cond-mat0604329/sw.tex
1: \documentclass{elsart}
2: \usepackage{graphicx}
3: \renewcommand{\vec}[1]{\mbox{\boldmath ${#1}$}}
4: \begin{document}
5: \begin{frontmatter}
6: \title{Spin-wave instability for parallel pumping in ferromagnetic thin
7:  films under oblique field}
8: \author{Kazue Kudo\corauthref{cor}}, 
9: \corauth[cor]{Corresponding author. Tel.: +81 6 6605 2768; 
10:  fax: +81 6 6605 2768.}
11: \ead{kudo@a-phys.eng.osaka-cu.ac.jp}
12: \author{Katsuhiro Nakamura}
13: \address{Department of Applied Physics, Osaka City University,
14:               Osaka 558-8585, Japan }
15: \begin{abstract}
16: Spin-wave instability for parallel pumping is studied theoretically.
17: The spin-wave instability threshold is calculated in ferromagnetic
18:  thin films under oblique field which has an oblique angle to the film
19:  plane. 
20: The butterfly curve of the threshold usually has a cusp at a certain
21:  value of the static external field. While the static field value
22: of the cusp point
23:  varies as the oblique angle changes, the general properties of
24:  the butterfly curve show little noteworthy change. 
25:  For very thin films, however, multiple cusps of the butterfly
26:  curve can appear due to different standing spin-wave modes, which
27:  indicate a novel feature for thin films under oblique field. 
28: \end{abstract}
29: \begin{keyword}
30:  spin-wave instability \sep parallel pumping \sep thin film \sep
31:  butterfly curve 
32: \PACS 76.50.+g \sep 75.30.Ds
33: \end{keyword}
34: \end{frontmatter}
35: 
36: \section{Introduction}
37: 
38: The first experiment for the parametric excitation of spin wave by
39: parallel pumping was given by Schl\"oman {\it et al.}~\cite{Schlomann}. 
40: In a parallel pumping experiment, a microwave magnetic field is applied
41: parallel to the external static field. When the microwave field
42: amplitude $h$ exceeds a certain spin-wave instability amplitude 
43: $h_{\rm crit}$, the parametric spin-wave excitation occurs.
44: The excited spin waves have half the pumping frequency.
45: The $h_{\rm crit}$ curve plotted against the static field is called
46: ``butterfly curve''.
47: The spin-wave instability for parallel pumping is often investigated 
48: for the purpose to
49: study relaxation phenomena in ferromagnetic materials, since the
50: instability threshold depends on the balance between the driving power
51: and the damping of spin waves.
52: 
53: Theoretical studies have not been very successful to explain 
54: the instability threshold for parallel pumping:    
55: a combination of 
56: more than two equations (theories) is needed to obtain the butterfly
57: curve. In other words, 
58: even a qualitative explanation for the butterfly curve cannot be
59: successful so long as one tries to use only the Landau-Lifshitz (LL)
60: equation governing the magnetization dynamics.  
61: By contrast the Suhl instability
62: for perpendicular pumping has been well explained
63: theoretically~\cite{Suhl}. 
64: The Suhl instability can be described by using only the LL equation. 
65: For parallel pumping, however, the LL equation 
66: is not enough to explain the instability, 
67: and the spin-wave line width $\Delta H_k$
68: is necessary to describe the relaxation of spin-waves as well as the
69: instability.  
70: 
71: A typical butterfly curve for parallel pumping has a cusp at a certain
72: external static field, which can be
73: explained if the spin-wave line width $\Delta H_k$ is given. For
74: example, Patton {\it et al.} proposed some trial $\Delta H_k$ functions
75: and successfully explained butterfly curves~\cite{Patton}.
76: Such a typical feature of butterfly curves usually does not depend on the
77: shape of samples. For very thin films, however, more interesting features
78: can appear. In fact, butterfly curves with multiple cusps were
79: observed for $0.5$-$\mu$m yttrium iron garnet (YIG) films under in-plane
80: external field~\cite{Kalin}. Those cusps were due to
81: quantized standing-wave modes across the film. 
82: It is highly desirable to develop the theory of multi-cusp butterfly
83: curves in the case of a general oblique field.
84: 
85: In this paper, we examine the instability threshold for parallel
86: pumping in ferromagnetic thin films  theoretically. 
87: The butterfly curve of the threshold will be calculated in cases where
88: the external field has an oblique angle to the film plane. 
89: For the calculation of the threshold, 
90: we use a trial $\Delta H_k$ function proposed by
91: Patton {\it et. al.}~\cite{Patton},
92: which was proposed originally for spherical samples but can be applied
93: to thin films~\cite{Wiese,Kabos}.
94: For usual thin films, the butterfly curve of the threshold has a single cusp at
95: a certain value of 
96: the static external field. We will show how the cusp shifts when the
97: oblique angle changes. Moreover, we will reveal that multi-cusp butterfly
98: curves can be seen for very thin films.
99: 
100: \section{Equations of motion for the spin-wave amplitudes}
101: 
102: The dynamics of magnetization field $\vec{M}(\vec{r})$ is governed by
103: the Landau-Lifshitz (LL) equation,
104: \begin{equation}
105: \frac1{\gamma}\partial_t \vec{M} =\vec{M}\times\vec{H}_{\rm eff}
106:  -\frac{\lambda}{M_0}\vec{M}\times (\vec{M}\times\vec{H}_{\rm eff}).
107:  \label{eq:LL}
108: \end{equation}
109: Here, $\gamma$ is the gyromagnetic ratio ($\gamma <0$ for spins);
110:  $M_0$ is the value of the
111: magnetization in thermal equilibrium; $\lambda$ is a damping constant
112: parameter ($\lambda <0$ in this case); 
113: $\vec{H}_{\rm eff}$ is the effective magnetic field:
114: \begin{equation}
115: \vec{H}_{\rm eff}=D\nabla^2\vec{M} +\vec{H}^{\rm d}+\vec{H}^{\rm a} 
116: +\vec{H}_0 +\vec{h}\cos\omega t.
117: \label{eq:Heff}
118: \end{equation}  
119: The first term comes from the exchange interaction; the third term is
120: the anisotropy field, which is omitted below for convenience;
121: the fourth and fifth
122: terms are the external static and pumping fields, respectively. For
123: parallel pumping, $\vec{H}_0 // \vec{h}$ and they are parallel to the
124: $z$ axis.
125: The second term on right hand side of Eq.~(\ref{eq:Heff}) is the
126: demagnetizing filed given by the gradient of the magneto-static potential
127: $\phi$: 
128: \begin{equation}
129:  \vec{H}^{\rm d}=-\vec{\nabla}\phi .
130: \end{equation}
131: The magneto-static potential obeys the Poisson equation:
132: \begin{equation}
133:  \nabla^2\phi = \left\{ 
134: \begin{array}{rl}
135:  4\pi\vec{\nabla}\cdot\vec{M}, & \mbox{inside the sample}, \\
136:  0, & \mbox{outside the sample}.
137: \end{array}
138: \right.
139: \label{eq:poi}
140: \end{equation}
141: 
142: The number of independent components of $\vec{M}$ is two, since the
143: length of the magnetization vector is invariant ($|\vec{M}|=M_0$).
144: The normalized magnetization $\vec{S}=\vec{M}/M_0$ can be
145: represented by a point on a unit sphere.
146: It is convenient to project the unit sphere stereographically onto a
147: complex variable $\psi (\vec{r},t)$~\cite{Laksh}:
148: \begin{equation}
149:  \psi=\frac{S_x+\mathrm{i}S_y}{1+S_z},
150: \end{equation}
151: where
152: \begin{equation}
153:  S_x=\frac{\psi+\psi^*}{1+\psi\psi^*}, \quad 
154:  S_y=\frac{\mathrm{i}(\psi^* -\psi)}{1+\psi\psi^*}, \quad
155:  S_z=\frac{1-\psi\psi^*}{1+\psi\psi^*}.
156: \label{eq:Ss}
157: \end{equation}
158: In terms of $\psi$, the LL equation~(\ref{eq:LL}) is rewritten as 
159: \begin{eqnarray}
160:  \partial_t \psi -\mathrm{i}(1-\mathrm{i}\lambda )&\gamma&
161:  \left\{ D M_0 \left[ \nabla^2\psi 
162:  -\frac{2\psi^*(\nabla\psi)^2}{1+\psi\psi^*} \right]
163:  -\frac12 (1-\psi^2)\partial_x\phi \right. \nonumber\\
164:  &-&\left. \frac{\rm i}2 (1+\psi^{*2})\partial_y\phi 
165:  +(\partial_z\phi-H_0-h\cos\omega t)\psi 
166: \right\} =0.
167: \label{eq:LLs}
168: \end{eqnarray}
169: Inside the sample, $\phi$ satisfies
170: \begin{eqnarray}
171:  \nabla^2\phi&=&\frac{4\pi M_0}{(1+\psi\psi^*)^2}\left\{
172:  (1-\psi^{*2})\partial_x\psi+(1-\psi^2)\partial_x\psi^* \right.\nonumber\\
173:  &+&\left. \mathrm{i}[(1+\psi^2)\partial_y\psi^*-(1+\psi^{*2})\partial_y\psi ]
174:  -2(\psi\partial_z\psi^* +\psi^*\partial_z\psi) \right\}.
175: \label{eq:pois}
176: \end{eqnarray}
177: 
178: %%% Figure 1 %%%
179: \begin{figure}
180: \begin{center}
181:  \includegraphics[width=8cm]{fig1.eps}
182: % \includegraphics[width=8cm]{setting.eps}
183: \end{center}
184: \caption{Schematic picture of sample setting. The sample extends on the
185:  $\tilde{x}$-$\tilde{y}$ plane. The $z$ axis corresponds to the direction of
186:  the external filed and has an angle $\alpha$ to the $\tilde{z}$ axis.}
187: \label{fig:setting}
188: \end{figure}
189: %%%%%%%%%%%%%%%
190: Now let us consider the boundary conditions, which affect the
191: demagnetizing field. We assume a film of thickness $d$ infinitely
192: extended in the $\tilde{x}$-$\tilde{y}$ plane as shown in
193: Fig.~\ref{fig:setting}. The external field is at an angle $\alpha$ to
194: the $\tilde{z}$ axis and corresponds to the $z$ axis.
195: We assume unpinned surface spins, which satisfy Neumann-like boundary
196: conditions, 
197: \begin{equation}
198:  \partial_{\tilde{z}} \vec{S} |_{\tilde{z}=\pm d/2} =0,
199: \end{equation}
200: namely,
201: \begin{equation}
202:   \left. \frac{\partial}{\partial\tilde{z}}\psi
203:   \right|_{\tilde{z}=\pm d/2}
204:  =\left. \frac{\partial}{\partial\tilde{z}}\psi^*
205:   \right|_{\tilde{z}=\pm d/2}=0.
206: \label{eq:BC}
207: \end{equation}
208: 
209: Before proceeding to the linear instability of spin waves, we introduce the
210: following dimensionless time and space units~\cite{Elmer}:
211: \begin{equation}
212:  t\to \frac{t}{4\pi |\gamma |M_0}, \quad \vec{r}\to\vec{r}d.
213: \end{equation}
214: Then we obtain the linearized equations of motion corresponding to
215: Eqs.~(\ref{eq:LLs}) and (\ref{eq:pois}): 
216: \begin{equation}
217: \partial_t\psi +\mathrm{i}(1-\mathrm{i}\lambda)\left[
218: l^2\nabla^2\psi -\frac12(\partial_x +\mathrm{i}\partial_y)\Phi
219: +(\partial_z\Phi -\omega_H -\omega_h\cos\omega_{\rm p}t)\psi
220: \right] =0,
221: \label{eq:LLl} 
222: \end{equation}
223: \begin{equation}
224:  \nabla^2\Phi =(\partial_x -\mathrm{i}\partial_y)\psi
225:  +(\partial_x +\mathrm{i}\partial_y)\psi^* \quad 
226:  \mbox{when $-\frac12 < \tilde{z} < \frac12$},
227: \label{eq:poil} 
228: \end{equation}
229: where
230: \begin{equation}
231:  l^2=\frac{D}{4\pi d^2}, \quad \Phi =\frac{\phi}{4\pi M_0d}, \quad
232:  \omega_H =\frac{H_0}{4\pi M_0}, \quad \omega_h =\frac{h}{4\pi M_0},
233:  \quad \omega_{\rm p}=\frac{\omega}{4\pi M_0|\gamma |}.
234: \end{equation}
235: 
236: Now we consider the undriven case (i.e., $\omega_h=0$),
237: and expand $\psi(\vec{r},t)$ and $\psi^*(\vec{r},t)$ 
238: so that they fulfills the boundary
239: conditions~(\ref{eq:BC}). For even modes, $\tilde{k}_z=2m\pi$ ($m$:
240: integer),
241: \begin{eqnarray}
242:  \psi (\tilde{\vec{r}},t)&=&\sum_{\tilde{\vec{\scriptstyle k}}}
243:  a_{\tilde{\vec{\scriptstyle k}}}(t) 
244:  \e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}
245:  \cos\tilde{k}_z\tilde{z} \nonumber\\
246:  \psi^* (\tilde{\vec{r}},t)&=&\sum_{\tilde{\vec{\scriptstyle k}}}
247:  a^*_{-\tilde{\vec{\scriptstyle k}}}(t) 
248:  \e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}
249:  \cos\tilde{k}_z\tilde{z}. 
250: \label{eq:expan_a}
251: \end{eqnarray}
252: For odd modes, $\tilde{k}_z=(2m+1)\pi$ ($m$: integer),
253: \begin{eqnarray}
254:  \psi (\tilde{\vec{r}},t)&=&\sum_{\tilde{\vec{\scriptstyle k}}}
255:  a_{\tilde{\vec{\scriptstyle k}}}(t) 
256:  \e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}
257:  \sin\tilde{k}_z\tilde{z} \nonumber\\
258:  \psi^* (\tilde{\vec{r}},t)&=&-\sum_{\tilde{\vec{\scriptstyle k}}}
259:  a^*_{-\tilde{\vec{\scriptstyle k}}}(t) 
260:  \e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}
261:  \sin\tilde{k}_z\tilde{z}. 
262: \label{eq:expan_b}
263: \end{eqnarray}
264: Here, $\tilde{k}_x=k_x\cos\alpha +k_z\sin\alpha$, $\tilde{k}_y=k_y$, and
265: $\tilde{k}_z=k_z\cos\alpha -k_x\sin\alpha$.
266: Using the expansions~(\ref{eq:expan_a}) and (\ref{eq:expan_b}), we obtain 
267: solutions of Eq.~(\ref{eq:poil}): for even modes,
268: \begin{eqnarray}
269:  \Phi (\tilde{\vec{r}})&=&-\sum_{\tilde{\vec{\scriptstyle k}}}
270:  \frac{\e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}}
271:  {\tilde{k}^2} \left[
272: \mathrm{i}(\tilde{k}_{-}a_{\tilde{\vec{\scriptstyle k}}}
273: +\tilde{k}_{+}a^*_{-\tilde{\vec{\scriptstyle k}}})
274: \left\{ \cos\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
275: \cosh\tilde{k}_\perp\tilde{z}  \right\} \right. \nonumber\\
276: &+& \left. \tilde{k}_z\sin\alpha\cdot (a_{\tilde{\vec{\scriptstyle k}}} +
277: a^*_{-\tilde{\vec{\scriptstyle k}}})\left\{
278: \sin\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
279: (\tilde{k}_z/\tilde{k}_\perp )
280: \sinh\tilde{k}_\perp\tilde{z} \right\}
281: \right],
282: \end{eqnarray}
283: for odd modes,
284: \begin{eqnarray}
285:   \Phi (\tilde{\vec{r}})&=&-\sum_{\tilde{\vec{\scriptstyle k}}}
286:  \frac{\e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}}
287:  {\tilde{k}^2} \left[
288: \mathrm{i}(\tilde{k}_{-}a_{\tilde{\vec{\scriptstyle k}}}
289: -\tilde{k}_{+}a^*_{-\tilde{\vec{\scriptstyle k}}})
290: \left\{ \sin\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
291: \sinh\tilde{k}_\perp\tilde{z}  \right\} \right. \nonumber\\
292: &-& \left. \tilde{k}_z\sin\alpha\cdot (a_{\tilde{\vec{\scriptstyle k}}} -
293: a^*_{-\tilde{\vec{\scriptstyle k}}})\left\{
294: \cos\tilde{k}_z\tilde{z}+(-1)^m\e^{-\tilde{k}_\perp /2}
295: (\tilde{k}_z/\tilde{k}_\perp )
296: \cosh\tilde{k}_\perp\tilde{z} \right\}
297: \right],
298: \end{eqnarray}
299: where $\tilde{k}_{+}=\tilde{k}_x\cos\alpha +\mathrm{i}\tilde{k}_y$, 
300: $\tilde{k}_{-}=\tilde{k}_x\cos\alpha -\mathrm{i}\tilde{k}_y$ and
301: $\tilde{k}_\perp =\sqrt{\tilde{k}_x^2+\tilde{k}_y^2}$.
302: The detailed derivation of these solutions is shown in
303: Appendix~\ref{sec:app}.
304: Here we approximate $\partial_z\Phi=-H^{\rm d}_z/4\pi M_0$ 
305: with the value of uniform magnetization, $\vec{k}=0$: 
306: $H^{\rm d}_z=H^{\rm d}_{\tilde{x}}\sin\alpha 
307: +H^{\rm d}_{\tilde{z}}\cos\alpha =-4\pi 
308: (N_{\tilde{x}}M_{\tilde{x}}\sin\alpha 
309: +N_{\tilde{z}} M_{\tilde{z}}\cos\alpha)$, where $N_{\tilde{x}}$ and 
310: $N_{\tilde{z}}$ are
311: demagnetizing factors. In this case, $N_{\tilde{x}}=N_{\tilde{y}}=0$ and
312: $N_{\tilde{z}}=1$ .
313: Assuming $M_z\simeq M_0$, namely, 
314: $M_{\tilde{z}}\simeq M_0\cos\alpha$, we obtain  
315: $\partial_z\Phi =-H^{\rm d}_z/4\pi M_0=\cos^2 \alpha$.
316: Then Eq.~(\ref{eq:LLl}) for $\omega_h=0$ is rewritten as
317: \begin{equation}
318: \partial_t\psi +\mathrm{i}(1-\mathrm{i}\lambda)\left[
319: l^2\nabla^2\psi -\frac12(\cos\alpha\partial_{\tilde{x}} 
320:  +\mathrm{i}\partial_{\tilde{y}}-\sin\alpha\partial_{\tilde{z}})\Phi
321: +(\cos^2 \alpha -\omega_H )\psi \right] =0.
322: \label{eq:LLl2}
323: \end{equation}
324: Combining Eqs.~(\ref{eq:expan_a})-(\ref{eq:LLl2}), we obtain
325: \begin{equation}
326: \partial_t a_{\vec{\scriptstyle k}}
327:  +\mathrm{i}(1-\mathrm{i}\lambda) A_{\vec{\scriptstyle k}}
328:    a_{\vec{\scriptstyle k}}
329:  +\mathrm{i}(1-\mathrm{i}\lambda) B_{\vec{\scriptstyle k}}
330:    a^*_{-\vec{\scriptstyle k}}=0, 
331: \label{eq:al}
332: \end{equation}
333: where
334: \begin{equation}
335:  A_{\vec{\scriptstyle k}}= C_{\vec{\scriptstyle k}}+ 
336:  \mathrm{Re}\tilde{C_{\vec{\scriptstyle k}}}, \quad
337:  B_{\vec{\scriptstyle k}}= (-1)^n\left(
338: D_{\vec{\scriptstyle k}}
339: \e^{2\mathrm{i}\varphi_{\vec{\scriptscriptstyle k}}}
340: +\tilde{C_{\vec{\scriptstyle k}}}
341: \right). 
342: \end{equation}
343: Here, $n$ is an integer, 
344: $\cos\varphi_{\vec{\scriptstyle k}}=k_x/k_\perp$,
345: $\sin\varphi_{\vec{\scriptstyle k}}=k_y/k_\perp$,
346: $k_{\perp}=\sqrt{k_x^2+k_y^2}$, and
347: \begin{eqnarray}
348: C_{\vec{\scriptstyle k}}&=&\cos^2 \alpha -l^2 k^2 -\omega_H-\frac12 
349: f_{\vec{\scriptstyle k}}\sin^2\theta_{\vec{\scriptstyle k}},\nonumber\\
350: \tilde{C_{\vec{\scriptstyle k}}}&=&-\frac{\tilde{k}_z}{k^2}\sin\alpha
351: \left( k_{\perp}\e^{\mathrm{i}\varphi_{\vec{\scriptscriptstyle k}}}
352: +\tilde{k}_z\sin\alpha \right) f_{\vec{\scriptstyle k}},\nonumber\\
353: D_{\vec{\scriptstyle k}}&=&-\frac12 f_{\vec{\scriptstyle k}}
354: \sin^2\theta_{\vec{\scriptstyle k}},\nonumber\\
355: f_{\vec{\scriptstyle k}}&=&1-(2-\delta_{0n})\left(
356: 1-\exp \left[ -\sqrt{k^2-\tilde{k}_z^2}\right] \right)
357: \frac{\sqrt{k^2-\tilde{k}_z^2}}{k^2},\nonumber\\
358: &&\sin\theta_{\vec{\scriptstyle k}}=\frac{k_\perp}{k}, \quad 
359: \tilde{k}_z =\pi n.
360: \label{eq:CDf}
361: \end{eqnarray} 
362: The detailed derivation is given in Appendix~\ref{sec:app_b}. 
363: 
364: Equation~(\ref{eq:al}) represents two coupled harmonic oscillators 
365: $a_{\vec{\scriptstyle k}}$, $a^*_{-\vec{\scriptstyle k}}$.
366: We diagonalize Eq.~(\ref{eq:al}) by means of the Holstein-Primakoff
367: transformation: 
368: \begin{eqnarray}
369:  a_{\vec{\scriptstyle k}}&=&
370:  \nu_{\vec{\scriptstyle k}}b_{\vec{\scriptstyle k}}
371: -\mu_{\vec{\scriptstyle k}}b^*_{-\vec{\scriptstyle k}}\nonumber\\
372:  a^*_{-\vec{\scriptstyle k}}&=&
373:  \nu_{\vec{\scriptstyle k}}b^*_{-\vec{\scriptstyle k}}
374: -\mu^*_{\vec{\scriptstyle k}}b_{\vec{\scriptstyle k}}, 
375: \label{eq:H-P}
376: \end{eqnarray}
377: where
378: \begin{eqnarray}
379:   \nu_{\vec{\scriptstyle k}}&=&
380:  \cosh\frac{\chi_{\vec{\scriptstyle k}}}{2},\nonumber\\
381:  \mu_{\vec{\scriptstyle k}}&=&
382:  \e^{\mathrm{i}\beta_{\vec{\scriptscriptstyle k}}}
383:  \sinh\frac{\chi_{\vec{\scriptstyle k}}}{2},\nonumber\\
384:  \cosh\chi_{\vec{\scriptstyle k}}&=&
385:  \frac{\left| C_{\vec{\scriptstyle k}}+
386:  \mathrm{Re}\tilde{C_{\vec{\scriptstyle k}}}\right| }
387: {\left[ (C_{\vec{\scriptstyle k}}+
388:  \mathrm{Re}\tilde{C_{\vec{\scriptstyle k}}})^2-(1+\lambda^2)
389: \left| D_{\vec{\scriptstyle k}}
390:  \e^{2\mathrm{i}\varphi_{\vec{\scriptscriptstyle k}}}+
391: \tilde{C_{\vec{\scriptstyle k}}}\right|^2 \right]^{1/2}}.
392: \end{eqnarray}
393: Substituting Eq.~(\ref{eq:H-P}) into Eq.~(\ref{eq:al}), we obtain
394: \begin{equation}
395:  \partial_t b_{\vec{\scriptstyle k}} -\mathrm{i}
396: (\omega_{\vec{\scriptstyle k}}+\mathrm{i}\eta_{\vec{\scriptstyle k}})
397:  b_{\vec{\scriptstyle k}}=0, 
398: \end{equation}
399: where
400: \begin{eqnarray}
401:  \omega_{\vec{\scriptstyle k}}^2 &=&(C_{\vec{\scriptstyle k}}+
402:  \mathrm{Re}\tilde{C_{\vec{\scriptstyle k}}})^2
403:  -(1+\lambda^2) \left| D_{\vec{\scriptstyle k}}
404:  \e^{2\mathrm{i}\varphi_{\vec{\scriptscriptstyle k}}}+
405:  \tilde{C_{\vec{\scriptstyle k}}}\right|^2,
406:  \label{eq:disp} \\
407:  \eta_{\vec{\scriptstyle k}}&=&\lambda (C_{\vec{\scriptstyle k}}+
408:  \mathrm{Re}\tilde{C_{\vec{\scriptstyle k}}})
409: \label{eq:damp}
410: \end{eqnarray}
411: Equations~(\ref{eq:disp}) and (\ref{eq:damp}) express the dispersion
412: relation and a damping rate, respectively.
413: 
414: \section{The instability threshold}
415: 
416: Now we consider the case where the microwave field $\vec{h}\cos\omega t$
417: is applied. The equation of motion of $a_{\vec{\scriptstyle k}}$
418: corresponding to Eq.~(\ref{eq:al}) is 
419: \begin{equation}
420:  \partial_t a_{\vec{\scriptstyle k}}
421:  +\mathrm{i}(1-\mathrm{i}\lambda) A_{\vec{\scriptstyle k}}
422:    a_{\vec{\scriptstyle k}}
423:  +\mathrm{i}(1-\mathrm{i}\lambda) B_{\vec{\scriptstyle k}}
424:    a^*_{-\vec{\scriptstyle k}}
425:  -\mathrm{i}(1-\mathrm{i}\lambda) \omega_h\cos\omega_{\rm p}t\cdot
426:   a_{\vec{\scriptstyle k}}=0. 
427: \label{eq:ap}
428: \end{equation}
429: After the the Holstein-Primakoff transformation (\ref{eq:H-P}), 
430: we substitute the following equations into Eq.~(\ref{eq:ap}),
431: \begin{eqnarray}
432:  b_{\vec{\scriptstyle k}}(t)&=&b^o_{\vec{\scriptstyle k}}(t)
433: \exp [\mathrm{i}(\omega_{\rm p}/2)t-\eta_{\vec{\scriptstyle k}}t],
434: \nonumber\\
435:  b^*_{-\vec{\scriptstyle k}}(t)&=&b^{o*}_{-\vec{\scriptstyle k}}(t)
436: \exp [-\mathrm{i}(\omega_{\rm p}/2)t-\eta_{\vec{\scriptstyle k}}t],
437: \end{eqnarray}
438: since the resonance occurs at 
439: $\omega_{\vec{\scriptstyle k}}=\omega_{\rm p}/2$.
440: Then the slowly-varying variable $b^o_{\vec{\scriptstyle k}}$ satisfies
441: \begin{equation}
442:  \partial_t^2 b^o_{\vec{\scriptstyle k}} +
443:  \left[ \left( \omega_{\vec{\scriptstyle k}} 
444:  -\frac{\omega_{\rm p}}{2}\right)^2 
445:  - \left| \rho_{\vec{\scriptstyle k}} \right|^2 \right]
446:  b^o_{\vec{\scriptstyle k}}=0,
447: \label{eq:bp}
448: \end{equation} 
449: where
450: \begin{equation}
451:  \left|  \rho_{\vec{\scriptstyle k}} \right| =
452:  \omega_h\sqrt{1+\lambda^2}\frac{\left| D_{\vec{\scriptstyle k}}
453:  \e^{2\mathrm{i}\varphi_{\vec{\scriptscriptstyle k}}}+
454:  \tilde{C_{\vec{\scriptstyle k}}}\right| }
455:  {2\omega_{\vec{\scriptstyle k}}}. 
456: \label{eq:rho}
457: \end{equation}
458: Therefore, the exponentially increasing solution for 
459: $b_{\vec{\scriptstyle k}}$ is 
460: \begin{equation}
461:  b_{\vec{\scriptstyle k}}\propto \exp \left[
462: \left( \left| \rho_{\vec{\scriptstyle k}} \right| 
463: -\eta_{\vec{\scriptstyle k}} \right)t +\mathrm{i}(\omega_{\rm p}/2)
464: t \right],
465: \end{equation}
466: where
467: \begin{equation}
468:  \left| \rho_{\vec{\scriptstyle k}} \right| > \eta_{\vec{\scriptstyle k}}.
469: \end{equation}
470: The instability threshold $\omega_h^{\rm crit}$ is now given as
471: \begin{equation}
472:  \omega_h^{\rm crit}=\frac{\omega_{\rm p}}{\sqrt{1+\lambda^2}} 
473:  \min_{\vec{\scriptstyle k}} \left\{ 
474:  \frac{\eta_{\vec{\scriptstyle k}}}{\left| D_{\vec{\scriptstyle k}}
475:  \e^{2\mathrm{i}\varphi_{\vec{\scriptscriptstyle k}}}+
476:  \tilde{C_{\vec{\scriptstyle k}}}\right| } \right\}.
477: \label{eq:dam_dH}
478: \end{equation}
479: The threshold, Eq.~(\ref{eq:dam_dH}), is obtained by using only the LL
480: equation, but cannot explain the experimentally-observed instability
481: threshold for parallel pumping.
482: In fact, Eq.~(\ref{eq:dam_dH}) proves to give an instability curve
483: totally different from the real butterfly curve.
484: Since the relaxation phenomenon is essentially nonlinear one, 
485: the linearizion analysis is not sufficient to discuss the instability
486: threshold. The problem is beyond the purpose of this paper, and 
487: it will be discussed in a forthcoming paper~\cite{Kudo}.
488: This difficulty, however, can be overcome by replacing the damping rate 
489: $\eta_{\vec{\scriptstyle k}}$ by a suitable spin-wave line width.
490: Using the spin-wave line width $\Delta H_{\vec{\scriptstyle k}}$, 
491: we rewrite Eq.~(\ref{eq:dam_dH}) as
492: \begin{equation}
493:   \omega_h^{\rm crit}=\omega_{\rm p}
494:  \min_{\vec{\scriptstyle k}} \left\{ 
495:  \frac{\Delta H_{\vec{\scriptstyle k}}}{\left| D_{\vec{\scriptstyle k}}
496:  \e^{2\mathrm{i}\varphi_{\vec{\scriptscriptstyle k}}}
497:  \tilde{C_{\vec{\scriptstyle k}}}\right| } \right\}.
498: \label{eq:dH}
499: \end{equation}
500: Here we adopt a simple trial $\Delta H_{\vec{\scriptstyle k}}$
501: function~\cite{Patton},
502: \begin{equation}
503:  \Delta  H_{\vec{\scriptstyle k}}=A_0+A_1\sin^2(2\theta_k) +A_2k,
504: \label{eq:dH2}
505: \end{equation}
506: where $A_0$, $A_1$ and $A_2$ are adjustable parameters.
507: 
508: \section{Butterfly curves}
509: 
510: Typical butterfly curves of the threshold $\omega_h^{\rm crit}$ 
511: have a cusp at a
512: certain static field: as the static field increasing, $\omega_h^{\rm crit}$ 
513: decreases
514: below the cusp point and increases above that.
515: For static fields below the cusp point, the minimum threshold modes corresponds
516: to a spin wave propagating with $\theta_{\vec{\scriptstyle k}}=\pi /2$.
517: As the static field increases, the wave vector $k$ of the threshold
518: modes decreases, and $k \simeq 0$ at the cusp.
519: For static fields above the cusp point, the wave number remains at $k \simeq 0$
520: and $\theta_{\vec{\scriptstyle k}}$ decreases from $\pi /2$ to $0$.
521: 
522: %%% Figure 2 %%%
523: \begin{figure}
524:  \begin{center}
525:   \includegraphics[width=14cm]{fig2.eps}
526: %  \includegraphics[width=14cm]{thick.eps}
527:  \end{center}
528: \caption{Theoretical butterfly curves for $d=15\mu$m for some values of
529:  the oblique angle $\alpha$.}
530: \label{fig:15m}
531: \end{figure}
532: %%%%%%%%%%%%%%%%
533: Figure~\ref{fig:15m} shows theoretical butterfly curves calculated by
534: using 
535: Eqs.~(\ref{eq:dH}) and (\ref{eq:dH2}).  
536: The material parameters used in the calculation 
537: are typical values for
538: yttrium iron garnet (YIG) materials: 
539: $|\gamma |=1.77\times 10^7$rad/(s$\cdot$Oe), $4\pi M_0=1.75\times10^3$Oe,
540: $D=5.4\times 10^{-9}$Oe$\cdot$cm$^2$/rad$^2$. 
541: The other parameters are as follows: 
542: $\omega /(2\pi )=9.5\times 10^9$Hz, $d=15\mu$m, 
543: $A_0=3.0\times 10^{-2}/(4\pi M_0)$,
544: $A_1/A_0=0.4$, $A_2/A_0=3.0\times 10^{-4}/d$.  
545: We calculate the static field value for the cusp point with $k=0$ and 
546: $\theta_{\vec{\scriptstyle k}}$ in a bulk
547: approximation: we assume $f_{\vec{\scriptstyle k}}=1$ in
548: Eq.~(\ref{eq:CDf}). Setting 
549: $\omega_{\vec{\scriptstyle k}}=\omega_{\rm p}/2$ in Eq.~(\ref{eq:disp}),
550: we obtain the static field for the cusp point
551: \begin{equation}
552:  \omega_H^{\rm crit}=\cos^2\alpha 
553: -\frac12+\frac12\sqrt{\omega_{\rm p}^2+1}.
554: \end{equation}
555: While the cusp point shifts to lower static fields as the oblique angle
556: $\alpha$ increases, the typical shape of curves is found to show little
557: noteworthy change.
558: 
559: In some cases, however, 
560: curious butterfly curves can emerge. 
561: %%% Figure 3 %%%
562: \begin{figure}
563:  \begin{center}
564:   \includegraphics[width=7cm]{fig3.eps}
565: %  \includegraphics[width=7cm]{thin.eps}
566:  \end{center}
567: \caption{Theoretical butterfly curve for $d=5\mu$m. The external field
568:  is applied parallel to the film plane.}
569: \label{fig:5m}
570: \end{figure}
571: %%%%%%%%%%%%%%%%
572: Figure~\ref{fig:5m} is a theoretical butterfly curve for the case of
573: $d=5\mu$m and $\alpha=\pi/2$. 
574: The other parameters used in the calculation are the same as used in 
575: Fig.~\ref{fig:15m}.
576: There appear multiple cusps in the low field region of the butterfly
577: curve. 
578: These cusps are due to different standing spin-wave modes across the
579: film thickness. 
580: Let us recall that the $\tilde{z}$ component of wave vectors are
581: quantized. Effects of quantization are essential when the boundary
582: conditions become important. Comparing the results in
583: Figs.~\ref{fig:15m} and \ref{fig:5m}, one may say 
584: multi-cusp feature of a butterfly curve can
585: be seen when the thickness of a film is very small. 
586: This finding is supported by some experimental
587: studies~\cite{Kalin,Wiese}. In Ref.~\cite{Wiese}, a standard butterfly
588: curve of the instability was shown for 15.9-$\mu$m-thick yttrium iron
589: garnet (YIG) film under in-plane external field. However, in
590: Ref.~\cite{Kalin}, multiple butterfly curves appeared for a YIG film
591: under in-plane external field. The difference between the two is just
592: the thickness. The thickness of the film in Ref.~\cite{Kalin} was
593: 0.5$\mu$m.
594: 
595: \section{Discussion}
596: 
597: We have revealed how butterfly curves depend on the oblique angle between
598: the external field and the film plane. 
599: We have calculated theoretical butterfly curves by using the LL equation
600: together with the spin-wave line width  $\Delta H_k$.
601: The calculation was performed
602: under the assumption that parameters $A_0$, $A_1$ and $A_2$ for 
603: $\Delta H_k$ do not depend
604: on the oblique angle.  
605: These parameters were originally introduced
606: to fit a theoretical butterfly curve to experimental
607: data~\cite{Patton}, and
608: might depend on the oblique angle.
609: 
610: We have also shown qualitative features of the novel aspect of butterfly
611: curves with multiple cusps for parallel pumping.
612: Those multi-cusp curves come from quantized standing-wave modes
613: across the film. However, it is not clear how such cusps appear, 
614: since there remains an ambiguity about how to
615: evaluate $A_0$, $A_1$ and $A_2$. 
616: Figure~\ref{fig:5m} is one of possible theoretical curves. 
617: To proceed to quantitative evaluations, the
618: experimental test is highly desirable to confirm the
619: features predicted here.
620: 
621: \section*{Acknowledgments}
622: 
623: The authors thank to Prof. Mino of Okayama university for useful
624: discussion. One of the authors (K. K.) is supported by JSPS Research
625: Fellowships for Young Scientists.
626: 
627: \appendix
628: 
629: \section{\label{sec:app} Demagnetization field}
630: 
631: First, we should use the $(\tilde{x},\tilde{y},\tilde{z})$ coordinate
632: system rather than $(x,y,z)$ 
633: to use Eqs.~(\ref{eq:expan_a}) and (\ref{eq:expan_b}) in the
634: Poisson equation~(\ref{eq:poi}), and employ the transformation:
635: $\nabla^2\to \tilde{\nabla}^2$, 
636: $\partial_x\to\cos\alpha\partial_{\tilde{x}}-\sin\alpha\partial_{\tilde{z}}$.
637: We consider only the case of $-\frac12 < \tilde{z} < \frac12$.
638: For even modes, $\tilde{k}_z=2m\pi$ ($m$: integer), the Poisson
639: equation is rewritten as 
640: \begin{eqnarray}
641:  \tilde{\nabla}^2\Phi =\sum_{\tilde{\vec{\scriptstyle k}}}
642:  \e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}&&
643:  \left[ (\mathrm{i}\tilde{k}_{-}\cos\tilde{k}_z\tilde{z}
644:  +\tilde{k}_z\sin\alpha\sin\tilde{k}_z\tilde{z})
645:  a_{\vec{\scriptstyle k}}\right. \nonumber\\
646:  &+&\left. (\mathrm{i}\tilde{k}_{+}\cos\tilde{k}_z\tilde{z}
647:  +\tilde{k}_z\sin\alpha\sin\tilde{k}_z\tilde{z})
648:  a^*_{-\vec{\scriptstyle k}} \right],
649: \label{eq:A-poia}
650: \end{eqnarray}
651: where $\tilde{k}_{+}=\tilde{k}_x\cos\alpha +\mathrm{i}\tilde{k}_y$ and 
652: $\tilde{k}_{-}=\tilde{k}_x\cos\alpha -\mathrm{i}\tilde{k}_y$.
653: For odd modes, $\tilde{k}_z=(2m+1)\pi$ ($m$: integer),
654: \begin{eqnarray}
655:  \tilde{\nabla}^2\Phi =\sum_{\tilde{\vec{\scriptstyle k}}}
656:  \e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}&&
657:  \left[ (\mathrm{i}\tilde{k}_{-}\sin\tilde{k}_z\tilde{z}
658:  -\tilde{k}_z\sin\alpha\cos\tilde{k}_z\tilde{z})
659:  a_{\vec{\scriptstyle k}}\right. \nonumber\\
660:  &-&\left. (\mathrm{i}\tilde{k}_{+}\sin\tilde{k}_z\tilde{z}
661:  -\tilde{k}_z\sin\alpha\cos\tilde{k}_z\tilde{z})
662:  a^*_{-\vec{\scriptstyle k}} \right].
663: \label{eq:A-poib}
664: \end{eqnarray}
665: 
666: Here we note a well known fact: the Poisson equation,
667: \begin{eqnarray}
668:  \nabla^2\phi(\vec{r})=-4\pi\rho (\vec{r}),
669: \end{eqnarray} 
670: has the solution as
671: \begin{eqnarray}
672:  \phi(\vec{r})=\int\frac{\rho (\vec{r}')\d
673:   \vec{V}'}{|\vec{r}'-\vec{r}|}.
674: \label{eq:a_phi}
675: \end{eqnarray}
676: Applying Eq.~(\ref{eq:a_phi}), 
677: we obtain the solution of Eqs.~(\ref{eq:A-poia}) 
678: and (\ref{eq:A-poib}): for even modes,
679: \begin{eqnarray}
680:  \Phi(\tilde{\vec{r}})=-\frac1{4\pi}&\sum_{\tilde{\vec{\scriptstyle k}}}&
681: \left\{
682: \mathrm{i}(\tilde{k}_{-}a_{\tilde{\vec{\scriptstyle k}}}+
683:  \tilde{k}_{+}a^*_{-\tilde{\vec{\scriptstyle k}}})
684: \int\frac{\d \tilde{\vec{V}}'}{|\tilde{\vec{r}}'-\tilde{\vec{r}}|}
685: \e^{\mathrm{i}(\tilde{k}_x\tilde{x}'+\tilde{k}_y\tilde{y}')}
686: \cos\tilde{k}_z\tilde{z}' \right. \nonumber\\
687: &+&\left. \tilde{k}_z\sin\alpha\cdot (a_{\tilde{\vec{\scriptstyle k}}}+
688: a^*_{-\tilde{\vec{\scriptstyle k}}} )
689: \int\frac{\d \tilde{\vec{V}}'}{|\tilde{\vec{r}}'-\tilde{\vec{r}}|}
690: \e^{\mathrm{i}(\tilde{k}_x\tilde{x}'+\tilde{k}_y\tilde{y}')}
691: \sin\tilde{k}_z\tilde{z}'
692: \right\};
693: \label{eq:a5}
694: \end{eqnarray}
695: for odd modes,
696: \begin{eqnarray}
697:  \Phi(\tilde{\vec{r}})=-\frac1{4\pi}&\sum_{\tilde{\vec{\scriptstyle k}}}&
698: \left\{
699: \mathrm{i}(\tilde{k}_{-}a_{\tilde{\vec{\scriptstyle k}}}-
700:  \tilde{k}_{+}a^*_{-\tilde{\vec{\scriptstyle k}}})
701: \int\frac{\d \tilde{\vec{V}}'}{|\tilde{\vec{r}}'-\tilde{\vec{r}}|}
702: \e^{\mathrm{i}(\tilde{k}_x\tilde{x}'+\tilde{k}_y\tilde{y}')}
703: \sin\tilde{k}_z\tilde{z}'\right.  \nonumber\\
704: &-&\left. \tilde{k}_z\sin\alpha\cdot (a_{\tilde{\vec{\scriptstyle k}}}-
705: a^*_{-\tilde{\vec{\scriptstyle k}}} )
706: \int\frac{\d \tilde{\vec{V}}'}{|\tilde{\vec{r}}'-\tilde{\vec{r}}|}
707: \e^{\mathrm{i}(\tilde{k}_x\tilde{x}'+\tilde{k}_y\tilde{y}')}
708: \cos\tilde{k}_z\tilde{z}'
709: \right\}.
710: \label{eq:a6}
711: \end{eqnarray}
712: After integration, Eqs.~(\ref{eq:a5}) and (\ref{eq:a6}) become
713: \begin{eqnarray}
714:  \Phi(\tilde{\vec{r}})=&-&\sum_{\tilde{\vec{\scriptstyle k}}}\frac
715: {\e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}}
716: {\tilde{k}^2}
717: \left\{
718: \mathrm{i}(\tilde{k}_{-}a_{\tilde{\vec{\scriptstyle k}}}+
719:  \tilde{k}_{+}a^*_{-\tilde{\vec{\scriptstyle k}}})
720: \left[ \cos\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
721: \cosh\tilde{k}_\perp\tilde{z} \right]\right. \nonumber\\
722: &+&\left. \tilde{k}_z\sin\alpha\cdot (a_{\tilde{\vec{\scriptstyle k}}}+
723: a^*_{-\tilde{\vec{\scriptstyle k}}} )
724: \left[ \sin\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
725: \frac{\tilde{k}_z}{\tilde{k}_\perp}
726: \sinh\tilde{k}_\perp\tilde{z} \right]
727: \right\},
728: \end{eqnarray}
729: and
730: \begin{eqnarray}
731:  \Phi(\tilde{\vec{r}})=&-&\sum_{\tilde{\vec{\scriptstyle k}}}\frac
732: {\e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}}
733: {\tilde{k}^2}
734: \left\{
735: \mathrm{i}(\tilde{k}_{-}a_{\tilde{\vec{\scriptstyle k}}}-
736:  \tilde{k}_{+}a^*_{-\tilde{\vec{\scriptstyle k}}})
737: \left[ \sin\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
738: \sinh\tilde{k}_\perp\tilde{z} \right]\right. \nonumber\\
739: &-&\left. \tilde{k}_z\sin\alpha\cdot (a_{\tilde{\vec{\scriptstyle k}}}-
740: a^*_{-\tilde{\vec{\scriptstyle k}}} )
741: \left[ \cos\tilde{k}_z\tilde{z}+(-1)^m\e^{-\tilde{k}_\perp /2}
742: \frac{\tilde{k}_z}{\tilde{k}_\perp}
743: \cosh\tilde{k}_\perp\tilde{z} \right]
744: \right\},
745: \end{eqnarray}
746: respectively.
747: 
748: \section{\label{sec:app_b} Equation of motion for $a_{\vec{\scriptstyle k}}$}
749: 
750: First, let us calculate the derivative of $\Phi$ in Eq.~(\ref{eq:LLl2}).
751: For even modes, $\tilde{k}_z=2m\pi$ ($m$: integer),
752: \begin{eqnarray}
753:  \lefteqn{(\cos\alpha\partial_{\tilde{x}}+\mathrm{i}\partial_{\tilde{y}}
754: -\sin\alpha\partial_{\tilde{z}})\Phi
755: =\sum_{\tilde{\vec{\scriptstyle k}}}
756: \frac{\e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}}
757: {\tilde{k}^2}}
758: \nonumber\\
759: \times\biggl[  &-& \left\{
760: -(\tilde{k}_{+}\tilde{k}_{-}a_{\tilde{\vec{\scriptstyle k}}}
761: +\tilde{k}_{+}^2a^*_{-\tilde{\vec{\scriptstyle k}}})
762: \left[ \cos\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
763: \cosh\tilde{k}_\perp\tilde{z} \right] \right. \nonumber\\
764: &&+\left. \mathrm{i}\tilde{k}_{+}\tilde{k}_z\sin\alpha\cdot
765: (a_{\tilde{\vec{\scriptstyle k}}}+a^*_{-\tilde{\vec{\scriptstyle k}}})
766: \left[ \sin\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
767: (\tilde{k}_z /\tilde{k}_\perp )\sinh\tilde{k}_\perp\tilde{z} \right]
768: \right\} \nonumber\\
769: &+& \sin\alpha\left\{
770: \mathrm{i}(\tilde{k}_{-}a_{\tilde{\vec{\scriptstyle k}}}
771: +\tilde{k}_{+}a^*_{-\tilde{\vec{\scriptstyle k}}})
772: \left[ -\tilde{k}_z\sin\tilde{k}_z\tilde{z}-(-1)^m
773: \e^{-\tilde{k}_\perp /2}
774: \tilde{k}_\perp\sinh\tilde{k}_\perp\tilde{z} \right] 
775: \right. \nonumber\\
776: &&+\left. \tilde{k}_z^2\sin\alpha\cdot 
777: (a_{\tilde{\vec{\scriptstyle k}}}+a^*_{-\tilde{\vec{\scriptstyle k}}})
778: \left[ \cos\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
779: \cosh\tilde{k}_\perp\tilde{z} \right]
780: \right\} \biggr],
781: \label{eq:deri_a}
782: \end{eqnarray} 
783: where $\tilde{k}_{+}=\tilde{k}_x\cos\alpha +\mathrm{i}\tilde{k}_y$ and 
784: $\tilde{k}_{-}=\tilde{k}_x\cos\alpha -\mathrm{i}\tilde{k}_y$.
785: For odd modes, $\tilde{k}_z=(2m+1)\pi$ ($m$: integer),
786: \begin{eqnarray}
787:  \lefteqn{(\cos\alpha\partial_{\tilde{x}}+\mathrm{i}\partial_{\tilde{y}}
788: -\sin\alpha\partial_{\tilde{z}})\Phi
789: =\sum_{\tilde{\vec{\scriptstyle k}}}
790: \frac{\e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}}
791: {\tilde{k}^2}}
792: \nonumber\\
793: \times\biggl[  &-& \left\{
794: -(\tilde{k}_{+}\tilde{k}_{-}a_{\tilde{\vec{\scriptstyle k}}}
795: -\tilde{k}_{+}^2a^*_{-\tilde{\vec{\scriptstyle k}}})
796: \left[ \sin\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
797: \sinh\tilde{k}_\perp\tilde{z} \right] \right. \nonumber\\
798: &&-\left. \mathrm{i}\tilde{k}_{+}\tilde{k}_z\sin\alpha\cdot
799: (a_{\tilde{\vec{\scriptstyle k}}}+a^*_{-\tilde{\vec{\scriptstyle k}}})
800: \left[ \cos\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
801: (\tilde{k}_z /\tilde{k}_\perp )\cosh\tilde{k}_\perp\tilde{z} \right]
802: \right\} \nonumber\\
803: &+& \sin\alpha\left\{
804: \mathrm{i}(\tilde{k}_{-}a_{\tilde{\vec{\scriptstyle k}}}
805: -\tilde{k}_{+}a^*_{-\tilde{\vec{\scriptstyle k}}})
806: \left[ \tilde{k}_z\cos\tilde{k}_z\tilde{z}-(-1)^m
807: \e^{-\tilde{k}_\perp /2}
808: \tilde{k}_\perp\cosh\tilde{k}_\perp\tilde{z} \right] 
809: \right. \nonumber\\
810: &&+\left. \tilde{k}_z^2\sin\alpha\cdot 
811: (a_{\tilde{\vec{\scriptstyle k}}}-a^*_{-\tilde{\vec{\scriptstyle k}}})
812: \left[ \sin\tilde{k}_z\tilde{z}-(-1)^m\e^{-\tilde{k}_\perp /2}
813: \sinh\tilde{k}_\perp\tilde{z} \right]
814: \right\} \biggr].
815: \label{eq:deri_b}
816: \end{eqnarray} 
817: Let us project Eqs.~(\ref{eq:deri_a}) and (\ref{eq:deri_b}) onto 
818: $\cos\tilde{k}_z\tilde{z}$ and $\sin\tilde{k}_z\tilde{z}$, respectively.
819: For $-1/2 < \tilde{z} < 1/2$, 
820: the projection $F(\tilde{z})$ of a function $f(\tilde{z})$ onto 
821: $\cos\tilde{k}_z\tilde{z}$ is given by, 
822: \begin{equation}
823:  F(\tilde{z})=\int_{-1/2}^{1/2}\d\tilde{z} 
824:  f(\tilde{z})\cos\tilde{k}_z\tilde{z}
825:   \left/ \int_{-1/2}^{1/2}\d\tilde{z} \cos^2\tilde{k}_z\tilde{z}.\right.
826: \end{equation}
827: After the projection, Eqs.~(\ref{eq:deri_a}) and (\ref{eq:deri_b}) are
828: reduced to
829: \begin{eqnarray}
830:  (\cos\alpha\partial_{\tilde{x}}&+&\mathrm{i}\partial_{\tilde{y}}
831: -\sin\alpha\partial_{\tilde{z}})\Phi=
832: \sum_{\tilde{\vec{\scriptstyle k}}}
833: \frac{\e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}}{k^2}
834: \left\{ 1-(2-\delta_{m0})\left( 1-\e^{-\tilde{k}_\perp}\right) 
835:  \frac{\tilde{k}_\perp}{k^2}\right\} \nonumber\\
836: && \times\left\{ (\tilde{k}_{+}\tilde{k}_{-}+\tilde{k}_z^2\sin^2\alpha )
837: a_{\tilde{\vec{\scriptstyle k}}} 
838: + (\tilde{k}_{+}^2+ \tilde{k}_z^2\sin^2\alpha)
839: a^*_{-\tilde{\vec{\scriptstyle k}}} \right\}\cos\tilde{k}_z\tilde{z},
840: \end{eqnarray}
841: where $\delta_{ij}$ is the Kronecker delta, and
842: \begin{eqnarray}
843:  (\cos\alpha\partial_{\tilde{x}}&+&\mathrm{i}\partial_{\tilde{y}}
844: -\sin\alpha\partial_{\tilde{z}})\Phi=
845: \sum_{\tilde{\vec{\scriptstyle k}}}
846: \frac{\e^{\mathrm{i}(\tilde{k}_x\tilde{x}+\tilde{k}_y\tilde{y})}}{k^2}
847: \left\{ 1-2\left( 1-\e^{-\tilde{k}_\perp}\right) 
848: \frac{\tilde{k}_\perp}{k^2} \right\} \nonumber\\
849: && \times\left\{ (\tilde{k}_{+}\tilde{k}_{-}+\tilde{k}_z^2\sin^2\alpha )
850: a_{\tilde{\vec{\scriptstyle k}}} 
851: - (\tilde{k}_{+}^2 +\tilde{k}_z^2\sin^2\alpha)
852: a^*_{-\tilde{\vec{\scriptstyle k}}} \right\}\sin\tilde{k}_z\tilde{z}, 
853: \end{eqnarray}
854: respectively. 
855: Let us impose a restriction on 
856: the kinds of parameters about wave vectors to describe
857: the equation of motion: we only use $k$, 
858: $k_{\perp}=\sqrt{k_x^2+k_y^2}$, $\tilde{k}_z=\pi n$ ($n$: integer), 
859: $\theta_{\vec{\scriptstyle k}}$ and
860: $\varphi_{\vec{\scriptstyle k}}$. 
861: Here, $\sin\theta_{\vec{\scriptstyle k}}=k_{\perp}/k$;
862: $\cos\varphi_{\vec{\scriptstyle k}}=k_x/k_{\perp}$;
863: $\sin\varphi_{\vec{\scriptstyle k}}=k_y/k_{\perp}$.
864: The restriction leads to the following:
865: \begin{eqnarray}
866:  \tilde{k}_{+}\tilde{k}_{-}+\tilde{k}_z^2\sin^2\alpha &=&
867: k_\perp^2 +2 (k_\perp
868: \cos\varphi_{\tilde{\vec{\scriptstyle k}}} +\tilde{k}_z\sin\alpha)
869: \tilde{k}_z\sin\alpha;
870: \nonumber\\
871: \tilde{k}_{+}^2 +\tilde{k}_z^2\sin^2\alpha &=&
872: k_\perp^2\e^{2\mathrm{i}\varphi_{\vec{\scriptstyle k}}}
873: +2 (k_\perp
874: \e^{\mathrm{i}\varphi_{\vec{\scriptstyle k}}}
875: +\tilde{k}_z\sin\alpha)\tilde{k}_z\sin\alpha; \nonumber\\
876: \tilde{k}_\perp &=& \sqrt{k^2-\tilde{k}_z^2}.
877: \end{eqnarray}
878: From the above two equations and Eq.~(\ref{eq:LLl2}), we obtain
879: Eq.~(\ref{eq:al}):
880: \begin{equation}
881: \partial_t a_{\vec{\scriptstyle k}}
882:  +\mathrm{i}(1-\mathrm{i}\lambda) A_{\vec{\scriptstyle k}}
883:    a_{\vec{\scriptstyle k}}
884:  +\mathrm{i}(1-\mathrm{i}\lambda) B_{\vec{\scriptstyle k}}
885:    a^*_{-\vec{\scriptstyle k}}=0. 
886: \end{equation} 
887: For even modes,
888: \begin{equation}
889:  A_{\vec{\scriptstyle k}}= C_{\vec{\scriptstyle k}}+ 
890:  \mathrm{Re}\tilde{C_{\vec{\scriptstyle k}}}, \quad
891:  B_{\vec{\scriptstyle k}}= \left(
892: D_{\vec{\scriptstyle k}}
893: \e^{2\mathrm{i}\varphi_{\vec{\scriptscriptstyle k}}}
894: +\tilde{C_{\vec{\scriptstyle k}}}
895: \right),  
896: \label{eq:ABa}
897: \end{equation}
898: and for odd modes,
899: \begin{equation}
900:  A_{\vec{\scriptstyle k}}= C_{\vec{\scriptstyle k}}+ 
901:  \mathrm{Re}\tilde{C_{\vec{\scriptstyle k}}}, \quad
902:  B_{\vec{\scriptstyle k}}= -\left(
903: D_{\vec{\scriptstyle k}}
904: \e^{2\mathrm{i}\varphi_{\vec{\scriptscriptstyle k}}}
905: +\tilde{C_{\vec{\scriptstyle k}}}
906: \right).  
907: \label{eq:ABb}
908: \end{equation}
909: Here we note that 
910:  $C_{\vec{\scriptstyle k}}$, $\tilde{C_{\vec{\scriptstyle k}}}$
911: and $D_{\vec{\scriptstyle k}}$ are described in Eq.~(\ref{eq:CDf}).
912: The difference between Eqs.~(\ref{eq:ABa}) and (\ref{eq:ABb}) is just
913: the sign of $B_{\vec{\scriptstyle k}}$. Therefore, for both even and odd
914:  modes, we may write
915: \begin{equation}
916:  B_{\vec{\scriptstyle k}}= (-1)^n\left(
917: D_{\vec{\scriptstyle k}}
918: \e^{2\mathrm{i}\varphi_{\vec{\scriptscriptstyle k}}}
919: +\tilde{C_{\vec{\scriptstyle k}}}
920: \right). 
921: \end{equation}
922: 
923: \begin{thebibliography}{99}
924:  \bibitem{Schlomann} E.~Schl\"omann, J.~J.~Green and U.~Milano,
925: 		     J. Appl. Phys. {\bf 31}, 386S (1960).
926:  \bibitem{Suhl} H.~Suhl, J. Phys. Chem. Solids {\bf 1}, 209 (1957).
927:  \bibitem{Patton} See, for example, M.~Chen and C.~E.~Patton, 
928:  in {\it Nonlinear Phenomena and Chaos in Magnetic Materials}, edited by
929: 	 P.~E.~Wigen (World Scientific, Singapore, 1994), pp. 33-82.
930:  \bibitem{Kalin} B.~A.~Kalinikos, N.~G.~Kovshikov and N.~V.~Kozhus,
931: 	 Sov. Phys. Solid State {\bf 27}, 1681 (1986). [Fiz. Tverd. Tela
932: 	 (Leningrad) {\bf 27}, 2794 (1985).]
933:  \bibitem{Wiese} G.~Wiese, L.~Buxmzn, P.~Kabos and C.~E.~Patton,
934: 	 J.Appl. Phys. {\bf 75}, 1041 (1994).
935:  \bibitem{Kabos} P.~Kabos, M~Mendik, G.~Wiese and C.~E.~Patton,
936: 	 Phys. Rev. B {\bf 55}, 11457 (1997).
937:  \bibitem{Laksh} M.~Lakshmanan and K.~Nakamura, Phys. Rev. Lett. 
938: 	 {\bf 53}, 2497 (1984). 
939:  \bibitem{Elmer} F.~J.~Elmer, Phys. Rev. B {\bf 53}, 14323 (1996).
940:  \bibitem{Kudo} K.~Kudo and K.~Nakamura, in preparation.
941: \end{thebibliography}
942: \end{document}