1: %\documentstyle[twocolumn,eclepsf]{jpsj}
2: %\documentstyle[preprint]{jpsj}
3: \documentstyle[twocolumn,psfig]{jpsj}
4: %\renewcommand\figureheight[1]{\vspace{24pt}\mbox{\rule{0cm}{#1}}}
5: \def\runtitle{Low-temperature dependence of the London penetration depth}
6: \def\runauthor{Takao {\sc Morinari} and Manfred {\sc Sigrist}}
7:
8: \title{The influence of chiral surface states on the London penetration
9: depth in Sr$_2$RuO$_4$}
10:
11: \author{Takao {\sc Morinari} and Manfred {\sc Sigrist}}
12: \inst{Yukawa Institute for Theoretical Physics, Kyoto University,
13: Kyoto 606-8502, Japan}
14:
15: \recdate
16: {\today}
17:
18: \abst
19: {The London penetration depth for the unconventional superconductor
20: Sr$_2$RuO$_4$ is analyzed assuming an order parameter which breaks time
21: reversal symmetry and parity simultaneously. Such a superconducting
22: state possesses chiral quasiparticle states with subgap energies at
23: the surface. We show that these subgap states can give a significant
24: contribution to the low-temperature behavior of the London penetration
25: depth yielding a $ T^2 $ power-law even though bulk quasiparticle
26: spectrum is gapped. The presence of several electron bands gives rise
27: to interband transition among the subgap surface states and influences
28: the properties of the surface impedance. Furthermore, the surface
29: states lead also to a non-linear Meissner effect. }
30:
31: \kword
32: {
33: Sr$_2$RuO$_4$, $p$-wave superconductor, time-reversal
34: breaking state, chiral surface state}
35:
36: \begin{document}
37: \sloppy
38: \maketitle
39:
40: Several years of intense experimental research have established
41: the unconventional nature of superconductivity in
42: Sr$_2$RuO$_4$ \cite{MAENO1,NATRICE}. This compound has a layered
43: perovskite structure representing a basically two-dimensional metal with
44: three almost cylindrical Fermi surfaces.
45: The symmetry of the superconducting state is very likely
46: odd in parity, which implies the spin-triplet configuration analogous
47: to superfluid $^3$He \cite{RICE,ISHIDA1}. Muon
48: spin rotation experiments provide evidence for broken
49: time reversal symmetry \cite{LUKE}, a fact that strongly suggests that
50: the gap function has the basic form,
51: \begin{equation}
52: {\bf d} ({\bf k}) = \Delta_0 \hat{{\bf z}} \frac{k_x \pm i k_y}{k_F}
53: \label{chiralpwave}
54: \end{equation}
55: which is a chiral $p$-wave state, here written in the vector
56: representation, assuming cylindrical symmetry. The Cooper pairs possess
57: an internal orbital angular momentum which is oriented along the
58: $z$-axis. A consequence of this topological property of the
59: superconducting phase is the presence of chiral surface states at the
60: surface \cite{VOLOVIK,MATSU,YASU}.
61: While the chiral $p$-wave state has a basically gapful quasiparticle
62: spectrum, the surface states correspond to subgap quasiparticle
63: excitations with a continuous spectrum down to zero energy
64: \cite{MATSU,YASU}. These quasiparticle states are Andreev bound states
65: and extend only over a coherence length towards the bulk.
66: In this letter we consider the contribution of these states to the
67: temperature dependence of the London penetration depth. The London
68: penetration depth $ \lambda_{\parallel} $ for currents within the
69: plane and the in-plane coherence length $ \xi_{\parallel} $ are very
70: similar giving a Ginzburg-Landau parameter $ \kappa =
71: \lambda_{\parallel}/ \xi_{\parallel} \approx 2.6 $. Therefore, the
72: presence of the surface
73: states can lead to a visible reduction of the screening effect and
74: could even
75: dominate the low-temperature behavior $ \lambda_{\parallel} $, in
76: particular, if the bulk quasiparticle spectrum is gapped.
77: We show here that power-law temperature dependence can result from
78: the surface states, which is usually taken as an evidence for
79: nodes in the bulk quasiparticle gap.
80:
81: The discussion of the London penetration depth requires a careful
82: analysis of the current-current response to an external field which can
83: be written in general as
84: \begin{equation}
85: j_\mu ({\bf r},t) = - \frac{c}{4 \pi}\sum_{\nu} \int dt' \int d^3r'
86: K_{\mu \nu}({\bf r},t; {\bf r}',t') A_\nu ({\bf r}',t')
87: \end{equation}
88: where only the transverse component of $ A_\nu ({\bf r}',t') $ enters.
89: The kernel $ K_{\mu \nu}({\bf r},t; {\bf r}',t') $ is obtained from the
90: current-current correlation function. In our
91: case this response consists of two contributions: the bulk part due to
92: the continuum of quasiparticle states above the gap and the part due
93: to the surface states. We consider from now on the specific case of a
94: surface with normal vector along the $x$-axis and an external field
95: parallel to the $z$-axis. Consequently we have to deal with the
96: transverse vector potential and screening current along the
97: $y$-axis. The relevant terms are then,
98: \begin{equation}
99: j_y ({\bf r},t) = - \frac{1}{c}\int d^4r' \left[
100: \Pi_{yy} (r; r')
101: - \frac{c^2 \delta^{(4)}(r- r')}{4 \pi \lambda_0^2} \right]
102: A_y (r')
103: \label{eq_j}
104: \end{equation}
105: where the integral runs of the four coordinates $ r'= ({\bf r}', t') $
106: and $\lambda_0 $ corresponds to the bare ``London penetration depth''
107: of the bulk regime which can be considered as basically
108: temperature-independent for very low temperatures. For this bulk part
109: we take the local approximation, while for the first part connected
110: with the surface states nonlocality is important, as we will see
111: below.
112:
113: The surface states can be easily described within the Bogolyubov-de
114: Gennes formalism if we neglect self-consistency of the gap $ \Delta_0 $
115: which we choose to be constant everywhere inside the superconductor.
116: For the sake of simplicity we assume also that the surface provides
117: specular reflection of quasiparticles and the gap has no anisotropy on
118: the Fermi surface. The electron band in our model has cylindrical
119: symmetry and is represented by the parabolic form,
120: $ \varepsilon_{\bf k} = \{(k_x^2 + k_y^2) - k_{\rm F}^2 \} /
121: 2m $ neglecting any dispersion along the $z$-axis.
122: Using these simplifications the wave function of the subgap states
123: localized at the surface is given by
124: \begin{equation}
125: \left( \begin{array}{c}
126: u_{\bf k} \left( {\bf r} \right) \\
127: v_{\bf k} \left( {\bf r} \right)
128: \end{array} \right) \simeq \sqrt{\frac{2}{\xi_0L_y d}} ~ {\rm
129: e}^{ik_y y - \frac{x}{\xi_0}}~ \sin (k_x x)
130: \left( \begin{array}{c}
131: 1 \\
132: -i \end{array} \right).
133: \label{uv}
134: \end{equation}
135: Since only states very close to Fermi surface are important for the
136: low-temperature properties the wave vector
137: can be represented essentially as $ (k_x , k_y) =k_F (\cos \theta,\sin
138: \theta)$ for $ k_F \xi_0 \gg 1 $. Further,
139: $ L_y $ is the extension of the system along the $ y $-direction with
140: periodic boundary conditions and $
141: d$ is the interlayer spacing (the wave function is renormalized per
142: layer). The energy of the surface states
143: is given by $E_{k_y} = \eta \Delta_0 {k_y}/k_F$ with $
144: \eta = \pm 1 $ denoting the sign of the chirality of $ {\bf d} ({\bf
145: k}) = \hat{{\bf z}} (k_x \pm i k_y)/k_F $ \cite{HONER}.
146: Defining the current operators as $ j_{\mu} = (\hbar e/2mi)
147: (\hat{\Psi}^{\dag} \partial_{\mu} \hat{\Psi} - h.c.) $, where
148: \begin{equation}
149: \hat{\Psi} ({\bf r})=\left( \begin{array}{c}
150: \psi_{\uparrow} ({\bf r}) \\
151: \psi_{\downarrow}^{\dag} ({\bf r})
152: \end{array} \right) = \sum_{\bf k} \left[
153: \begin{array}{cc} u_{\bf k} ({\bf r}) & -v_{\bf k}^* ({\bf r}) \\
154: v_{\bf k} ({\bf r}) & u_{\bf k}^* ({\bf r}) \end{array} \right]
155: \left( \begin{array}{c} \gamma_{{\bf k} \uparrow} \\ \gamma_{{\bf
156: k}\downarrow}^{\dag} \end{array} \right),
157: \label{N_f}
158: \end{equation}
159: is the Nambu field operator with $\gamma^{({\dag})}_{{\bf k}\sigma}$
160: the Bogolyubov quasi-particle operator, we can express the
161: current-current correlation function $\Pi_{yy} (r;r')$ as
162: \begin{eqnarray}
163: \Pi_{yy} (r;r')
164: &=& -\frac{\hbar^2 e^2}{4m^2 } \lim_{r_1,r_2 \rightarrow r,~r_1',r_2'
165: \rightarrow r'} \nonumber \\
166: & & \hspace*{-15mm}\times (\partial_{y_1} - \partial_{y_2})
167: (\partial_{y_2'} - \partial_{y_1'})
168: {\rm Tr} \left[ {\bf G}(r_1; r_1') {\bf G}(r_2'; r_2) \right]
169: \label{j_j}
170: \end{eqnarray}
171: where ${\bf G} (r;r')$ is the Nambu-Gor'kov Green's
172: function in real space and $ \partial_y $ denotes the derivative with
173: respect to the spatial $ y$-coordinate.
174: The Green's function can be expressed as
175: \begin{equation}
176: {\bf G} ({\bf r},{\bf r}^{\prime};i\omega_n)
177: \simeq \frac{\phi (x) \phi (x^{\prime})}{L_y d} \sum_{0<k<k_F}
178: \sum_{s=\pm} \frac{{\hat{\sigma}}_0 - s {\hat{\sigma}}_2}{i\omega_n -
179: s E_k} {\rm e}^{isk(y-y^{\prime})}
180: \label{green_fn}
181: \end{equation}
182: with $\phi (x) = \sqrt{2/\xi_0}~ \exp^{-x/\xi_0} \sin k_F x$,
183: ${\hat{\sigma}}_0$ the unit matrix and
184: ${\hat{\sigma}}_2$ the second Pauli matrix,
185: and $\omega_n$ being the fermionic Matsubara frequency.
186: Using Eqs. (\ref{j_j}) and (\ref{green_fn}), we calculate
187: the current-current correlation function. The translational invariance
188: along the $ y $-direction allows us to transform the $ y$-coordinate
189: into momentum space,
190: \begin{eqnarray}
191: \lefteqn{\Pi_{yy} (x,x^{\prime};q,i\Omega_n)} \nonumber \\
192: & & \quad \simeq -\frac{32\hbar^2 e^2}{m^2L_y d } g(x) g(x^{\prime})
193: \sum_{0< k <k_F} k^2
194: \frac{f(E_{k+q})-f(E_k)}{i\Omega_n - E_{k+q}+E_k} \nonumber \\
195: & & \quad \simeq - \frac{8\pi \hbar^2 k_F^3}{3d m^2 \Delta_0}
196: \left( \frac{k_B T}{\Delta_0} \right)^2
197: \frac{g(x) g(x^{\prime})}{1-i\Omega_n k_F /q \Delta_0},
198: \label{pi}
199: \end{eqnarray}
200: in the limit $ T \ll T_c $,
201: where $g(x)\simeq \exp(-\frac{2x}{\xi_0}) \sin^2 (k_F x)/ \xi_0 $ is
202: the square of the amplitude of the surface state wave function and $ i
203: \Omega_n $ is the bosonic Matsubara frequency.
204: In deriving Eq. (\ref{pi}), we restrict ourselves to the leading
205: contribution for $q \ll k_F$ and $k_B T \ll \Delta_0$.
206: The nonlocal nature of response enters via the product form $ g(x)
207: g(x') $ which accounts for the fact that each of the
208: quasiparticle state is localized at the surface. The field at the
209: particular point $ (x',y', z') $ couples to the surface state in the
210: same layer with a weight $ g(x') $ and yields consequently a response
211: at any other point $ (x,y,z') $ with weight $ g(x) $. This is a
212: feature of the effectively one-dimensional character of the surface
213: states within each layer. A local approach in this place would
214: underestimate the role of the surface states in the low-temperature
215: response.
216:
217: Combining Eqs. (\ref{eq_j}) and (\ref{pi}), where we further use the
218: analytic continuation $i\Omega_n \rightarrow \hbar \omega +
219: i\delta$, with the Maxwell equation
220: $\nabla^2 A_y ({\bf r},\omega) = -\frac{4\pi}{c} j_y ({\bf r},\omega)$
221: we obtain an integro-differential equation for $A_y ({\bf
222: r},t)$. The boundary condition is given by
223: $\partial_x A_y ({\bf r},\omega)|_{x=0} =
224: B_z(q,\omega)$, where $ B_z(q,\omega)$ is the external magnetic field
225: at the surface parallel to the $ z$-axis. We solve this equation using
226: an approximation $g(x) \simeq \exp(-2x/\xi_0)/2 \xi_0 $ in the
227: integrand (we ignore the fast oscillations), which is certainly valid
228: for $k_F \xi_0 \gg 1$. This allows us to calculate the surface
229: impedance, $Z(q,\omega) = 4\pi c E_y(x=0,q,\omega)/B(q,\omega) = -
230: 4\pi i \omega A_y (x=0,q,\omega)/B(q,\omega)$. We then obtain the
231: penetration depth using the relation $4\pi \omega \lambda (q,\omega) =
232: {\rm Im} Z(q, \omega)$. Taking a static limit $\hbar \omega
233: k_F/q\Delta_0 \rightarrow 0$ and $q \rightarrow 0$, we find in the
234: regime of $k_B T \ll \Delta_0$, $\Delta \lambda (T) = \lambda(T) -
235: \lambda(0) $ has $T^2$-behavior:
236: \begin{equation}
237: \Delta \lambda (T)/\lambda_0 \simeq \frac{4\pi^2}{3}
238: \frac{\kappa}{(2\kappa+1)^2} \left( k_B T/\Delta_0 \right)^2.
239: \label{lambda}
240: \end{equation}
241: Setting $\kappa \simeq 2.6$, which is a typical value of
242: Sr$_2$RuO$_4$, we obtain $\Delta \lambda(T)/\lambda_0 \simeq 0.14
243: \times \left( T/T_c \right)^2$, if we assume the weak-coupling
244: relation $ \Delta_0 = 1.76 k_B T_c $. We ignored the temperature
245: dependence of $ \lambda_0 $ as it is exponential in the
246: low-temperature regime in our model.
247:
248: \begin{figure}
249: \begin{center}
250: \psfig{file=figure1.eps,height=7.0cm}
251: \end{center}
252: \caption{Schematic spectrum of the chiral surface states in the
253: multi-band case of Sr$_2$RuO$_4$. The Brillouin zone contains
254: three Fermi surfaces. The electron-like $ \beta $
255: and $ \gamma $-Fermi surfaces yield a chiral surface state spectrum
256: centered around $ k_y=0 $, while the spectrum of the surface state due
257: to the hole-like $ \alpha $-Fermi surface located at the Brillouin zone
258: boundary with opposite sign of chirality. The gap magnitudes are in
259: general different. We assigned, however, the same magnitude for $
260: \alpha $ and $ \beta $-Fermi surface for simplicity.}
261: \end{figure}
262:
263: It is important to notice that this contribution is independent of
264: the sign of the chirality and the charge of the carriers, i.e., whether
265: the superconducting state is $ {\bf d} \propto \hat{{\bf z}} (k_x + i
266: k_y) $ or $ \hat{{\bf z}}(k_x
267: - i k_y) $) and the Fermi surface is electron- or hole-like. Therefore,
268: the
269: formation of domains of the two superconducting states would not
270: lead to a significant change of the result.
271: Furthermore, for the
272: case of several superconducting bands the contributions of the surface
273: states of each band add up to enlarges the prefactor of the $ T^2
274: $-law. The coherence length
275: as the extension of the surface states towards the interior is
276: different for each band, since the Fermi velocities and the gap
277: magnitudes are different. The coherence length, experimentally
278: determined via the measurement of $ H_{c2}$, giving $ \kappa \simeq
279: 2.6 $ is the shortest among all. Therefore, the enhancement of the
280: surface state contribution can be sizable in the multiband case. We
281: consider here the case of three bands, as schematically shown in
282: Fig.1, where each band yields its own surface state described by a
283: Green's function ${\bf G}^{(j)} (r;r')$, (the superscript $j$ labels
284: the $j^{th}$band). We assume that the reflection of quasiparticles on
285: the surface does not lead to transitions among the different
286: bands. Each band is characterized by a Fermi vector $ k^{(j)}_{F} $,
287: the effective band mass $ m_j $ and the superconducting gap $\Delta_j
288: $. We now use Eq.(\ref{j_j}) to calculate the contribution of each
289: band to the current-current correlation function and then analyze the
290: resulting equation for the transverse vector potential as in the
291: single band case. This leads to the low-temperature behavior of the
292: London penetration depth,
293: \begin{equation}
294: \frac{\Delta \lambda (T)}{\lambda_0} \simeq \frac{4\pi^2}{3}
295: \sum^3_{j=1} \frac{\kappa^{(j)}}{(2 \kappa^{(j)}+1)^2}
296: \left(\frac{\lambda_0 \Delta_0}{\lambda^{(j)} \Delta_j}\right)^2
297: \left( \frac{k_B T}{\Delta_0} \right)^2
298: \label{lambda2}
299: \end{equation}
300: where
301: \begin{equation}
302: \frac{1}{\lambda_0^2} = \sum_j \frac{1}{\lambda^{(j)2}} = \sum_j
303: \frac{2 e^2 k^{(j)2}_F}{m_j c^2 d}
304: \label{eq_eta}
305: \end{equation}
306: and $ \kappa^{(j)} = \lambda_0 / \xi^{(j)}_0 $ with $ \xi^{(j)}_0 =
307: \hbar^2 k^{(j)}_F / m_j \Delta_j $. We choose $ \Delta_0 $ to
308: reproduce again the weak coupling relation, $ \Delta_0 = 1.76 k_B T_c
309: $. The contribution of all three bands can easily give a prefactor to
310: the $ (T/T_c)^2 $-law of order one, consistent with recent
311: measurements for fields along the $ z $-axis.
312:
313: In the discussion of the multi-band situation we
314: neglected the interband effects. Oscillatory fields, for example
315: appearing in microwave experiments, can yield interband
316: transitions. The matrix elements for the transition depends on various
317: details of the orbital and band structure. We do not go into these
318: complex details here, but assume that the interband transition can be
319: described by an ordinary current operator, $ j^{(i,j)}_\mu = (\hbar
320: e/2m'i) (\hat{\Psi}^{(i)\dag} \partial_\mu \hat{\Psi}^{(j)} - h.c.)$
321: where $ m' $ is a phenomenological parameter accounting for the matrix
322: element. $\hat{\Psi}^{(i)}$ is the Nambu field operator of the $i$-th
323: band as in Eq. (\ref{N_f}). In the small-momentum $(q)$ and
324: small-frequency $ (\omega) $ limit only surface states are important
325: which share the same zero-energy momentum. As shown in Fig.1 this is
326: the case for the $ \beta $- and $ \gamma $-band. The $ \alpha $-band
327: is unimportant because for interband transitions a large momentum
328: transfer ($ q \sim \pi $) is necessary. Analogous to Eq.(\ref{pi}) we
329: can derive the correlation function,
330: \begin{equation} \begin{array}{l} \displaystyle
331: \Pi^{(\beta, \gamma)}_{yy}(x_1,x_2,q; i \Omega_n)
332: \simeq \frac{16 \hbar^2 e^2}{m'^2d L_y } \tilde{g}(x_1) \tilde{g}(x_2)
333: \sum_{k} k^2 \\ \\ \displaystyle
334: \times \left [ \frac{f(E^{\beta}_{k+q}) - f(E^{\gamma}_k)}{i\Omega_n -
335: E^{\beta}_{k+q} + E^{\gamma}_k}+ \frac{f(E^{\gamma}_{k+q}) -
336: f(E^{\beta}_k)}{i\Omega_n - E^{\gamma}_{k+q} + E^{\beta}_k} \right]
337: \end{array}
338: \label{ib-pi}
339: \end{equation}
340: where $ \tilde{g}(x) \approx \exp(-2 x / \tilde{\xi})/2
341: \sqrt{\xi^{(\beta)} \xi^{(\gamma)}} $ with $ \tilde{\xi}^{-1} =
342: \xi^{(\beta)-1} + \xi^{(\gamma) -1} $. The surface state spectra in the
343: two bands is approximated by $ E^{(i)}_k = v_i k $, with $ v_i =
344: \Delta_i/k^{(i)}_F $. We now take the analytic continuation $ i
345: \Omega_n \to \hbar \omega + i \delta $ and set $ q=0 $ in
346: Eq.(\ref{ib-pi}). For the limit of very small $ \hbar \omega $ ($ \ll
347: k_B T $) the surface impedance gets the following contribution from
348: the interband transitions,
349: \begin{equation} \begin{array}{l} \displaystyle
350: Im Z_{ib}(\omega) = \frac{4 \pi^3 \tilde{\xi} \xi^{(\gamma)}}{3
351: \lambda^{(\gamma)2}} \frac{\tilde{\kappa}^3}{(2 \tilde{\kappa}+1)^2}
352: \left(\frac{\Delta_0 m_{\gamma}}{\Delta_\gamma m'}\right)^2
353: \frac{\omega v_{\gamma}}{\tilde{v}} \\ \\
354: \displaystyle
355: \qquad \times \left[ \left(\frac{k_B T}{\Delta_0} \right)^2
356: \left(\frac{v_{\gamma}^2}{v_{\beta}^2}-1 \right)
357: + \frac{2}{\pi} \left(\frac{\hbar \omega}{\Delta_0} \right)^2 {\rm ln}
358: \frac{v_{\gamma}}{v_{\beta}} \right]\end{array}
359: \end{equation}
360: for the imaginary part which yields in the zero frequency-limit also
361: a $ T^2 $-contribution to the London penetration depth ($
362: \tilde{\kappa} = \lambda / \tilde{\xi} $ and $ \tilde{v} = v_{\gamma}
363: - v_{\beta} > 0 $). For the real part we obtain,
364: \begin{equation} \begin{array}{l} \displaystyle
365: Re Z_{ib}(\omega) = \frac{2 \pi^3 \tilde{\xi} \xi^{(\gamma)}}{
366: \lambda^{(\gamma)2}} \frac{\tilde{\kappa}^3}{(2 \tilde{\kappa}+1)^2}
367: \left(\frac{\Delta_0 m_{\gamma}}{\Delta_\gamma m'}\right)^2
368: \\ \\ \displaystyle
369: \qquad \times \left(\frac{v_{\gamma}}{\tilde{v}} \right)^3
370: \omega \left(\frac{\hbar
371: \omega}{\Delta_0} \right)^2 \frac{\hbar \omega}{k_B T}
372: \end{array}
373: \end{equation}
374: for given $ T $ and small $ \hbar \omega $ ($ \ll k_B T \ll \Delta_0 $).
375: The imaginary part, the inductive resistance, shows a $ \omega
376: $-linear plus $ \omega^3 $-behavior, while the real part, the surface
377: resistance follows an $ \omega^4 $-law.
378: In the opposite limit where $ k_B T \ll \hbar \omega \ll \Delta_0 $ the
379: surface impedance due to interband transitions has to vanish. The
380: reason is that for $ q =0 $ the initial and final states are either
381: both empty or occupied in the zero-temperature limit. A simple
382: analysis shows the following behavior,
383: \begin{equation} \begin{array}{l} \displaystyle
384: Re Z_{ib}(\omega) \propto \frac{\omega^4}{T} ~{\rm e}^{ -
385: \frac{v_{\gamma}+ v_{\beta}}{2 \tilde{v}} \frac{\hbar \omega}{k_B T}},
386: ~Im Z_{ib}(\omega) \propto \frac{T^4}{\omega^2}
387: \end{array} \end{equation}
388: The surface resistance and the inductive resistance vanish
389: exponentially and with a power-law, respectively, in the
390: zero-temperature limit.
391:
392: We now consider the possibility of a so-called nonlinear Meissner
393: effect.The application of a magnetic field
394: introduces a Doppler shift which changes the quasiparticle energy,
395: \begin{equation}
396: E'_{\bf k} = E^{(sg)}_{\bf k} + \frac{e v_{Fy}}{c} A_y
397: \end{equation}
398: if we again consider the case of $ {\bf n} = (1,0,0) $. We can expand
399: the current-current correlation function for small $ A_y $ and analyze
400: the contribution to the London penetration depth in the same way as
401: done above. Restricting to the single band model we obtain,
402: \begin{equation} \begin{array}{l} \displaystyle
403: \frac{\Delta \lambda(T)}{\lambda_0} \simeq \frac{4 \pi^2}{3}
404: \frac{\kappa}{(2 \kappa +1)^2} \left( \frac{k_B T}{\Delta_0} \right)^2
405: \left( 1 - \eta \frac{3 \kappa}{2} \frac{H}{H_{c2}}
406: \right)
407: \label{nonlin-pi}
408: \end{array} \end{equation}
409: which leads to a non-linear correction in
410: the external field. This effect is a
411: consequence of the angular momentum of the Cooper pairs coupling to
412: the field along the $ z$-axis. The sign of this correction
413: depends on the chirality (direction of angular momentum) and the
414: character of the Fermi surface, i.e.,
415: a different sign appears for the $ \alpha $-band than for the $ \beta
416: $- and $ \gamma $-band in Fig.1. Therefore,
417: the presence of electron-like and hole-like Fermi surfaces as well as
418: the formation of domains of the two chiral states lead to compensations
419: which diminish the effect.
420:
421: In recent experiments by Bonalde et al., the temperature
422: dependence of the
423: London penetration depth was determined
424: using a self-inductive technique\cite{BONALD}.
425: It was found that the low-temperature behavior is
426: indeed governed by a $ T^2 $-behavior. This is not compatible
427: with a simple interpretation in terms of line nodes in the gap as
428: originally proposed based on the $ T $-power laws in specific heat and
429: NQR \cite{NISHI}, since this would lead to a linear $ T $-dependence
430: \cite{MINEEV}. The more sophisticated approach based on a nonlocal
431: response theory by Kostin and Leggett (KL) (for the reason that $
432: \kappa $ is small), however, would yield a $ T^2 $-behavior
433: \cite{KOSTIN}. On the other hand, in this letter we propose an
434: alternative mechanism for a $ T^2$-behavior based on the contributions
435: of the surface states. In both theories it is expected that this
436: power-law behavior is absent for $ \lambda_{\perp} $, for screening
437: currents flowing along the $ z$-axis. The $z$-axis current is not
438: proportional to the surface state energy as required to obtain the $
439: T^2 $-behavior in our theory. Furthermore, $ \kappa_{\perp} $ is about
440: 20 times larger than the in-plane $ \kappa $ so that contribution of
441: the surface states as well as the nonlocal effect by KL are rather
442: small. However, the measurements for fields in the plane, probing the
443: $ z $-axis current show a similar $ T^2 $-behavior. This is in
444: apparent conflict with both interpretations. Since in this case,
445: however, not only $ \lambda_{\perp} $ but also the contribution from
446: in-plane currents from the surfaces normal to the $z$-axis are involved,
447: the final answer will be given only when these geometrical aspects
448: have been thoroughly investigated \cite{VANHARL}.
449:
450: In our model the gap size is isotropic on the Fermi surface.
451: Anisotropy in turn modifies the surface state spectrum to $ E_{k_y} =
452: v k_y + v' k_y^3 + ... $ without destroying the particle-hole symmetry
453: ($ v v' < 0 $ in general). This yields an additional $ T^4
454: $-contribution which may not be so small. Together with other
455: correction in this order this leads to
456: \begin{equation}
457: \frac{\Delta \lambda(T)}{\lambda_0} = a (T/T_c)^2 + b (T/T_c)^4 + ...
458: \end{equation}
459: with $ a \sim b >0 $.
460: In an intermediate range the second term generates a
461: T-dependence which over some temperature range appears to be close to
462: a $ T^3$-behavior and only at rather low temperatures the $ T^2 $-law
463: would dominate. For one sample Bonalde et al. could indeed fit their
464: data reasonably well with a $ T^3 $-curve \cite{VANHARL}. While this
465: sample happened to be dirty it is not clear from the experiment what
466: is the intrinsic origin for the apparently different
467: power-law. Therefore, the additional $ T^4 $-contribution, which due
468: to the surface orientation and disorder is larger than usual, may be
469: one possible explanation.
470:
471: A further aspect noteworthy here is that our mechanism is active at the
472: surface only, while the KL scheme also applies also in the bulk of the
473: superconductor. Therefore, the London penetration
474: depth governing the magnetic interaction between vortices should have
475: different temperature dependence in the two scenarios. Measurements of
476: in the mixed phase by $ \mu $SR suggest that London penetration
477: depth saturates faster than $ T^2 $ at low temperature \cite{MSR}.
478: In contrast to the surface-sensitive experiment mentioned above
479: \cite{BONALD}, $\mu$SR is indeed a bulk probe.
480: Unfortunately, it has less accuracy in determining the
481: temperature dependence of the London penetration depth so that we
482: cannot draw a strong conclusion to date. Nevertheless,
483: the present experimental result is consistent with the interpretation
484: based on surface states.
485:
486: We would like to thank A. Furusaki, M. Matsumoto, T.M. Rice,
487: C. Honerkamp, J. Goryo,
488: Y. Maeno, D. Van Harlingen and I. Bonalde for many
489: stimulating discussions.
490: This work was supported by a Grant-in-Aid of the Japanese Ministry of
491: Education, Science, Culture and Sports.
492:
493: \begin{thebibliography}{99}
494: \bibitem{MAENO1}
495: Y. Maeno et al.: Nature {\bf 372} (1994) 532.
496: Physica C {\bf 282-287} (1997) 206.
497: \bibitem{NATRICE}
498: T.M. Rice: Nature {\bf 396} (1998) 627.
499:
500: \bibitem{RICE} T. M. Rice and M. Sigrist:
501: J. Phys. Condens. Matter. {\bf 7} (1995) L643;
502: G. Baskaran, Physica {\bf B 223-224} (1996) 490.
503:
504: \bibitem{ISHIDA1} K. Ishida et al.: Nature {\bf 396} (1998) 658.
505:
506: \bibitem{LUKE} G.M. Luke et al.: Nature {\bf 394} (1998) 558.
507:
508: \bibitem{VOLOVIK} G.E. Volovik: JETP Lett. {\bf 66} (1997) 523 and
509: references cited therein.
510:
511: \bibitem{YASU} M. Sigrist, A. Furusaki, C. Honerkamp, M. Matsumoto,
512: K.K. Ng and Y. Okuno: J. Phys. Soc. Jpn. Suppl. (2000).
513:
514:
515: \bibitem{MATSU} M. Matsumoto and M. Sigrist:
516: J. Phys. Soc. Jpn. {\bf 68} (1999) 994; ibid. {\bf 68} (1999) 3120.
517:
518: \bibitem{HONER} C. Honerkamp and M. Sigrist: J. Low. Temp. Phys. {\bf
519: 111} (1998) 895.
520:
521: \bibitem{BONALD} I. Bonalde, B.D. Yanoff, M.B. Salamon, D.J. Van
522: Harlingen and E.M.E. Chia: preprint.
523:
524: \bibitem{NISHI} S. Nishizaki et al.: J. Phys. Soc. Jpn. {\bf 67}
525: (1998) 560.
526:
527: \bibitem{MINEEV} V.P. Mineev and K.V. Samokhin, {\it Introduction to
528: Unconventional Superconductivity}, Gordon and Breach Science
529: Publisher, Amsterdam 1999.
530:
531: \bibitem{KOSTIN} I. Kosztin and A.J. Leggett: Phys. Rev. Lett. {\bf
532: 79} (1997) 135.
533:
534: \bibitem{VANHARL} D.J. Van Harlingen: private communications.
535:
536: \bibitem{MSR} G.M. Luke: private communications.
537:
538: \end{thebibliography}
539:
540: \end{document}
541: