ wave state realized in centrosymmetric superconductors,
437: where the coherence factor of $1/T_1T$ disappears.
438: This is understood as follows.
439: For simplicity, we ignore electron correlation effects.
440: Then, in noncentrosymmetric superconductors,
441: the nuclear relaxation rate is given by,~\cite{sch}
442: \begin{eqnarray}
443: &&\frac{1}{T_1T}\propto \nonumber \\
444: &&\lim_{\omega\rightarrow 0}
445: \frac{1}{\omega}{\rm Im}\Bigl[T\sum_{\varepsilon_m}\sum_{p,p'}
446: \{{\rm Tr}[\frac{\sigma^{+}}{2}\hat{G}(p,\varepsilon_m+\omega_n)
447: \frac{\sigma^{-}}{2}\hat{G}(p',\varepsilon_m)] \nonumber \\
448: &&-{\rm Tr}[\frac{\sigma^{+}}{2}
449: \hat{F}(p,\varepsilon_m+\omega_n)
450: \frac{\sigma^{-}}{2}\hat{F}(p',\varepsilon_m)]\}
451: |_{i\omega_n\rightarrow \omega+i\delta}\Bigr] \nonumber \\
452: &&=\int \frac{d\varepsilon}{2\pi}\frac{1}{2T\cosh^2\frac{\varepsilon}{2T}}
453: \{[N_n(\varepsilon)]^2+[N_a(\varepsilon)^2]\}, \label{t1}
454: \end{eqnarray}
455: with $N_n(\varepsilon)$ and $N_a(\varepsilon)$ defined by the retarded Green's
456: functions as,
457: \begin{eqnarray}
458: N_n(\varepsilon)=-\sum_p\sum_{\tau=\pm}{\rm Im}
459: G^R_{\tau}(p,\varepsilon),
460: \end{eqnarray}
461: \begin{eqnarray}
462: N_a(\varepsilon)=-\sum_p\sum_{\tau=\pm}{\rm Im}
463: F^R_{\tau}(p,\varepsilon).
464: \end{eqnarray}
465: The expression of $1/T_1T$ (\ref{t1})
466: does not rely on the phase factor $\vec{\hat{t}}_p\times \vec{n}$
467: of the triplet component of the gap function (\ref{gap}).
468: The second term of the right-hand side of (\ref{t1}) gives
469: the non-zero contribution from the coherence factor,
470: as in the case of conventional $s$ wave superconductors.
471: This property enhances the coherence peak of $1/T_1T$
472: prominently.
473: It may be important to take into account this unusual coherence effect
474: for clarification of the origin of the notable coherence peak of $1/T_1T$
475: in CePt$_3$Si.~\cite{yogi}
476: It is also intriguing to explore the unusual
477: coherence effect on other response functions,
478: %in noncentrosymmetric $p$ wave superconductors,
479: such as the ultrasonic attenuation.
480: We would like to address this issue elsewhere.
481: %However, we can not completely exclude the possibility that
482: %the line nodes reflect intrinsically the pairing symmetry, and
483:
484:
485: \begin{figure}[h]
486: \includegraphics*[width=4cm]{invfig1}
487: \caption{Four point vertex in the particle-particle channel
488: }
489: \end{figure}
490:
491:
492: \section{Many-body effects on the magnetoelectric transport
493: in the superconducting state}
494:
495: In this section, we consider electron correlation effects on
496: the magnetoelectric transport
497: in the superconducting state first found by Edelstein and later
498: discussed in detail by Yip; i.e. the emergence of the paramagnetic
499: supercurrent induced by the Zeeman magnetic field in the direction
500: normal to the $\vec{n}$ vector.~\cite{ede1,yip}
501:
502: We consider the charge current flowing in the $y$ direction which
503: is defined as,
504: \begin{eqnarray}
505: J_y=T\sum_{\varepsilon_n,p}\frac{1}{2}{\rm Tr}[\hat{V}_{0y}\mathcal{G}(p)]
506: \end{eqnarray}
507: with
508: \begin{equation}
509: \hat{V}_{0y}=
510: \left(
511: \begin{array}{cc}
512: \hat{v}_{0y}(\vec{p}) & 0 \\
513: 0 & -\hat{v}_{0y}^{t}(-\vec{p})
514: \end{array}
515: \right)
516: \end{equation}
517: and
518: $\hat{v}_{0y}=\partial_{p_y}\varepsilon_p+\alpha(\vec{n}\times \vec{\sigma})$.
519:
520: The transport coefficient which characterizes the Edelstein's
521: magnetoelectric effect is given by,
522: \begin{eqnarray}
523: &&\mathcal{K}_{yx}\equiv \left.\frac{\partial J_y}{\partial H_x}\right|_{H_x=0}
524: \nonumber \\
525: &&=-T\sum_{n,p}\frac{1}{2}
526: {\rm Tr}[\hat{V}_{0y}\mathcal{G}(p)
527: \frac{\partial \mathcal{G}^{-1}(p)}
528: {\partial H_x}\mathcal{G}(p)]|_{H_x=0} \label{k1}
529: \end{eqnarray}
530:
531: Following ref.3,
532: to simplify the expression of $\mathcal{K}_{yx}$
533: we use the Ward's identity for the current vertex,
534: \begin{eqnarray}
535: \frac{\partial\hat{G}(p)}{\partial p_y}&=&
536: \hat{G}(p)\hat{\tilde{V}}_y(\vec{p})\hat{G}(p)
537: +\hat{F}(p)\hat{\tilde{V}}^t_y(-\vec{p})
538: \hat{F}^{\dagger}(\vec{p},-i\varepsilon) \nonumber \\
539: &&+\hat{R}(p), \label{wa}
540: \end{eqnarray}
541: where $\hat{\tilde{V}}_y(\vec{p})=
542: \hat{v}_{0y}+\partial\hat{\Sigma}(p)/\partial p_y$, and
543: \begin{eqnarray}
544: \hat{R}(p)=
545: \left(
546: \begin{array}{cc}
547: 0 & r(p) \\
548: r^{*}(\vec{p},-i\varepsilon) & 0
549: \end{array}
550: \right),
551: \end{eqnarray}
552: \begin{eqnarray}
553: r(p)=\left[\frac{G_{+}-G_{-}}{2\alpha|\vec{t}_p|}
554: -G_{+}G_{-}+F_{+}F_{-}\right]
555: \hat{t}_{+}\Lambda_{+-}^{cy}(p),
556: \end{eqnarray}
557: \begin{eqnarray}
558: \Lambda_{+-}^{cy}(p)=\hat{t}_x
559: (\alpha+\frac{\partial {\rm Re}\Sigma_{\uparrow\downarrow}}{\partial p_y})
560: -\hat{t}_y
561: \frac{\partial {\rm Im}\Sigma_{\uparrow\downarrow}}{\partial p_y}.
562: \end{eqnarray}
563: Then, from (\ref{g1}), (\ref{k1}), and (\ref{wa}), we obtain,
564: \begin{eqnarray}
565: \mathcal{K}_{yx}&=&
566: -2T\sum_{n,p}{\rm Tr}\left[\hat{F}(p)\hat{v}_{0y}^t(-\vec{p})
567: \hat{F}^{\dagger}(p)\left(\mu_{\rm B}\sigma_x-
568: \frac{\partial \hat{\Sigma}}{\partial H_x}\right)\right] \nonumber \\
569: &-&2\mu_{\rm B}\alpha T\sum_{n,p}F_{+}F_{-}\hat{t}_x
570: \Lambda_{+-}^{sx}(p) \nonumber \\
571: &+&T\sum_{n,p}{\rm Tr}\left[\frac{\partial \hat{\Sigma}}{\partial p_y}
572: \frac{\partial \hat{G}}{\partial H_x}\right]
573: -T\sum_{n,p}{\rm Tr}\left[\frac{\partial \hat{\Sigma}}{\partial H_x}
574: \frac{\partial \hat{G}}{\partial p_y}\right], \label{sc}
575: \end{eqnarray}
576: where the three-point vertex function is,
577: \begin{eqnarray}
578: \Lambda_{+-}^{sx}(p)=\hat{t}_x(1-\frac{1}{\mu_{\rm B}}\frac{\partial {\rm Re}
579: \Sigma_{\uparrow\downarrow}}{\partial H_x})+\frac{\hat{t}_y}{\mu_{\rm B}}
580: \frac{\partial {\rm Im}
581: \Sigma_{\uparrow\downarrow}}{\partial H_x}.
582: \end{eqnarray}
583: The last two terms of (\ref{sc}), which emerge as a result of
584: Fermi liquid corrections, can be rewritten by using the Luttinger-Ward's
585: identity generalized to the superconducting state.
586: The Luttinger-Ward's identity in the normal state reads,~\cite{lut,lut2}
587: \begin{eqnarray}
588: T\sum_{n,p}{\rm Tr}[\hat{\Sigma}\frac{\partial \hat{G}}{\partial p_{\mu}}]=0.
589: \end{eqnarray}
590: This relation is obtained by differentiating all closed linked
591: diagrams with respect to $p_{\mu}$.~\cite{lut,lut2}
592: In the superconducting state, a similar analysis leads,
593: \begin{eqnarray}
594: T\sum_{n,p}{\rm Tr}[\hat{\Sigma}\frac{\partial \hat{G}}{\partial p_{\mu}}]
595: &+&T\sum_{n,p}{\rm Tr}[\hat{\Delta}^{\dagger}\frac{\partial \hat{F}}
596: {\partial p_{\mu}}] \nonumber \\
597: &+&T\sum_{n,p}{\rm Tr}[\hat{\Delta}\frac{\partial \hat{F}^{\dagger}}
598: {\partial p_{\mu}}]=0. \label{lw}
599: \end{eqnarray}
600: Differentiating eq.(\ref{lw}) with respect to $H_x$, and
601: integrating over $p_y$ by parts, we have,
602: \begin{eqnarray}
603: &&T\sum_{n,p}{\rm Tr}\left[\frac{\partial \hat{\Sigma}}{\partial p_y}
604: \frac{\partial \hat{G}}{\partial H_x}\right]
605: -T\sum_{n,p}{\rm Tr}\left[\frac{\partial \hat{\Sigma}}{\partial H_x}
606: \frac{\partial \hat{G}}{\partial p_y}\right] \nonumber \\
607: &&=
608: -T\sum_{n,p}{\rm Tr}\left[\frac{\partial \hat{\Delta}^{\dagger}}{\partial p_y}
609: \frac{\partial \hat{F}}{\partial H_x}\right]
610: +T\sum_{n,p}{\rm Tr}\left[\frac{\partial \hat{\Delta}^{\dagger}}{\partial H_x}
611: \frac{\partial \hat{F}}{\partial p_y}\right] \nonumber \\
612: &&-T\sum_{n,p}{\rm Tr}\left[\frac{\partial \hat{\Delta}}{\partial p_y}
613: \frac{\partial \hat{F}^{\dagger}}{\partial H_x}\right]
614: +T\sum_{n,p}{\rm Tr}\left[\frac{\partial \hat{\Delta}}{\partial H_x}
615: \frac{\partial \hat{F}^{\dagger}}{\partial p_y}\right]. \label{lw2}
616: \end{eqnarray}
617: We see from eqs.(\ref{sc}) and (\ref{lw2}) that
618: $\mathcal{K}_{yx}$ vanishes exactly in the normal state,
619: as is consistent with
620: the thermodynamic argument that a static magnetic field can not induce
621: non-equilibrium current flows.~\cite{ede1,yip}
622: The right-hand side of eq.(\ref{lw2}) consists of the terms which have
623: the form $\sum_{n,p} G_{\tau}F_{\tau'}$.
624: The ratio of the contributions from these terms to those from
625: other terms of eq.(\ref{sc}) is of order $\Delta/E_F$, and thus
626: we can neglect the last two terms of (\ref{sc}) approximately.
627: Then, the magnetoelectric coefficient is expressed as,
628: \begin{eqnarray}
629: &&\frac{\mathcal{K}_{yx}}{e\mu_{\rm B}}= \nonumber \\
630: &&\sum_p\sum_{\tau=\pm 1}\tau v_{0y\tau}
631: \frac{z_{p\tau}^2\Delta_{p\tau}^2}{E_{p\tau}^{2}}
632: \left[\frac{{\rm ch}^{-2}\frac{E_{p\tau}}{2T}}{2T}-
633: \frac{{\rm th}\frac{E_{p\tau}}{2T}}{E_{p\tau}}\right]
634: \Lambda^{sx}_{\tau}(p)
635: \nonumber \\
636: &&+2\alpha\sum_p\frac{z_{p+}z_{p-}\Delta_{p+}\Delta_{p-}}
637: {E_{p+}^2-E_{p-}^2}\left[\frac{{\rm th}\frac{E_{p+}}{2T}}{E_{p+}}
638: -\frac{{\rm th}\frac{E_{p-}}{2T}}{E_{p-}}\right]
639: \hat{t}_x\Lambda^{sx}_{+-}(p) \label{kyx}
640: \end{eqnarray}
641: where
642: \begin{eqnarray}
643: \Lambda^{sx}_{\tau}(p)=\hat{t}_y(1-\frac{1}{\mu_{\rm B}}\frac{\partial {\rm Re}
644: \Sigma_{\uparrow\downarrow}}{\partial H_x})-\frac{\hat{t}_x}{\mu_{\rm B}}
645: \frac{\partial {\rm Im}
646: \Sigma_{\uparrow\downarrow}}{\partial H_x}
647: -\frac{\tau}{\mu_{\rm B}}\frac{\partial {\rm Re}\Sigma}{\partial H_x}.
648: \end{eqnarray}
649: It is noted that the vertex corrections due to electron correlation,
650: $\Lambda^{sx}(p)$,
651: which appear in the above expression (\ref{kyx})
652: is nothing but the vertex corrections to
653: the uniform spin susceptibility,
654: \begin{eqnarray}
655: &&\chi_{xx}= \nonumber \\
656: &&\mu_{\rm B}^2\sum_p\sum_{\tau=\pm 1}
657: \frac{z_{p\tau}}{2E_{p\tau}^{2}}
658: \left[\frac{\varepsilon_{p\tau}^{*2}{\rm ch}^{-2}\frac{E_{p\tau}}{2T}}{2T}-
659: \frac{\Delta_{p\tau}^2{\rm th}\frac{E_{p\tau}}{2T}}{E_{p\tau}}\right]
660: \hat{t}_y\Lambda^{sx}_{\tau}(p) \nonumber \\
661: &&+ \mu_{\rm B}^2\sum_p\left[
662: \frac{\varepsilon_{p+}^{*}{\rm th}\frac{E_{p+}}{2T}}{2E_{p+}}
663: -\frac{\varepsilon_{p-}^{*}{\rm th}\frac{E_{p-}}{2T}}{2E_{p-}}
664: \right]
665: \frac{\hat{t}_x}{\alpha|\vec{t}_p|}\Lambda^{sx}_{+-}(p). \label{chi}
666: \end{eqnarray}
667: Eq.(\ref{chi}) is easily obtained by differentiating
668: the $x$ component of the total magnetization
669: $S^x=\mu_{\rm B}T\sum_{n,p}{\rm Tr}[\sigma_{x}\hat{G}(p)]$ with respect
670: to $H_x$. Note that the first term of the right-hand side of (\ref{chi})
671: is the Pauli paramagnetic contribution and the second one is
672: the ``van Vleck'' term which arises from excitations between
673: spin-orbit split two bands.
674: Generally, in heavy-fermion systems,
675: the magnitude of the uniform spin susceptibility is
676: enhanced by the vertex corrections $\Lambda^{sx}(p)$.
677: In typical heavy-fermion systems including
678: CePt$_3$Si, the Wilson ratio
679: $R_W=T\chi/C/(T\chi_0/C_0)\sim 2$, which implies that
680: the vertex corrections $\Lambda^{sx}$ is approximately of order
681: the mass enhancement factor $1/z_{p\tau}$.~\cite{take}
682: Therefore, in eq.(\ref{kyx}) effects of the vertex corrections
683: and the mass renormalization factors $z_{p\tau}$
684: cancel with each other.
685: This cancellation holds as long as there is no strong ferromagnetic
686: spin fluctuation which increases notably
687: the magnitudes of the vertex corrections $\Lambda^{sx}$.
688: Another important feature of eq.(\ref{kyx}) is the absence of
689: the backflow term of the charge current, which usually exists in
690: the non-equilibrium current flow. (See the discussion on the usual
691: Meissner current in the next section.)
692: This is related to the fact that the current induced by a static magnetic field
693: is a dissipationless equilibrium flow.
694: As a result, {\it the Fermi liquid corrections do not exist
695: in this magnetoelectric coefficient for heavy fermion superconductors,
696: provided that there is no ferromagnetic fluctuation.}
697: This is one of the main result of this paper.
698: In terms of the Kubo formula,
699: the absence of electron correlation effects for $\mathcal{K}_{yx}$
700: is understood as follows.
701: $\mathcal{K}_{yx}$ is given by the correlation function
702: of the current and spin density operators.
703: The spin density vertex is renormalized by electron correlation in
704: the opposite way to the current vertex, resulting in the cancellation
705: of the mass renormalization factors.
706: The important implication of this result is that
707: the Zeeman-field-induced paramagnetic supercurrent is not suppressed by
708: electron correlation effects in contrast with
709: the usual diamagnetic supercurrent of which the magnitude is much
710: reduced by the large mass enhancement in heavy fermion systems.
711: This property may make the experimental observation of the magnetoelectric
712: effect easier, as discussed in the next section.
713:
714: \section{Implications for experimental observations}
715:
716: On the basis of the formula (\ref{kyx}), we would like to
717: discuss how the Zeeman-field-induced paramagnetic supercurrent
718: is experimentally observed.
719:
720: When the magnetic field is applied in the $x$ direction,
721: the London equation is modified to,
722: \begin{eqnarray}
723: \vec{J}_s=-\frac{c}{4\pi \lambda^2}\vec{A}+\mathcal{K}_{yx}(\vec{n}\times
724: \vec{H}_x), \label{lon}
725: \end{eqnarray}
726: where $\vec{A}$ is the vector potential.
727: Since the applied magnetic field always induces both the diamagnetic
728: and the paramagnetic supercurrent in the system
729: with the Rashba spin-orbit interaction,
730: it is important for the experimental observation of this effect
731: to discriminate between these two supercurrents.
732: If one measures currents simply attaching leads to the sample and applying
733: an in-plane magnetic field, the Zeeman-field-induced
734: paramagnetic supercurrent can not flow in the sample because it generates
735: the Joule heat in the normal metal leads.~\cite{com} In this case,
736: the magnetoelectric effect is canceled with
737: the non-zero phase gradient of the superconducting order parameter
738: in the equilibrium state.
739:
740: Here, to highlight the observation of the paramagnetic supercurrent,
741: we consider an experimental setup composed of two superconducting samples
742: joined together as depicted in FIG.2.
743: The $\vec{n}$ vectors of the sample A and B are, respectively,
744: given by $\vec{n}_{\rm A}=(0,0,1)$,
745: and $\vec{n}_{\rm B}=-\vec{n}_{\rm A}$.
746: The joined surface at the junction is normal to the $\vec{n}$ vectors.
747: The applied magnetic field in the $x$ direction
748: $\vec{H}=(H,0,0)$ gives rise to
749: the paramagnetic supercurrent in the sample A (B) in the direction
750: $\vec{n}_{\rm A}\times \vec{H}$ ($-\vec{n}_{\rm A}\times \vec{H}$).
751: Then, electrons accumulated in the right (left) edge of the sample A (B)
752: transfer to the right (left) edge of the sample B (A)
753: to decrease the chemical potential difference between the samples A and B,
754: resulting in supercurrent flows circulating in the system.
755: The applied magnetic field also induces the Meissner diamagnetic supercurrent.
756: As explained below, the paramagnetic contribution can be discriminated from
757: the diamagnetic current by using the Volovik effect.~\cite{vol}
758:
759: Let us first consider
760: the coefficient of the diamagnetic Meissner supercurrent
761: with the Fermi liquid corrections at zero temperature,
762: which is equal to the Drude weight in the normal state,~\cite{leg}
763: \begin{eqnarray}
764: \frac{c}{4\pi \lambda^2}=\frac{e^2}{c}
765: \sum_{p,\tau=\pm 1}v_{p\tau}^{\mu *}J_{p\tau}^{\mu *}
766: \delta(\mu-\varepsilon_{p\tau}^{*}), \label{pen}
767: \end{eqnarray}
768: where the quasi-particle velocity is
769: $v_{p\tau}^{\mu *}=\partial \varepsilon_{p\tau}^{*}/\partial p_{\mu}$, and
770: the charge current is
771: \begin{eqnarray}
772: J_{p\tau}^{\mu *}=v_{p\tau}^{\mu *}
773: +\sum_pf_{p\tau,p'\tau'}\delta (\mu-\varepsilon^{*}_{p\tau})
774: v_{p'\tau'}^{\mu *}.
775: \label{cur}
776: \end{eqnarray}
777: Here $f_{p\tau,p'\tau'}$ is the interaction between two quasiparticles.
778: The second term of (\ref{cur}) is the backflow term.
779: In heavy fermion systems, the mass renormalization factor $z_{p\tau}$
780: and the backflow term in the current $J_{p\tau}^{*}$ give rise to
781: the pronounced reduction of the Meissner coefficient.
782: For example, in CePt$_3$Si, $z_{p\tau}$ is estimated
783: as $\sim 1/100$.~\cite{bau,hari}
784: If we assume the spherical Fermi surface,
785: eq.(\ref{pen}) reduces to $(e^2/c)v^{*}_Fn_s/p_F$.
786: Here $v^{*}_F$ is the renormalized Fermi velocity, and
787: $n_s$ the superfluid density, $p_F$ the Fermi momentum.
788: It should be notified that $n_s$ is renormalized by the backflow effect
789: of $J_{p\tau}^{*}$, and is not equal to the carrier density even
790: at zero temperature.
791: In particular, for heavy fermion systems in which Umklapp scattering
792: is expected to be strong, $n_s$ is smaller than the carrier density.
793: In contrast to the diamagnetic Meissner current,
794: the Zeeman-field-induced paramagnetic supercurrent is not influenced by
795: the many-body effects described above, as discussed in the previous section.
796: This difference of electron correlation effects between the diamagnetic and
797: paramagnetic supercurrents can be utilized
798: to detect the magnetoelectric effect.
799:
800: \begin{figure}[h]
801: \includegraphics*[width=5cm]{invfig2}
802: \caption{An experimental setup for the detection of
803: the Zeeman-field-induced supercurrent.
804: The $\vec{n}$ vectors of the two superconducting samples A and B
805: (depicted by the gray arrows)
806: are aligned in the directions $(0,0,1)$ and $(0,0,-1)$, respectively.
807: The in-plane magnetic field $\vec{H}$ is applied in the $x$ direction.
808: The paramagnetic supercurrent circulates in the system as depicted by
809: the black thin arrows.
810: }
811: \end{figure}
812:
813: If the superconducting gap has a nodal structure
814: as is often realized in some heavy-fermion superconductors,
815: the existence of the paramagnetic supercurrent is indirectly
816: observed through the Volovik effect on the single particle density of
817: states under the applied in-plane magnetic field for $H_{c1}<H<H_{c2}$.
818: Here $H_{c1}$ and $H_{c2}$ are, respectively,
819: the lower and upper critical field.
820: In fact, the recent thermal transport measurements for CePt$_3$Si supports
821: the existence of the line node in the superconducting
822: state of this system.~\cite{matsu}
823: Applying the semi-classical approximation based upon
824: the Doppler shift effect,~\cite{vol} and assuming
825: a spherical Fermi surface, we calculate the local density of states
826: from the modified London equation (\ref{lon}).
827: In the calculation of $\mathcal{K}_{yx}$, we use the fact that
828: the vertex corrections $\Lambda^{sx}_{\tau (+-)}$ is appropriately
829: approxiamted as $\sim z_{p\tau}^{-1}$ in typical heavy-ferimon systems
830: as discussed in Sec.IV., and
831: expand eq.(\ref{kyx}) in terms of $\alpha p_F/E_F$ up to the lowest order.
832: The result at $T=0$ is
833: \begin{eqnarray}
834: \delta D_{\rm loc}(0)\sim \left|\sqrt{\frac{H}{H_{c2}}}
835: \frac{e^2v_{F}^{*}\Phi_0}{c\xi}
836: %\frac{p_Fc\Phi_0}{4\pi^2\xi\lambda^2n_s}
837: \pm \frac{H}{H_{c2}}\frac{e\mu_{\rm B}\alpha p_F\Phi_0n_0}{\pi\xi^2E_Fn_s}
838: \right|.
839: \label{dos}
840: \end{eqnarray}
841: Here $\Phi_0=hc/(2e)$, and $n_0$ is the density of electrons.
842: $E_F$ is the unrenormalized Fermi energy.
843: The first term of the right-hand side of (\ref{dos}) is due to
844: the usual Volovik effect, and the second term linearly proportional to
845: the applied magnetic field stems from
846: the Zeeman-energy-induced paramagnetic supercurrent.
847: Thus, the magnetic-field-dependence distinguishes between the paramagnetic
848: and diamagnetic currents.
849: The above behavior of the local density of states may be observed by
850: the measurement of the specific heat coefficient or
851: the thermal conductivity in sufficiently
852: low field regions.~\cite{hir,fra,vec,cho}
853: In the above expression of $\delta D_{loc}(0)$, it is seen that
854: the conventional diamagnetic contribution is suppressed by
855: the mass renormalization factor $z_{p\tau}$ which appears through $v^{*}_F$,
856: whereas the magnetoelectric contribution is not affected by this correlation
857: effect.
858: It is also noted that the carrier density which enters into the paramagnetic
859: term is not $n_s$ but equal to the electron density $n_0$.
860: This is due to the absence of the backflow term in the Zeeman-energy-induced
861: supercurrent.
862: As mentioned before, $n_s$ is affected by the backflow term.
863: For simplicity, we assume that $n_s\approx n_0$ for a while.
864:
865: In the case of CePt$_3$Si, according to the measurement of $H_{c2}$,
866: the coherence length $\xi\sim8.1\times 10^{-7}$ cm.~\cite{bau}
867: It is a bit difficult to estimate
868: the renormalized Fermi velocity from experimental measurements.
869: Bauer et al. obtained $v_F^{*}\sim 5.29\times 10^{5}$ cm/s
870: from the data of $dH_{c2}/dT$ and the specific heat coefficient,
871: assuming a spherical Fermi surface.~\cite{bau}
872: This value of $v_F^{*}$ is almost of the same order as
873: that obtained by combining the unrenormalized
874: Fermi velocity computed from the LDA method and
875: the mass enhancement factor $z^{-1}\sim 100$ estimated from the specific heat
876: measurement.~\cite{hari,bau}
877: %If all $f$ electrons of which the number density is one per a Ce site
878: %are itinerant, and form the Fermi surface,
879: %the renormalized Fermi velocity is estimated as $v_{F}^{*}\sim 10^{5}$ cm/s
880: %by using the fact
881: %that the mass renormalization factor is $\sim 100$ as suggested from
882: %the specific heat data, and
883: %the unrenormalized Fermi energy obtained from
884: %the band calculation is $\sim 1$ eV.
885: %Thus, it is proper to assume that
886: %the order of $v_{F}^{*}$ of CePt$_3$Si is $\sim 10^{5}$ cm/s.
887: According to the LDA band calculations,~\cite{hari,sam}
888: the spin-orbit splitting is not so small compared to
889: the Fermi energy, and may be approximated as $\alpha p_F/E_F\sim 0.1$.
890: Then, for CePt$_3$Si, we have,
891: \begin{eqnarray}
892: \delta D_{\rm loc}(0)\sim \left|
893: 1.0\times10^{-24}\sqrt{\frac{H}{H_{c2}}}\pm 0.48\times
894: 10^{-24}\frac{H}{H_{c2}}\right|. \label{dos2}
895: \end{eqnarray}
896: It is remarkable that the contribution from the paramagnetic supercurrent
897: (the second term of (\ref{dos2})) is
898: comparable to that from the Meissner supercurrent (the first term of
899: (\ref{dos2})).
900: It should be stressed
901: that the feasibility of the experimental observation
902: of the Edelstein's magnetoelectric effect is due to
903: the large mass enhancement in the heavy fermion system,
904: which suppresses strongly the Meissner supercurrent, but in contrast,
905: does not affect the Zeeman-field-induced paramagnetic supercurrent.
906: Moreover, if the superfluid density $n_s$ is reduced by
907: the backflow effect, the Meissner term of (\ref{dos}) is more suppressed
908: compared with the paramagnetic term, and thus
909: the observation of the magnetoelectric effect may become easier .
910:
911:
912:
913: \section{Summary and Discussion}
914:
915: We have investigated electron correlation effects on
916: the magnetoelectric transport phenomena
917: in superconductors without inversion symmetry.
918: It is found that, in contrast to the Meissner diamagnetic supercurrent
919: which is much reduced by the mass enhancement factor in the absence of
920: translational symmetry,
921: the Zeeman-field-induced paramagnetic supercurrent
922: is not affected by the strong electron correlation
923: provided that ferromagnetic fluctuation is not developed.
924: Because of this remarkable property,
925: the experimental detection of the magnetoelectric effect may be
926: more feasible in heavy-fermion superconductors without inversion symmetry such
927: as CePt$_3$Si, where the enormous mass enhancement suppresses
928: the magnitude of the Meissner supercurrent, than in
929: conventional metals with moderate effective electron mass.
930: We have proposed the experimental setup for the observation of
931: the magnetoelectric effect in CePt$_3$Si which utilizes the Volovik effect.
932: It has been also pointed out that in noncentrosymmetric $p$ wave
933: superconductors, the coherence effect on the nuclear relaxation rate
934: $1/T_1T$ is similar to that of conventional $s$ wave superconductors.
935:
936: Finally, we would like to comment on the implication of our results
937: for UIr, which is the recently discovered ferromagnetic superconductor
938: without inversion symmetry.~\cite{uir}
939: UIr exhibits superconductivity under high pressure
940: in the vicinity of the phase boundary between ferromagnetic
941: and non-magnetic states.
942: The resistivity of this system increases remarkably as the applied pressure
943: approaches to the critical value at which the ferromagnetism disappears,
944: indicating the existence of ferromagnetic critical fluctuation.
945: In this case, the magnetoelectric coefficient eq.(\ref{kyx})
946: may be enhanced by the three-point vertex functions $\Lambda^{sx}_{\tau(+-)}$,
947: of which the magnitudes are much increased by ferromagnetic fluctuation,
948: provided that the spin easy axis is taken as the $x$ axis.
949: Since, the crystal structure of UIr is monoclinic, and does not
950: possess any mirror planes,
951: the Rashba spin-orbit interaction with the $\vec{n}$ vector perpendicular to
952: the spin easy axis should always exists.
953: Thus, the magnetoelectric effect strongly enhanced by
954: ferromagnetic fluctuation may be observed in UIr under
955: an applied magnetic field parallel to the spin easy axis.
956:
957:
958: \acknowledgments{}
959:
960: The author would like to thank K. Yamada,
961: Y. Matsuda, and H. Ikeda for invaluable discussions.
962: This work was partly supported by a Grant-in-Aid from the Ministry
963: of Education, Science, Sports and Culture,
964: Japan.
965:
966:
967: %\begin{references}
968: \begin{thebibliography}{99}
969: \bibitem{lev} L. S. Levitov, Yu. V. Nazarov, an G. M. Eliashberg,
970: Sov. Phys. JETP {\bf 61}, 133 (1985)
971:
972: \bibitem{ede0} V. M. Edelstein, Solid State Commun.{\bf 73}, 233 (1990).
973:
974: \bibitem{ede1} V. M. Edelstein, Sov. Phys. JETP {\bf 68}, 1244 (1989).
975:
976: \bibitem{ede2} V. M. Edelstein, Phys. Rev. Lett. {\bf 75}, 2004 (1995).
977:
978: \bibitem{yip} S. K. Yip, Phys. Rev. B {\bf 65}, 144508 (2002)
979:
980: \bibitem{bau} E. Bauer, G. Hilscher, H. Michor, Ch. Paul, E. W. Scheidt,
981: A. Gribanov, Yu. Seropegin, H. No\"el, M. Sigrist, and P. Rogl,
982: Phys. Rev. Lett. {\bf 92}, 027003 (2004).
983:
984: \bibitem{uir} T. Akazawa, H. Hidaka, H. Kotegawa, T. Kobayashi,
985: T. Fujiwara, E. Yamamoto, Y. Haga, R. Settai, and Y. $\bar{\rm O}$nuki,
986: J. Phys. Soc. Jpn. {\bf 73}, 3129 (2004).
987:
988: \bibitem{cd} M. Hanawa, J. Yamaura, Y. Muraoka, F. Sakai, and
989: Z. Hiroi, J. Phys. Chem. Solids {\bf 63} 1027 (2002).
990:
991: \bibitem{take} T. Takeuchi, S. Hashimoto, T. Yasuda, H. Shishido, T. Ueda,
992: M. Yamda, Y. Obiraki, M. Shiimoto, H. Kohara, T. Yamamoto, K. Sugiyama,
993: K. Kindo, T. D. Matsuda, Y. Haga, Y. Aoki, H. Sato, R. Settai, and
994: Y. $\bar{\rm O}$nuki, J. Phys.: Condens. Matter {\bf 16}, L333 (2004).
995:
996: \bibitem{hari} S. Hashimoto, T. Yasuda, T. Kubo, H. Shishido,
997: T. Ueda, R. Settai, T. D. Matsuda, Y. Haga, H. Harima, and
998: Y. $\bar{\rm O}$nuki,
999: J. Phys.: Condens. Matter {\bf 16}, L287 (2004).
1000:
1001: \bibitem{met} N. Metoki, K. Kaneko, T. D. Matsuda, A. Galatanu,
1002: T. Takeuchi, S. Hashimoto, T. Ueda, R. Settai, Y. $\bar{\rm O}$nuki,
1003: and N. Berngoeft, J. Phys.: Condens. Matter {\bf 16}, L207 (2004).
1004:
1005: \bibitem{yogi} M. Yogi, Y. Kitaoka, S. Hashimoto, T. Yasuda, R. Settai,
1006: T. D. Matsuda, Y. Haga, Y. $\bar{\rm O}$nuki, P. Rogl, and E. Bauer,
1007: Phys. Rev. Lett. {\bf 93}, 027003 (2004).
1008:
1009: \bibitem{ras} E. I. Rashba, Sov. Phys. Solid State {\bf 2},
1010: 1109 (1960).
1011:
1012: \bibitem{leg} A. J. Leggett, Phys. Rev. {\bf 140}, 1869 (1965);
1013: Phys. Rev. {\bf 147}, 119 (1966).
1014:
1015: \bibitem{mig} A. I. Larkin and A. B. Migdal, Sov. Phys. JETP {\bf 17},
1016: 1146 (1963).
1017:
1018: \bibitem{fuji} S. Fujimoto, J. Phys. Soc. Jpn.{\bf 61}, 765 (1992).
1019:
1020: \bibitem{sch} see, e.g., J. R. Schrieffer,{\it Theory of Superconductivity}
1021: (Addison Wesley, Reading, 1983).
1022:
1023: \bibitem{gor} L. P. Gor'kov and E. Rashba, Phys. Rev. Lett.
1024: {\bf 87}, 037004 (2001).
1025:
1026: \bibitem{fri} P. A. Frigeri, D. F. Agterberg, A. Koga, and M. Sigrist,
1027: Phys. Rev. Lett. {\bf 92}, 097001 (2004).
1028:
1029: \bibitem{ser} I. A. Sergienko and S. H. Curnoe, Phys. Rev. B{\bf 70},
1030: 214510 (2004).
1031:
1032: \bibitem{lut} J. M. Luttinger and J. C. Ward, Phys. Rev. {\bf 118}, 1417
1033: (1960).
1034:
1035: \bibitem{lut2} J. M. Luttinger, Phys. Rev. {\bf 119}, 1153 (1960).
1036:
1037: \bibitem{matsu} K. Izawa, Y. Kasahara, Y. Matsuda, K. Behnia, T. Yasuda,
1038: R. Settai, and Y. $\bar{\rm O}$nuki, unpublished.
1039:
1040: \bibitem{com} If the attached leads are made of superconductors
1041: with the transition temperature higher than that of the sample and
1042: the same pairing symmetry as that of $\Delta_s(p)$ and $\Delta_t(p)$,
1043: the Zeeman-field-induced supercurrent can flow in the system.
1044: Note that the tunneling amplitude between the leads
1045: and the noncentrosymmetric superconductor does not depend on
1046: the symmetric properties of the $\vec{d}$
1047: vector of the gap function (\ref{gap}),
1048: $\vec{\hat{t}}_p\times\vec{n}$,
1049: as pointed out by I. A. Sergienko and S. H. Curnoe, in ref. 18.
1050: The observation of the magnetoelectric effect
1051: may be possible also in this case, as discussed in Sec.V.
1052:
1053: \bibitem{vol} G. E. Volovik, JETP Lett. {\bf 58}, 469 (1993).
1054:
1055: \bibitem{hir} C. K\"ubert and P. J. Hirschfeld, Phys. Rev. Lett.
1056: {\bf 80}, 4963 (1998).
1057:
1058: \bibitem{fra} M. Franz, Phys. Rev. Lett. {\bf 82}, 1760 (1999).
1059:
1060: \bibitem{vec} I. Vekhter and A. Houghton, Phys. Rev. Lett. {\bf 83},
1061: 4626 (1999).
1062:
1063: \bibitem{cho} M. Chiao, R. W. Hill, C. Lupien, B. Popic, R. Gagnon,
1064: and L. Tallefer, Phys. Rev. Lett. {\bf 82}, 2943 (1999).
1065:
1066: \bibitem{sam} K. V. Samokhin, E. S. Zijlstra, and S. K. Bose,
1067: Phys. Rev. B{\bf 69}, 094514 (2004).
1068:
1069:
1070:
1071: %%%%%%%%%%%%%%%%%%%%%%%
1072:
1073: \end{thebibliography}
1074: %\end{references}
1075:
1076:
1077: \end{document}
1078: %%%%%%%%%%%%%%%%%%%%%%%%%%%% E N D O F F I L E %%%%%%%%%
1079: