1: % ****** Start of file F1.tex ******
2:
3:
4: %\documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
5: \documentclass[twocolumn,showpacs,showkeys,preprintnumbers,amsmath,amssymb]{revtex4}
6: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
7:
8: % Some other (several out of many) possibilities
9: %\documentclass[preprint,aps]{revtex4}
10: %\documentclass[preprint,aps,draft]{revtex4}
11: %\documentclass[prb]{revtex4}% Physical Review B
12:
13: \usepackage{graphicx}% Include figure files
14: \usepackage{dcolumn}% Align table columns on decimal point
15: \usepackage{bm}% bold math
16:
17: \newcommand{\bv}[1]{\mbox{\boldmath$#1$}}
18:
19:
20: %\nofiles
21:
22: \begin{document}
23:
24: \preprint{ }
25:
26: \title{A novel method to create a vortex in a Bose-Einstein condensate}% Force line breaks with \\
27:
28: \author{Shin-Ichiro Ogawa, Mikko M\"ott\"onen$^1$,
29: Mikio Nakahara$^{1,2}$, Tetsuo Ohmi$^{3}$,
30: and Hisanori Shimada$^{2}$\\}
31:
32:
33: \affiliation{%
34: Department of Physics, Osaka City University, Osaka 558-8585, Japan\\
35: $^1$Materials Physics Laboratory, Helsinki University of Technology, P.O. Box 2200 (Technical Physics), FIN-02015 HUT Finland\\
36: $^2$Department of Physics, Kinki University, Higashi-Osaka 577-8502, Japan\\
37: $^3$Department of Physics, Kyoto University, Kyoto 606-8502, Japan
38: }%
39:
40:
41:
42: \date{\today}% It is always \today, today,
43: % but any date may be explicitly specified
44:
45: \begin{abstract}
46: It has been shown that a vortex in a BEC with spin degrees of freedom
47: can be created by manipulating with external magnetic
48: fields. In the previous work, an optical plug along the vortex axis
49: has been introduced to avoid Majorana flips, which take place when the
50: external magnetic field vanishes along
51: the vortex axis while it is created. In the present work, in contrast, we
52: study the same scenario without introducing the optical plug.
53: The magnetic field vanishes only in the center of the vortex
54: at a certain moment of the evolution and hence we expect that the system will
55: lose only a fraction of the atoms by Majorana flips even
56: in the absence of an optical plug.
57: Our conjecture is justified by numerically solving the
58: Gross-Pitaevskii equation, where the full spinor degrees of freedom of the
59: order parameter are properly taken into account.
60: A significant simplification of the experimental realization of the scenario
61: is attained by the omission of the optical plug.
62: \end{abstract}
63:
64: \pacs{03.75.Fi, 67.57.Fg}% PACS, the Physics and Astronomy
65: % Classification Scheme.
66: \keywords{BEC, vortex, Gross-Pitaevskii equation, hyperfine spin}%Use showkeys class option if keyword
67: %display desired
68: \maketitle
69:
70:
71: \section{Introduction}
72:
73: Alkali atoms become a superfluid upon
74: Bose-Einstein condensation (BEC) [1, 2].
75: The superfluid properties of the system were considered to be
76: essentially the same as those of superfluid $^4$He, in spite
77: of the fact that the former is a weakly-coupled system while the latter
78: is coupled strongly. In contrast with $^4$He, however, alkali atoms
79: have internal degrees of freedom attributed to the
80: hyperfine spin ${\bv{F}}$ and, accordingly,
81: the order parameter has $2|F|+1$ components [3, 4]. The atoms
82: $^{23}$Na and $^{87}$Rb have $|F|=1$, for example, and their order parameters
83: have three components, similarly to the orbital or the spin part of
84: the superfluid $^3$He.
85:
86:
87: These degrees of freedom bring about remarkable
88: differences between the BEC of alkali atoms and that of $^4$He.
89: The hyperfine spin freezes along the direction of the local magnetic
90: field when a BEC is magnetically trapped.
91: If a BEC is trapped optically, on the other hand, these degrees of
92: freedom manifest themselves and various new phenomena, such as the phase
93: separation between the different spin states that have never been seen
94: in superfluid $^4$He, may be observed.
95:
96: It was suggested in [5] and [6] that a totally new process for
97: formation of a vortex in a BEC of alkali atoms
98: is possible by making use of the hyperfine
99: degrees of freedom to ``control'' the BEC.
100: Suppose a BEC is confined in a Ioffe-Pritchard trap.
101: Then a vortex state with two units of circulation can be continuously
102: created from the vortex free superfluid state by simply reversing
103: the axial magnetic field $B_z$, parallel to the Ioffe bars.
104: This field is created by a set of pinch coils and can be easily controlled.
105: Four groups have already reported the formation of vortices
106: with three independent methods [7-11].
107: Our method is totally different from the previous ones in
108: that the hyperfine spin degrees of freedom have been fully utilized.
109: The present paper is a sequel to [6]. Here, we conduct
110: further investigations
111: on the formation of a vortex by reversing the axial
112: field $B_z$. An optical plug along the axis was introduced in [5] and [6]
113: to avoid possible Majorana flips
114: near the axis, that may take place when $B_z$ passes through zero.
115: It is expected, however, that trap loss due to Majorana flips
116: may not be very significant if the
117: ``dangerous point'' $|B|=0$ is passed fast enough
118: and that considerable amount of the BEC remains in the trap as the result.
119: This process should not
120: be too fast, however, such that the adiabatic condition is still satisfied.
121: It is a very difficult task to introduce a sharply forcused
122: optical plug along the center of the condensate whose radius
123: is of the order of a few microns. Accordingly we expect that experimental
124: realization of our scenario will be much easier without the
125: optical plug. In the present paper, therefore, we analyze our scenario by
126: numerically integrating the multi-component Gross-Pitaevskii equation.
127:
128: This paper is organized as follows. In Sec. II, we outline
129: the order parameter and
130: the Gross-Pitaevskii equation for an $|F|=1$ BEC.
131: In Sec. III, the Gross-Pitaevskii equation is integrated numerically
132: to analyze the time dependence of the order parameter. It will be shown
133: that merely half of the condensate is lost from the trap if the
134: time dependence of the external magnetic field is chosen properly.
135: Sec. IV is devoted to conclusions and discussions.
136: In the Appendix, it is shown that the formation of a vortex
137: in the present scenario may be understood in terms of the Berry phase.
138:
139: \section{Order Parameter and Gross-Pitaevskii Equation of $|F|=1$ BEC}
140:
141: Let us briefly summarize the order parameter and the
142: Gross-Pitaevskii equation for a BEC with the
143: hyperfine spin $F=1$ to make this paper self-contained.
144: The readers should be referred to [3] and [6] for further details.
145:
146: The order parameter of a BEC with $|F|=1$
147: has three components $\Psi_i\ (i=-1, 0, +1)$ with respect to the
148: basis vectors $| i \rangle$ defined by $F_z| i\rangle = i|i\rangle$.
149: The order parameter $|\Psi \rangle$ is then expanded as
150: $$
151: |\Psi \rangle = \sum_{i=0, \pm 1} \Psi_i| i \rangle.
152: $$
153: It turns out, however, that another set of basis vectors $|a
154: \rangle\ (a=x,y,z)$, defined by $F_{a}|a\rangle=0$, is
155: more convenient for certain purposes. The order parameter is now expressed
156: as
157: $$
158: |\Psi \rangle =\sum_{a = x,y,z} \Psi_{a} |a \rangle.
159: $$
160: The transformation matrix from $\{\Psi_i\}$ to $\{\Psi_{a}\}$
161: is found in [6].
162:
163: The most general form of the Hamiltonian for $|F|=1$ atoms,
164: that is rotationally invariant except for the Zeeman term, is
165: \begin{eqnarray}
166: \hat{H}-\mu \hat{N} &=& \int \Big[
167: \psi^{\dagger}_{a}\left(-\frac{\hbar^2}{2m}\nabla^2-\mu \right)
168: \psi_{a} + \frac{g_1}{2} (\psi^{\dagger}_{a} \psi_{a})^2
169: \nonumber\\
170: & &+ \frac{g_2}{2} |\psi_{a} \psi_{a}|^2
171: + i \varepsilon_{a b c}\omega_{L c}
172: \psi^{\dagger}_{a} \psi_{b}\Big] d^3\bv{r},
173: \end{eqnarray}
174: where $\hbar \omega_{L a} = \gamma_{\mu} B_{a}$, $\gamma_{\mu}
175: \simeq \mu_B/2$
176: being the gyromagnetic ratio of the atom. The coupling constants
177: are given by
178: \begin{equation}
179: g_1 = \frac{4 \pi \hbar^2}{m} a_{2},\quad g_2 = \frac{4 \pi \hbar^2}{m}
180: \frac{a_0-a_2}{3},
181: \end{equation}
182: where $a_2 = 2.75\;\mathrm{nm}$ and $a_0 = 2.46\;\mathrm{nm}$ [12].
183:
184: The Heisenberg equation of motion derived from the above Hamiltonian is
185: \begin{eqnarray}
186: i \hbar
187: \frac{\partial \psi_{a}}{\partial t} &=&[\psi_{a}, \hat{H}-\mu \hat{N}]
188: \nonumber\\
189: &=& \left(-\frac{\hbar^2}{2m} \nabla^2-\mu \right) \psi_{a} + g_1
190: (\psi^{\dagger}_{b} \psi_{b}) \psi_{a}\nonumber\\
191: & &+ g_2 (\psi_{b} \psi_{b}) \psi_{a}^{\dagger} +
192: i \varepsilon_{a b c} \omega_{L c} \psi_{b}.
193: \end{eqnarray}
194: By taking the expectation value of the above equation in the
195: mean-field approximation, $\langle \psi \psi \psi \rangle \simeq \Psi \Psi
196: \Psi$, where $\Psi_{a} = \langle \psi_{a} \rangle$, we obtain
197: \begin{eqnarray}
198: i \hbar \frac{\partial \Psi_{a}}{\partial t} &=&
199: \left( -\frac{\hbar^2}{2m} \nabla^2-\mu \right) \Psi_{a}
200: + g_1|\Psi|^2 \Psi_{a}
201: \nonumber\\
202: & & + g_2 (\Psi)^2 \Psi^*_{a}+ i \varepsilon_{abc}
203: \omega_{Lc} \Psi_{b}.
204: \label{gp}
205: \end{eqnarray}
206: This is the fundamental equation which we will use in the rest of this paper.
207:
208: \section{Creation of a Vortex}
209:
210: \subsection{Magnetic Fields}
211:
212: Suppose we confine a BEC in a Ioffe-Pritchard trap
213: with a quadrupole magnetic field
214: \begin{equation}
215: \bv{B}_{\perp}(\bv{r}) = \left(
216: \begin{array}{c}
217: B_{\perp}(r) \cos(-\phi)\\
218: B_{\perp}(r) \sin (-\phi)\\
219: 0
220: \end{array} \right)
221: \end{equation}
222: and a uniform axial field
223: \begin{equation}
224: \bv{B}_{z}(t) = \left(
225: \begin{array}{c}
226: 0\\
227: 0\\
228: B_z(t)
229: \end{array} \right).
230: \end{equation}
231: Here $(r, \phi, z)$ are the cylindrical coordinates.
232: The magnitude $B_{\perp}(r)$ is proportional to $r$ near the axis $r \sim 0$;
233: $B_{\perp}(r) \simeq B' r$, with $B'$ being a constant.
234: The system is assumed to be uniform along the $z$ direction for
235: calculational simplicity. Our analysis should apply to a cigar-shaped
236: system as well.
237:
238: Suppose we reverse the $\bv{B}_z$ field slowly,
239: while keeping $\bv{B}_{\perp}$ fixed as shown in Fig. 1.
240: It was shown in [5, 6] that a vortex with two units of circulation
241: will be formed if we start with a vortex-free BEC.
242: In these papers,
243: an optical plug was introduced along the vortex axis
244: to prevent the atoms from escaping from the trap when
245: $\bv{B} = \bv{B}_{\perp} + \bv{B}_z$ vanishes at $r=0, t= T/2$.
246: Accordingly, the order parameter remains within the weak-field
247: seeking state (WFSS) throughout the scenario.
248: \\
249: \begin{figure}
250: \begin{center}
251: \includegraphics[width=5cm]{Fig1.eps}
252: \end{center}
253: \caption{The schematic time dependence of the axial field $B_z$.
254: The field $B_z$ reverses the sign in an interval of time $T$,
255: while the quadrupole field $\bv{B}_{\perp}=B' r$ remains fixed. The field
256: $B_z(t)$ is fixed at the value $B_z(T)=-B_z(0)$ for $t>T$ for
257: further evolution.}
258: \label{fig:1}
259: \end{figure}
260:
261: We suspect, however, that the BEC may be stable even without
262: the optical plug since $\bv{B}$ vanishes only in $r=0$ at the
263: time $t=T/2$. This will be justified by solving the Gross-Pitaevskii
264: equation numerically below.
265: In contrast with the previous work,
266: we have to take the full degrees of freedom of the order
267: parameter into account since the energies of the
268: three hyperfine states (the strong-field
269: seeking state, the weak-field seeking state and the neutral state)
270: are degenerate when $\bv{B}=0$.
271:
272: \subsection{Initial State}
273:
274: We first solve the Gross-Pitaevskii equation in the stationary state to find
275: the initial order parameter configuration and the corresponding
276: chemical potential. There is a sufficient gap
277: between the weak-field seeking state
278: and the other states at $t=0$ and the BEC
279: may be assumed to be purely in the weak-field seeking state.
280: Let us parametrize the external magnetic field $\bv{B}$ with use of the polar
281: angles $\alpha$ and $\beta$ as
282: \begin{equation}
283: \bv{B}(r, t) = \bv{B}_{\perp}(r) + \bv{B}_z(t) =|\bv{B}|
284: \left( \begin{array}{c}
285: \sin \beta \cos \alpha\\
286: \sin \beta \sin \alpha\\
287: \cos \beta
288: \end{array} \right),
289: \end{equation}
290: where $\alpha = -\phi$ for the quadrupole field and
291: \begin{equation}
292: \beta = \tan^{-1} \left[\frac{|\bv{B}_{\perp}(r)|}{|\bv{B}_z(t) |}\right].
293: \end{equation}
294: Since the hyperfine spin $\bv{F}$ is
295: antiparallel with $\bv{B}$, we must have
296: \begin{equation}
297: \hat{\bv{l}} =
298: \left( \begin{array}{c}
299: -\sin \beta \cos \alpha\\
300: -\sin \beta \sin \alpha\\
301: -\cos \beta
302: \end{array} \right),
303: \end{equation}
304: where $\hat{\bv{l}}$ is a unit vector parallel to $\bv{F}$, see Fig. 2.
305:
306: \begin{figure}
307: \begin{center}
308: \includegraphics[width=5cm]{fig2.eps}
309: \end{center}
310: \caption{Polar angles $(\alpha, \beta)$
311: parametrizing $\bv{B}$ and $\hat{\bv{l}} \parallel \bv{F}$.}
312: \label{fig:2}
313: \end{figure}
314:
315: The most general form for
316: the order parameter yielding the above $\hat{\bv{l}}$ vector is
317: \begin{equation}
318: \Psi_{a} = \frac{f_0}{\sqrt{2}} e^{-i \gamma}\left( \begin{array}{c}
319: \cos \beta \cos \alpha + i \sin \alpha\\
320: \cos \beta \sin \alpha - i \cos \alpha\\
321: -\sin \beta
322: \end{array} \right) \equiv f_0 v_{a},
323: \label{wfss}
324: \end{equation}
325: where $\{v_{a}\}$ represents the ``phase'',
326: while $f_0$ is the amplitude of the order parameter.
327: In mathematical terms, $\{v_{a}\}$ defines the local $SO(3)$ frame or the
328: ``triad'' of the real orthonormal vectors $\{\hat{\bv{m}}, \hat{\bv{n}},
329: \hat{\bv{l}\}}$, where
330: \begin{equation}
331: v_{a} = \frac{1}{\sqrt{2}}e^{-i\gamma} (\hat{\bv{m}}+i \hat{\bv{n}})_{a}, \quad \hat{\bv{l}} = \hat{\bv{m}} \times \hat{\bv{n}}.
332: \end{equation}
333: It should be noted that the above order parameter takes the same form as
334: that of the orbital part of the superfluid $^3$He-A.
335:
336: One may obtain more insight if the order parameter in Eq.~(\ref{wfss}) is rewritten
337: in the $\Psi_i$ basis as [13]
338: \begin{equation}
339: \Psi_{i} =f_0 e^{-i \gamma}
340: \left( \begin{array}{c}
341: \frac{1}{2} e^{-i \alpha}(1-\cos \beta)\\
342: -\frac{1}{\sqrt{2}} \sin \beta\\
343: \frac{1}{2} e^{i\alpha}(1+\cos \beta)
344: \end{array}
345: \right).
346: \end{equation}
347: The angle $\beta$ vanishes at $r=0$ (note that $\bv{B}_{\perp}=0$ at $r=0$),
348: hence
349: we find $\Psi_{+1}=
350: \Psi_{0} = 0$ while $\Psi_{-1} \propto e^{i(\alpha-\gamma)} \neq 0$ there.
351: For the order parameter to be smooth at $r=0$, we have to choose
352: \begin{equation}
353: \gamma = \alpha = -\phi.
354: \end{equation}
355: Accordingly, the order parameter takes the form
356: \begin{equation}
357: \Psi_{a} = \frac{f_0}{\sqrt{2}} e^{i \phi}\left( \begin{array}{c}
358: \cos \beta \cos \phi - i \sin \phi\\
359: -\cos \beta \sin \phi - i \cos \phi\\
360: -\sin \beta
361: \end{array} \right)
362: \label{wfa}
363: \end{equation}
364: in the $\Psi_{a}$ basis and
365: \begin{equation}
366: \Psi_{i}=\frac{f_0}{2}\left( \begin{array}{c}
367: (1-\cos \beta)e^{2i \phi}\\
368: -{\sqrt{2}} \sin \beta e^{i \phi}\\
369: 1+\cos \beta
370: \end{array} \right)
371: \label{wfi}
372: \end{equation}
373: in the $\Psi_i$ basis. It follows from Eq.~(\ref{wfi}) that the state with
374: $\hat{\bv{l}}=-\hat{\bv{z}}\ (\beta=0)$ has a vanishing winding number,
375: while that with $\hat{\bv{l}}=\hat{\bv{z}}\ (\beta=\pi)$ has the winding
376: number $2$. Accordingly, the former state may be continuously deformed
377: to the latter state by changing $\beta$ from $0$ to $\pi$ smoothly,
378: resulting in the formation of a vortex
379: with winding number 2 [14, 15]. It is shown in the
380: Appendix that this phase may also
381: be understood in terms of the Berry phase associated with the
382: adiabatic change of the local magnetic field.
383:
384: The stationary Gross-Pitaevskii equation is given by
385: \begin{equation}
386: -\frac{\nabla^2}{2m} \Psi_{a} + g_1 |\Psi|^2 \Psi_{a} + g_2 \Psi^2 \Psi_a^*
387: + i \varepsilon_{a b c} \Psi_{b}\omega_{L c}
388: = \mu \Psi_{a}.
389: \label{gps}
390: \end{equation}
391: By substituting the WFSS order parameter in Eq.~(\ref{wfa}) into Eq.~(\ref{gps}),
392: we obtain
393: \begin{eqnarray}
394: &&-\frac{\hbar^2}{2m}\Big[\nabla^2 -\frac{\beta^{\prime 2}}{2}-
395: \frac{1}{4r^2}(7-8 \cos \beta + \cos 2\beta)\Big] f_0\qquad \nonumber\\
396: & &\qquad \qquad + \gamma_{\mu} B(r) f_0 + g_1 f_0^3 = \mu f_0.
397: \end{eqnarray}
398: Note that there appears an extra term $\beta^{\prime 2}$ that is missing if
399: this BEC is described by a scalar (namely $U(1)$) order parameter. This term
400: originates from $v_{a}^{\dagger} \nabla^2 v_{a}$ and
401: represents the rotation of the local $SO(3)$ frame. In ordinary physical
402: settings, however, this term is smaller than the $B$-term by a factor of
403: $\omega/\omega_L \sim 10^{-3}$ and hence its effect may be negligible
404: in the following arguments.
405:
406: The amplitude of the external magnetic field has an
407: approximately harmonic
408: potential profile near the origin,
409: \begin{eqnarray}
410: B(r,t) &=&
411: \sqrt{B_{\perp}^2(r) + B_z^2(t)} \simeq \sqrt{B^{\prime 2} r^2 + B_z^2(t)}\nonumber\\
412: &\simeq& B_z(t) +\frac{B^{\prime 2}}{2B_z(t)}r^2.
413: \end{eqnarray}
414: It turns out to be convenient to scale the energies by the
415: energy-level spacing $\hbar \omega$ at $t=0$ and the
416: lengths by the harmonic-oscillator length $a_{\rm HO}$, where
417: \begin{eqnarray}
418: \omega &=& B' \sqrt{\frac{\gamma_{\mu}}{m B_z(0)}},\\
419: a_{\rm HO} &=& \sqrt{\frac{\hbar}{m \omega}}.
420: \end{eqnarray}
421: If we take $^{23}$Na and substitute $B_z (0)= 1 \;\mathrm{G}$ and $B' = 300
422: \;\mathrm{G/cm}$, we find $a_{\rm HO} \sim 9.14 \times 10^{-1}\;
423: \mathrm{\mu m}$
424: and $\hbar \omega \sim 3.49 \times 10^{-24}\;\mathrm{erg}$.
425: For the same choice of the parameters, we have $\omega_L(r=0, t=0) \sim
426: 4.40 \times 10^6\;{\mathrm{rad/s}} \sim 1.330 \times 10^3 \omega$.
427: It is reasonable to
428: assume
429: $\tau \equiv 2\pi/\omega_L(r=0, t=0)
430: \sim 1.43\;\mathrm{\mu s}$ to be the measure of the adiabaticity.
431:
432: After these scalings, the dimensionless Gross-Pitaevskii equation takes
433: a simpler form
434: \begin{eqnarray}
435: -\frac{1}{2}\tilde{\nabla}^2 \tilde{f}_0
436: &+&\left[\tilde{B}(r) +\frac{\beta^{\prime 2}}{4}
437: + \frac{1}{8\tilde{r}^2}
438: (7-8 \cos \beta + \cos 2\beta)\right] \tilde{f}_0\nonumber\\
439: & &+ \tilde{g}_1 \tilde{f}_0^3 = \tilde{\mu} \tilde{f}_0,
440: \end{eqnarray}
441: where $\tilde{r} =r/a_{\rm HO}, \tilde{f}_0 = f_0 a_{\rm HO}^{3/2},
442: \tilde{\mu} = \mu/\hbar \omega$, $\tilde{B} = \gamma_{\mu}B/\hbar \omega$
443: and $\tilde{g}_1 = g_1/a_{\rm HO}^3 \hbar \omega \sim 0.0378$.
444: A similar scaling for $g_2$ yields $\tilde{g}_2 \sim -0.00132$.
445: Note that the singularity at $\tilde{r}=0$ vanishes if $\beta(\tilde{r})$
446: approaches to 0 fast enough as $\tilde{r} \to 0$.
447: The tildes will be dropped hereafter whenever it does not cause
448: confusion. The eigenvalue $\mu$ may be obtained numerically. For
449: $^{23}$Na with $B_z$ and $B'$ given above, we find
450: ${\mu} - \hbar \omega_L(r=0, t=0) = 3.66$, which amounts
451: to $1.28 \times 10^{-23}\;{\mathrm erg}$ in dimensional units.
452: Figure 3 shows the corresponding condensate profile $f_0(r)$.
453: \\
454: \begin{figure}
455: \begin{center}
456: \includegraphics[width=6cm]{fig3.eps}
457: \end{center}
458: \caption{Condensate wave function $f_0(r)$.
459: The condensate wave function and the radial coordinate
460: are scaled by $a_{\rm HO}^{-3/2}$ and $a_{\rm HO}$, respectively,
461: and hence dimensionless.}
462: \label{fig:3}
463: \end{figure}
464:
465: \subsection{Time Development}
466:
467: The time development of the condensate wavefunction is obtained
468: by solving the Gross-Pitaevskii Eq. (\ref{gp}) which
469: is written in dimensionless form as
470: \begin{eqnarray}
471: i \frac{\partial \Psi_{a}}{\partial t} &=& -\frac{1}{2} \nabla^2
472: \Psi_{a}
473: + g_1|\Psi|^2 \Psi_{a} \nonumber\\
474: & &+ g_2 (\Psi)^2 \Psi^*_{a}
475: + i \varepsilon_{a b c} \Psi_{b} \omega_{Lc}.
476: \end{eqnarray}
477: The initial condition is
478: \begin{equation}
479: \Psi_{a} = f_0(r) v_{a},
480: \end{equation}
481: where $f_0(r)$ is obtained in the previous subsection
482: and $v_{a}$ is defined in Eq. (\ref{wfss}).
483: The condensate cannot remain within the weak-field seeking state during
484: the formation of a vortex and we have to utilize the full degrees of freedom
485: of the order parameter $\Psi_{a}$. The Gross-Pitaevskii equation
486: has been solved numerically to find the temporal development of the order
487: parameter
488: while $B_z$
489: changes as
490: \begin{equation}
491: B_z(t) = B_z(0) \left[ 1-\frac{2t}{T}\right] \quad (0 \leq t \leq T),
492: \end{equation}
493: with several choices for the reversing time $T$.
494: Adiabaticity is not guaranteed at $t\sim T/2$ when
495: the energy gaps among the weak-field seeking state (WFSS), the
496: neutral state (NS) and the strong-field seeking state (SFSS) disappear.
497: We expect, however, that this breakdown of
498: adiabaticity is not very significant to the condensate
499: since it takes place only along
500: the condensate axis for a short period of time.
501:
502: % 10 tau
503: \begin{figure}
504: \begin{center}
505: \includegraphics[width=8.5cm]{T10wav.eps}
506: %\epsfxsize=85mm
507: %\epsfbox{T10wav.eps}
508: \end{center}
509: \caption{Temporal development of the order parameters
510: $|\Psi_{\rm WFSS}|$, $|\Psi_{\rm NS}|$ and $|\Psi_{\rm SFSS}|$ for
511: $T= 10 \tau$, where $\tau \sim 1.43 \mu{\rm s}$.
512: The graphs are plotted for $t=0$ (initial state),
513: $t=T/2-\tau$ (slightly before $B_z=0$ is crossed), $t=T/2 + \tau$
514: (slightly after $B_z=0$ is crossed) and $t=T$ ($B_z$ is completely reversed).
515: Furthermore, $B_z$ is fixed to $B_z(T)=-B_z(0)$ for $t > T$.
516: The bottom row shows the order parameters at $t=1200 T$.
517: The order parameters and the radial coordinate are plotted in units of
518: $a_{\rm HO}^{-3/2}$ and $a_{\rm HO}$, respectively.}
519: \label{fig:4}
520: \end{figure}
521: %
522: \begin{figure}
523: \begin{center}
524: \includegraphics[width=7cm]{T10n.eps}
525: \end{center}
526: \caption{The ratio $N(t)/N(0)$
527: as a function of $t/T$ for the reversing time $T=10 \tau$. Here
528: $N(t)$ is the number of atoms$/$unit
529: length along the vortex axis at time $t$. The magnetic
530: field $B_z$ decreases monotonically
531: for $0 \leq t/T \leq 1$, but is kept fixed to $-B_z(0)$ for
532: $T \leq t$, see Fig. 1.}
533: \label{fig:5}
534: \end{figure}
535: %
536: % 100 tau
537: \begin{figure}
538: \begin{center}
539: \includegraphics[width=8.5cm]{T100wav.eps}
540: \end{center}
541: \caption{Same as in Fig. 4, but for
542: $T= 100 \tau$, $\tau \sim 1.43 \mu{\rm s}$.
543: The graphs are plotted for $t=0, t=T/2-\tau,
544: t=T/2 + \tau, t=T$, and $t=50 T$.}
545: \label{fig:6}
546: \end{figure}
547: %
548: \begin{figure}
549: \begin{center}
550: \includegraphics[width=7cm]{T100n.eps}
551: \end{center}
552: \caption{Same as in Fig. 5, but for $T=100 \tau$.}
553: \label{fig:7}
554: \end{figure}
555: %
556: %1000 tau
557: \begin{figure}
558: \begin{center}
559: \includegraphics[width=8.5cm]{T1000wav.eps}
560: \end{center}
561: \caption{Same as in Fig. 4, but for $T= 1000 \tau$, $\tau=1.43\mu{\rm s}$.
562: The graphs are plotted for $t=0, t=T/2-\tau,
563: t=T/2 + \tau, t=T$, and $t=4 T$.}
564: \label{fig:8}
565: \end{figure}
566: %
567: \begin{figure}
568: \begin{center}
569: \includegraphics[width=7cm]{T1000n.eps}
570: \end{center}
571: \caption{Same as in Fig. 5, but for $T=1000\tau$.}
572: \label{fig:9}
573: \end{figure}
574:
575: We project $\Psi_{a}$ thus obtained to the local
576: WFSS, NS and SFSS by defining the
577: projection operators to the respective states as
578: \begin{equation}
579: \begin{array}{c}
580: \Pi_{\rm{W}a b} = v_{a} v_{b}^{\dagger},\quad
581: \Pi_{\rm{S}ab} = u_{a} u_{b}^{\dagger},
582: \vspace{0.3cm}\\
583: \Pi_{\rm{N}ab} = \delta_{ab} - \Pi_{\rm{W}ab} -
584: \Pi_{\rm{S}ab}
585: \end{array}
586: \end{equation}
587: where
588: \begin{equation}
589: v_{a}= \frac{1}{\sqrt{2}}(\hat{\bv{m}}+i \hat{\bv{n}})_{a},
590: u_{a} = \frac{1}{\sqrt{2}}(\hat{\bv{m}}-i\hat{\bv{n}})_{a}
591: \end{equation}
592: define the local WFSS and SFSS $SO(3)$ frames, respectively.
593:
594: Figures 4, 6, and 8 depict the
595: amplitudes of the wave functions at $t=0, T/2-\tau, T/2 + \tau$ and $T$
596: for $T=10 \tau, 100 \tau$, and $1000 \tau$.
597: Figures 5, 7, and 9 show the particle numbers within the trap for
598: the same choices of $T$.
599: To simulate the trap loss, we introduced a function
600: \begin{equation}
601: h(r) = \frac{1}{2} \left[ 1 - \tanh\left(\frac{r-r_0}{\lambda} \right)
602: \right],
603: \end{equation}
604: with $r_0 = 30$ and $\lambda=2$. The wave function
605: $\Psi_{a}(r)$ is multiplied by $h(r)$ after each step
606: of the Crank-Nicholson algorithm. The parameter $r_0$ roughly corresponds to
607: the trap size, while $\lambda$ is taken large enough to prevent the
608: wavefunction being reflected at $r \sim r_0$.
609: Thus particles that reach beyond $r \sim r_0$ disappear from the trap.
610: The parameters $r_0$ and $\lambda$ are introduced for purely computational
611: purposes and should not be confused with any realistic experimental settings.
612: As $t \gg T$, the condensates in SFSS and NS disappear from the trap and the
613: particle number reaches its equilibrium value.
614: The bottom row of Figs. 4, 6, and 8 shows the wavefunctions after
615: the equilibrium is attained.
616:
617: Figure 4 shows the behavior of the condensate when $T=10 \tau$.
618: In this case,
619: $B_z(t)$ is reversed so fast that most of the particles
620: remain in the WFSS at $t =T/2-\tau$ and they are
621: suddenly converted into the SFSS at
622: $t=T/2+\tau$. These particles in the SFSS and the NS are eventually
623: lost from the trap as $t \to \infty$, see Fig. 5.
624:
625: Figures 6 and 8 show that the behavior of the condensate is qualitatively
626: similar for $T=100 \tau$ and $T=1000 \tau$. The condensate is transferred
627: from the WFSS to the SFSS more efficiently at $t=T/2 -\tau$ for $T=1000 \tau$
628: which leads to more WFSS components at $t=T/2 +\tau$ in this case.
629: At a later time $t > T$,
630: the condensate is found to oscillate in the trap with the
631: frequency $\sim \omega$.
632:
633: We have also analyzed the case $T=10000 \tau$ and found no qualitative
634: difference compared to the case $T=1000 \tau$.
635: There are slightly more of the NS component in the former case
636: at $t=T/2 -\tau$ than the latter case, which leads
637: to less particles in its equilibrium state at $t\gg T$ .
638:
639: It is certainly desirable to have more particles remaining in the trap
640: when a vortex is created. Figure 10 shows the ratio of the final
641: particle number to the initial particle number as a function of
642: the reversing time $T$. Note that the final particle number is evaluated
643: when the equilibrium is reached.
644: It is found that a
645: considerable amount of the condensate $(\gtrsim 1/3)$
646: is left in the trap for a wide range of the reversing times
647: $10^2 \lesssim T/\tau \lesssim 10^4$.
648: \begin{figure}
649: \begin{center}
650: \includegraphics[width=8cm]{Nlogt.eps}
651: \end{center}
652: \caption{Ratio of the residual particle number to the initial
653: particle number as a function of $\log_{10}(T/\tau)$.
654: The crosses are the results of our numerical calculation while
655: the interpolating curve is introduced as a guide.}
656: \label{fig:10}
657: \end{figure}
658:
659: In summary, our numerical calculations indicate that
660: a large fraction of the BEC remains in the trap when $B_z$ is
661: reversed with proper choices of the reversing time $T$.
662: Although there remain SFSS and NS components at $t=T$,
663: they eventually disappear from the trap at $t \gg T$ when the equilibrium
664: is reached. The condensate is converted into a state with two units of
665: circulation. Thus we conclude that a
666: vortex can be created, even in the absence of an optical plug,
667: by simply reversing the axial field $B_z$.
668: We expect that this makes the experimental
669: realization of the present scenario much easier than that with an
670: optical plug [5, 6].
671:
672: \section{Conclusions and Discussions}
673:
674: We have analyzed a simple scenario of vortex formation in a spinor BEC.
675: An axial magnetic field is slowly reversed while the quadrupole field
676: has been kept fixed in a Ioffe-Pritchard trap, which results in a
677: formation of a vortex with the winding number $2$. The spinor
678: degrees of freedom have been fully utilized which renders our
679: scheme much simpler compared with other methods.
680:
681: Since the vortex thus created has a higher winding number, it is
682: metastable and eventually breaks up to two singly quantized vortices.
683: The lifetime of the metastable state is quite an interesting quantity
684: to evaluate since its magnitude, compared with the field-reversal time $T$
685: and the trap time, will be crucial to determine the ultimate
686: fate of the vortex.
687:
688: A similar analysis with an $F=2$ BEC is under progress and will be
689: published elsewhere.
690:
691: \begin{acknowledgments}
692: We would like to thank Kazushiga Machida and Tomoya Isoshima
693: for useful discussions.
694: One of the authors (MN) would like to thank partial support by
695: Grant-in-Aid from Ministry of Education, Culture, Sports, Science and
696: Technology, Japan (Project Nos. 11640361 and 13135215).
697: He also thanks Martti M. Salomaa for support and
698: warm hospitality in the
699: Materials Physics Laboratory at Helsinki University of Technology,
700: Finland.
701: \end{acknowledgments}
702:
703:
704: \appendix
705: \section{Vortex Formation and Berry Phase}
706: It is shown that the formation of a vortex in
707: our scenario is understood
708: from a slightly different viewpoint. Namely, we show that the phase appearing
709: in the end of the process may be identified with the Berry phase
710: associated with the adiabatic change of the magnetic field.
711:
712: Let a point on the unit sphere in Fig. 11 (a) denote the hyperfine
713: spin state with $F=1$. All the spins near the origin $r=0$ have
714: $F_z \simeq -1$ at $t=0$
715: and hence correspond to the points near the south pole. Now the axial
716: field $B_z$ monotonically decreases such that eventually all the
717: spins near the origin have $F_z \simeq +1$ at $t=T$ and hence are expressed by
718: the points near the north pole. The path a spin follows on the unit
719: sphere depends on the position of the spin relative to the origin.
720: Figure 11 (b) shows the configuration of four
721: hyperfine spins $\bv{F}$ at $t=T/2$, when
722: $B_z$ vanishes and $\bv{F}$ is determined by the quadrupole field.
723: The path $C_1$ in Fig. 11 (a) shows the trajectory the spin 1 in Fig. 11 (b)
724: follows when $t$ is varied from $0\ (\beta =0)$ to $T\ (\beta=\pi)$. Similarly
725: the path $C_2$ in Fig. 11 (a) is the trajectory of the spin 2 in Fig. 11 (b),
726: and so on. When the spins 1 and 2
727: left the south pole at $t=0$, they had the common phase
728: factor, while at $t=T$ they obtain the relative phase equal to the
729: solid angle subtended by the trajectories $C_1$ and $C_2$.
730: The shaded area in Fig. 11 (a) shows this area which subtends
731: the solid angle $\pi$. Similarly the paths 1 and 3 (1 and 4) subtend the
732: solid angle $2 \pi$ ($3 \pi$), resulting in the relative phase $2\pi
733: (3\pi)$ between the spins 1 and 3 (1 and 4), respectively,
734: when $B_z$ is completely reversed. Accordingly, as one completes
735: a loop surrounding the origin, one measures the phase change of $4\pi$
736: observing that a vortex formation has taken place.
737: \begin{figure}
738: \begin{center}
739: \includegraphics[width=8cm]{berry.eps}
740: \end{center}
741: \caption{Formation of a vortex in the present scenario may be
742: understood in terms of the Berry phase associated with each
743: spin. When $|B_z| \gg |B_{\perp}|$ at $t=0$, all the spins near the
744: origin have $F_z \simeq -1$ with the same phase. Due to the ``twisting''
745: of the spins during the evolution, they obtain different phases
746: depending on the trajectories they follow. (b) shows four spins
747: at $t=T/2$ where $B_z$ vanishes. The spin 1 follows the trajectory
748: denoted by $C_1$ in (a) starting from the south pole and ending
749: at the north pole. Similarly the spins 2, 3 and 4 in (b) follows
750: the trajectories $C_2, C_3$ and $C_4$ in (a), respectively.
751: The shaded area $(\pi/2 \leq \phi\leq \pi, 0 \leq \beta \leq \pi)$
752: in (a) is the solid angle subtended by the paths $C_1$ and $C_2$.
753: Since this area measures $\pi$, the spin $2$ has a phase
754: $\pi$ relative to that of the spin $1$ at $t=T$.
755: Similarly the spins 3 and 4 obtain phases $2\pi$ and $3\pi$, respectively,
756: relative to that of the spin 1 at $t=T$.}
757: \label{fig:9}
758: \end{figure}
759:
760:
761:
762: \begin{thebibliography}{99}
763: \bibitem{rf:1} M.~Inguscio, S.~Stringari, and C. E.~Wieman (eds.),
764: {\it Bose-Einstein Condensation in Atomic Gases}, (IOS Press, Amsterdam,
765: 1999).
766: \bibitem{rf:2} S.~Martellucci,
767: A. N.~Chester, A.~Aspect, and M.~Inguscio (eds.),
768: {\it Bose-Einstein Condensates and Atomic Lasers}, (Kluwer Academic/
769: Plenum Publishers, New York, 2000).
770: \bibitem{rf:3} T.~Ohmi and K.~Machida, J. Phys. Soc. Jpn. {\bf 67}, 1822
771: (1998).
772: \bibitem{rf:4} T.-L.~Ho, Phys. Rev. Lett. {\bf 81}, 742 (1998).
773: \bibitem{rf:5} M.~Nakahara, T.~Isoshima, K.~Machida, S.-I.~Ogawa and
774: T.~Ohmi, Physica B {\bf 284-288}, 17 (2000).
775: \bibitem{rf:6} T.~Isoshima, M.~Nakahara, T.~Ohmi, and K.~Machida, Phys.
776: Rev. A {\bf 61}, 063610 (2000).
777: \bibitem{rf:7} M. R.~Matthews, B. P.~Anderson, P. C.~Haljan, D. S.~Hall,
778: C. E.~Wieman, and E. A.~Cornell, Phys. Rev. Lett. {\bf 83}, 2498 (1999).
779: \bibitem{rf:8} K. W.~Madison, F.~Chevy, W. Wohlleben, and J.~Dalibard,
780: Phys. Rev. Lett. {\bf 84}, 806 (2000).
781: \bibitem{rf:9} J. R.~Abo-Shaeer, C.~Raman, J. M.~Vogels, and W.~Ketterle,
782: Science {\bf 292}, 476 (2001).
783: \bibitem{rf:10} E.~Hodby, G.~Hechenblaikner, S. A.~Hopkins, O. M.~Marago,
784: and C.J.~Foot, Phys. Rev. Lett. {\bf 88} 010405 (2001).
785: \bibitem{rf:11} P. C.~Haljan, I.~Coddington, P.~Engels, and E. A.~Cornell,
786: Phys. Rev. Lett. {\bf 87}, 210403 (2001).
787: \bibitem{rf:12} D. M.~Stamper-Kurn and W.~Ketterle: in
788: R.~Kaiser, C.~Westbrook, F.~David (eds.),
789: {\it Coherent atomic matter waves - Ondes
790: de matiere coherentes}, (Springer, Heidelberg, 2001).
791: \bibitem{rf:13} K.~Maki and T.~Tsuneto: J. Low Temp. Phys.
792: {\bf 27}, 635 (1977).
793: \bibitem{rf:14} P. W. Anderson and G. Toulouse, Phys. Rev. Lett. {\bf 38}, 508 (1977).
794: \bibitem{ref:15} N. D. Mermin and T.-L. Ho, Phys. Rev. Lett. {\bf 36}, 594 (1976).
795:
796:
797: %\bibitem{rf:8} The e-mail address is jpsj-online@jps.or.jp.
798: \end{thebibliography}
799:
800:
801:
802: %\newpage %Just because of unusual number of tables stacked at end
803: %\bibliography{apssamp}% Produces the bibliography via BibTeX.
804:
805: \end{document}
806: %
807: % ****** End of file apssamp.tex ******
808: