1: \documentclass[twocolumn,pra]{revtex4}% Physical Review A
2: %\documentclass[onecolumn,pra]{revtex4}% Physical Review A
3: \usepackage{graphicx}% Include figure files
4: \usepackage{dcolumn}% Align table columns on decimal point
5: \usepackage{bm}% bold math
6: \begin{document}
7:
8: \preprint{APS/123-QED}
9:
10: \title{Probing dipolar effects with condensate shape oscillation}
11: \author{S. Yi and L. You}
12: \affiliation{School of Physics, Georgia Institute of Technology,
13: Atlanta, GA 30332-0430}
14:
15: \date{\today}
16:
17:
18: \begin{abstract}
19: We discuss the low energy shape oscillations of a magnetic trapped
20: atomic condensate including the spin dipole interaction.
21: When the nominal isotropic s-wave interaction strength becomes
22: tunable through a Feshbach resonance (e.g. as for $^{85}$Rb atoms),
23: anisotropic dipolar effects are shown to be detectable
24: under current experimental conditions
25: [E. A. Donley {\it et al.}, Nature {\bf 412}, 295 (2001)].
26: \end{abstract}
27:
28: \pacs{03.75.Fi, 05.30.-d, 32.80.Pj}
29: \maketitle
30: %\narrowtext
31:
32: Collective excitations play an important role in probing
33: microscopic interactions \cite{ph}.
34: The recently available gaseous atomic Bose-Einstein
35: condensates (BEC), have proven to be a profitable
36: testing ground for such studies \cite{castin}.
37: Atomic BEC are dilute with properties completely determined
38: by binary interactions.
39: Experimentally they are created at very low temperatures
40: when the short range atom-atom (collision) interaction can be
41: described by a single parameter: the s-wave scattering
42: length $a_{\rm sc}$. All higher order partial wave
43: collisions are suppressed at zero collision energy in
44: a short ranged potential. This enable atoms to be
45: modelled as hard spheres of radius $a_{\rm sc}$,
46: reflecting the isotropic ground state interaction.
47: Inside an electric or magnetic field, however,
48: ground state atoms may be polarized, e.g.
49: in a magnetic trap the direction of atomic spins
50: (of the valance electron for alkali) becomes aligned.
51: The resulting dipole interaction between condensed
52: atom pairs many not simply average out.
53: In this article, we investigate such spin dipole
54: effects on shape oscillation frequencies of an atomic condensate.
55: Although dipolar effects are typically small compared to
56: the dominant s-wave contact interaction, our study
57: indicates that these shifts
58: become observable in the $^{85}$Rb BEC \cite{e2,e3},
59: when a Feshbach resonance is used to tune
60: $a_{\rm sc}$ near zero \cite{e2}.
61:
62: Dipolar interaction in atomic BEC leads to physics beyond the
63: usual s-wave contact term, mainly because of the
64: modified low energy collision threshold behavior due to
65: the anisotropic nature of the interaction \cite{mircea}.
66: For magnetic spin dipoles, the net effect in the dilute
67: gas sample is rather small, one can therefor approximate the
68: complete two-body interaction by \cite{yi1,goral}
69: \begin{eqnarray}
70: V(\vec R)=g_0\delta(\vec R)+g_2{1-3\cos^2\theta_R\over R^3}
71: \end{eqnarray}
72: where $\vec R=\vec r-\vec r\,'$, and $g_0=4\pi\hbar^2a_{\rm sc}/M$
73: is the s-wave contact term. More generally, the dipole strength
74: $g_2$ equals $\alpha^2(0){\cal E}^2$ ($\alpha(0)$ atomic polarizability)
75: or $\mu^2$ ($\mu$ magnetic dipole moment) respectively for electric
76: or magnetic dipoles.
77: Figure \ref{fig1} illustrates the geometry for two
78: aligned dipoles along the local magnetic
79: field (z-axis). Several interesting properties of a dipolar
80: condensate have already been discussed \cite{yi1,goral,santos,yi2}.
81: In this paper, we focus on the shape oscillations
82: of a trapped dipolar condensate assuming a tunable
83: $a_{\rm sc}$.
84: \begin{figure}
85: \includegraphics[width=1.5in]{fig1.eps}
86: \caption{Geometry for the interaction of two aligned dipoles.}
87: \label{fig1}
88: \end{figure}
89:
90: In the standard approach, low energy collective excitations
91: are described by the Bogoliubov theory \cite{stringari}.
92: The condensate wavefunction $\phi(\vec r,t)$ is governed
93: by the Gross-Pitaevskii equation, while the non-condensed atoms
94: are described by quasi-particles \cite{keith}, some of
95: which have been studied experimentally \cite{jila,mit}.
96: The inclusion of the non-local dipolar interaction makes
97: the Bogoliubov approach impractical to implement numerically.
98: We therefore will rely on two alternative methods
99: to study the three characteristic shape modes \cite{jila}
100: as graphically illustrated for a cylindrical
101: symmetric trap in Fig. \ref{fig2}.
102: \begin{figure}
103: \includegraphics[width=2.25in]{fig2.eps}
104: \caption{Collective excitations of a
105: condensate with cylindrical symmetry.
106: Mode a is purely radial due to the cylindrical symmetry.
107: It's angular momentum projection along the z-axis $m=2$. Mode b
108: and c are respectively the quadrupole and monopole oscillations.
109: They are customarily called the low and high $m=0$ modes.}
110: \label{fig2}
111: \end{figure}
112:
113: First we adopt the highly successful time-dependent variation
114: approach used in Ref. \cite{zoller} by assuming a Gaussian ansatz
115: \begin{eqnarray}
116: \phi(x,y,z,t)=A(t)\prod_{\eta=x,y,z}e^{-\eta^2/2q^2_\eta+i\eta^2\beta_\eta(t)}.
117: \label{gausan}
118: \end{eqnarray}
119: For an harmonic trap
120: $V_t=M\sum_{\eta=x,y,z}\nu_\eta^2\eta^2/2$
121: ($\nu_\eta=\omega\lambda_\eta$),
122: the equations for variational parameters $q_\eta$
123: are equivalent to the classical motion of a particle
124: (with coordinate $q_\eta$) inside an effective potential
125: \begin{eqnarray}
126: &&U(q_x,q_y,q_z)\nonumber\\
127: &=&\sum_{\eta=x,y,z}\left({\hbar^2\over 2M q^2_\eta}
128: +{M\omega^2\over 2}\lambda^2_\eta q^2_\eta\right)
129: +{Ng_0\over (2\pi)^{3/2}q_xq_yq_z}\nonumber\\
130: &+&N{g_2\over (2\pi)^{3/2}}{1\over q_xq_yq_z}\int d{\vec r}
131: \,{1-3{\cos}^2\theta\over r^3}\,e^{-\sum_\eta{\eta^2/2q^2_\eta}},
132: \label{eqv}
133: \end{eqnarray}
134: where $N$ is the number of condensate atoms.
135: The equilibrium location ($q_\eta^0$) of Eq. (\ref{eqv})
136: then determines condensate size, while linearized shape
137: oscillation frequencies are determined by the second order
138: derivative $U_{\eta\eta'}(q_x,q_y,q_z)$ evaluated at $q_\eta^{0}$,
139: For a cylindrically symmetric trap ($\lambda_x=\lambda_y=1$),
140: we take $q_x=q_y=q_r$ and $\lambda_z=\lambda$.
141: The last integral in (\ref{eqv}) as well as
142: $U_{\eta\eta'}$ all become analytically computable \cite{yi2}.
143: The resulting $U_{\eta\eta'}$ matrix is symmetric
144: ($U_{\eta\eta'}=U_{\eta'\eta'}$), diagonalization of which gives
145: the three mode frequencies $\nu_a=\sqrt{U_{11}-U_{12}}$ and
146: \begin{eqnarray}
147: \nu_{b,c} ={1\over
148: \sqrt 2}\Big [&&U_{11}+U_{12}+U_{33}\nonumber\\
149: &&\pm\sqrt{(U^2_{11}+U^2_{12}-U^2_{33})^2+8U^2_{13}} \Big]^{1/2}.
150: \end{eqnarray}
151: \begin{figure}
152: \includegraphics[width=2.75in]{fig3.eps}
153: \caption{Mode frequencies for $N=10^4$ and $\lambda=1$. Solid
154: (dashed) lines are results excluding (including) dipole
155: interactions. $a_0$ stands for Bohr radius.} \label{fig3}
156: \end{figure}
157:
158: In the JILA experiments \cite{e2,e3} with $^{85}$Rb in the
159: $|F=2,M_F=2\rangle$ state, the valance electron spin gives rise to
160: an aligned magnetic dipole moment of $\mu=2\mu_B/3$ ($\mu_B$ being
161: the Bohr magneton). We take the radial frequency to be
162: $\nu_r=17.35$ (Hz), in addition to a tunable $a_{\rm sc}$
163: and trap aspect ratio $\lambda$ as in the experiment \cite{e3}.
164: Our results confirm that spin dipole effects become detectable
165: in terms of shifts of the shape oscillation frequencies.
166: We first report variational calculated results as the
167: analytic formulae obtained allow for a careful analysis of the
168: underline physics.
169: \begin{figure}
170: \includegraphics[width=2.75in]{fig4.eps}
171: \caption{The $\lambda$ dependence of the critical
172: scattering length $a_{\rm sc}^C$ for $N=10^4$.
173: The overall trend agrees with the variational result for
174: $a_{\rm eff}$ (in dashed line) derived earlier in Ref. [8].} \label{fig4}
175: \end{figure}
176:
177: Figure \ref{fig3} shows the three mode frequencies for
178: $N=10^4$ in a spherical trap ($\lambda=1$). The solid and
179: dashed lines denote respectively mode frequencies
180: without and with the dipolar interaction.
181: The vertical lines show the critical values of $a_{\rm sc}^C$
182: when the mode character switching
183: $b\leftrightarrow c$ occurs. This mode
184: switching occurs whenever the overall mean-field
185: condensate interaction changes from
186: repulsive to attractive or {\it vice versa}.
187: When squeezed along the radial direction,
188: mode c (b) became predominantly excited for an overall
189: attractive (repulsive) condensate as atoms are
190: pulled in (pushed out) along the z-direction.
191: Without the spin dipole interaction, this mode switch
192: always occurs at $a_{\rm sc}=0$. The anisotropic
193: dipole interaction affects the overall condensate
194: mean field and the stability depending on the trap aspect ratio
195: $\lambda$ \cite{yi1,santos}. Depending on
196: the configuration of the dipoles, dipole-dipole interactions
197: can be either attractive or repulsive. For two dipoles,
198: if they were placed in a plane perpendicular to their
199: polarization ($\uparrow\uparrow$), they repel each other.
200: On the other hand, they attract each other if
201: they were placed along the direction of their polarization ($\rightarrow\rightarrow$).
202: For a cloud of trapped dipoles one therefore expects that repulsive
203: interaction increases as one increases $\lambda$, which leads to
204: increased condensate stability.
205: In Fig. \ref{fig4}, we display the $\lambda$ dependence
206: of the critical scattering length $a_{\rm sc}^C$ as obtained
207: from the variational calculation.
208: We find that the effective scattering
209: length $a_{\rm eff}$ (introduced earlier by us in \cite{yi1})
210: provides a reasonable approximation to the sign of
211: the overall condensate mean field. In terms of
212: the actual values, however, the results in the figure show that
213: $a_{\rm eff}$ differs significantly from $a_{\rm sc}^C$
214: when $\lambda$ deviates from the neighborhood of unity.
215:
216: In Fig. \ref{fig5}, we show the $\lambda$-dependence of
217: the dipole induced fractional changes to the mode frequencies
218: for a condensate with $10^4$ atoms at $a_{\rm sc}=-1\,(a_0)$.
219: For strongly prolate or oblate traps, the shifts
220: are in a few percentage range even with such small
221: numbers of atom.
222: Figure \ref{fig6} summarizes our results for
223: the atom number dependence of the dipole induced frequency
224: shifts for the trap aspect ratio $\lambda=6.8/17.35$ \cite{e3}.
225: It was shown earlier \cite{yi2},
226: at increased values of $N$, a dipolar condensate always collapses
227: irrespective of the sign of $a_{\rm sc}$. The Gaussian ansatz
228: (\ref{gausan}) becomes questionable near collapse as reflected
229: in the seemly divergent results when N is increased.
230:
231: \begin{figure}
232: \includegraphics[width=2.75in]{fig5.eps}
233: \caption{Fractional change (\%) of mode frequencies for $a_{\rm
234: sc}=-1\,(a_0)$ and $N=10^4$. Similar results are obtained for
235: $a_{\rm sc}=0$ and $1\,(a_0)$.} \label{fig5}
236: \end{figure}
237:
238: \begin{figure}
239: \includegraphics[width=2.75in]{fig6.eps}
240: \caption{Atom number dependence of the dipole interaction induced
241: shape oscillation frequency shifts.
242: Mode a, b, and c are labelled respective with solid, dashed,
243: and dot-dashed lines.} \label{fig6}
244: \end{figure}
245:
246: These results (Figs. \ref{fig3}-\ref{fig6})
247: clearly show that dipolar effects are detectable
248: in the current $^{85}$Rb BEC setup, given the extraordinary
249: capability of $0.1\%$ frequency measurement \cite{pri}.
250: To confirm the validity of the variational calculations,
251: we have invested considerable effort in
252: an exact numerical method based on the time dependent
253: Gross-Pitaevskii equation \cite{keith2}.
254: By applying external periodic forcing terms
255: described by the potential
256: $$
257: V_F(\vec r)=\sum_{\eta=x,y,z} V_\eta\left\{
258: e^{-{[\eta+\Delta_\eta(t)]^2\over 2w^2}}+
259: e^{-{[\eta-\Delta_\eta(t)]^2\over 2w^2}}\right\},
260: $$
261: %(see Fig. \ref{fig7})
262: with
263: $\Delta_\eta(t)=\eta_0+\delta_\eta\sin(\Omega_\eta t+\phi_\eta)$
264: to the condensate ground state, selected shape oscillations
265: become predominantly excited, as first numerically
266: implemented by Ruprecht {\it et al.} \cite{keith2}.
267: %\begin{figure}
268: %\includegraphics[width=2.in]{fig7.eps}
269: %\caption{Deformed (bare) trap potentials.} \label{fig7}
270: %\end{figure}
271:
272: After a selected duration $T$, typically several periods of
273: the trap radial oscillation,
274: the forcing term $V_F$ is tuned off. The free
275: propagation of the time dependent Gross-Pitaevskii equation is
276: continued, and the dynamic condensate width $\sqrt{\langle
277: \eta^2(t)\rangle}$ is sampled. The shape oscillation frequencies
278: are then identified by taking the Fourier transformation of
279: $\sqrt{\langle \eta^2(t)\rangle}$. A typical result from this
280: calculation is given in Fig. \ref{fig8}, which shows
281: remarkably clear signal. By varying $V_\eta$, $\eta_0$,
282: $\delta_\eta$, and $\Omega_\eta$, our results are
283: self-consistently checked, i.e. to be independent of all
284: parameters involved as they should be in the small amplitude
285: oscillation limit and to be numerically accurate.
286: \begin{figure}
287: \includegraphics[width=2.75in,height=3.5in]{fig8.eps}
288: \caption{Typical numerical results for condensate width and its
289: corresponding Fouier transformed signal. Mode $b$ and $c$ are
290: clearly identified. The parameters are $a_{\rm sc}=0$,
291: $\lambda=1$, and $N=40000$.} \label{fig8}
292: \end{figure}
293: For the experimental parameter of $\nu_r=17.35$ (Hz) and
294: $\nu_z=6.8$ (Hz). We have computed the shape oscillation
295: frequencies numerically for $a_{\rm sc}=0,1,-2$ ($a_0$), and
296: compared them with the variational results in Table
297: \ref{t1}. The quoted errors for the numerical results
298: shown in the tables mainly come
299: from the limited sampling window in time. Further improvement
300: is hard as the calculations are time consuming and
301: numerical accuracies become difficult to control at long times.
302: \begin{table}
303: \caption{Mode frequencies (in unit of $\nu_r$) with dipolar interaction}
304: \begin{tabular}{cccc}
305: $a_{\rm sc}$($a_0$) &mode &numerical &variational\\ \hline
306: $-2$ & $\begin{array}{c}b\\c\end{array}$
307: &$\begin{array}{c}2.0010\pm 0.0040\\0.8578\pm 0.0085\end{array}$
308: &$\begin{array}{c}2.0016\\0.8657\end{array}$ \\ \hline
309: $0$ &$\begin{array}{c}b\\c\end{array}$ &$\begin{array}{c}1.9959\pm
310: 0.0028\\0.7874\pm 0.0022\end{array}$
311: &$\begin{array}{c}1.9964\\0.7895\end{array}$\\ \hline
312: $+1$ &$\begin{array}{c}b\\c\end{array}$ &$\begin{array}{c}0.7705\pm
313: 0.0060\\1.9974\pm 0.0032\end{array}$
314: &$\begin{array}{c}0.7675\\1.9974\end{array}$ \\ \hline
315: \end{tabular}
316: \label{t1}
317: \end{table}
318: We find that the variational results are
319: consistent with the exact numerical results. In order to affirm
320: such mode frequency shifts are indeed from the dipolar interaction,
321: rather than a mis-calibration of $a_{\rm sc}$, we also need to
322: validate the variational approach in the absence of the
323: dipole interaction. This was explored earlier in Ref. \cite{zoller},
324: where the numerical and variational results of the condensate
325: dynamics were compared. Our results are presented
326: in Table \ref{t2}. It convincingly proves that
327: dipolar efforts reported in the Table \ref{t1}
328: is due to the physics of spin dipole interaction.
329: \begin{table}
330: \caption{Mode frequencies (in unit of $\nu_r$) without dipolar interaction}
331: \begin{tabular}{cccc}
332: $a_{\rm sc}$($a_0$)&mode&numerical&variational \\ \hline $-2$
333: &$\begin{array}{c}b\\c\end{array}$ &$\begin{array}{c}2.0073\pm
334: 0.0025\\0.8546\pm 0.0051\end{array}$
335: &$\begin{array}{c}2.0075\\0.8591\end{array}$\\\hline $+1$
336: &$\begin{array}{c}b\\c\end{array}$ &$\begin{array}{c}0.7664\pm
337: 0.0054\\2.0004\pm 0.0024\end{array}$
338: &$\begin{array}{c}0.7622\\2.0004\end{array}$\\ \hline
339: \end{tabular}
340: \label{t2}
341: \end{table}
342:
343: In conclusion, we have studied low energy shape oscillations
344: of a trapped dipolar condensate. Using a magnetic field dependent
345: Feshbach resonance to tune the s-wave scattering length to around
346: zero, we have shown that the weak spin dipole interaction
347: becomes detectable as shifts to shape oscillation frequencies
348: under currently available experimental conditions \cite{e2,e3}.
349: These shifts grow with the number of trapped atoms, and
350: are around $1\%$ level with as few as $10^4$ atoms.
351: We have also independently verified the accuracy of the
352: variational calculation by developing a rigorous
353: numerical approach for the three predominant shape oscillation modes.
354: Near a Feshbach resonance, significant atom loss might occur
355: as found in the Na experiment \cite{mit2}. The subsequent damping
356: could lead to a broadening of the shape oscillation thus masking
357: the direct observation of the proposed spin dipole effects.
358: Fortunately for $^{85}$Rb atoms, the condition of $a_{\rm sc}(B)=0$
359: corresponds to far off resonance on the high B-field side, where
360: impressive controls have been demonstrated without any significant
361: loss \cite{e3}.
362: Over the last few years, mean field theory has proven to be
363: remarkably successful when applied to BEC physics.
364: The s-wave $g_0$ contact pseudo-potential has made the
365: concept of scattering length $a_{\rm sc}$ widely popular.
366: It is often said that scattering length is the only relevant
367: atomic parameter since the net interaction effect scales
368: as $Na_{\rm sc}/a_{\rm ho}$ for a harmonically trapped atomic
369: condensate, where $a_{\rm ho}$ is the trap size.
370: Dipole interactions as discussed in this article
371: points to interesting physics beyond the s-wave.
372: Within the current experimental regime of
373: dilute atomic gases, mean field theory remains applicable
374: and allows for the calculation of dipolar induced
375: shifts of collective excitation frequencies.
376: Successful experimental verification of our predictions
377: will shed new light on atomic BEC.
378:
379: We thank Drs. C. E. Wieman and E. Cornell for helpful discussions.
380: This work is supported by the NSF grant No. PHY-9722410.
381:
382: \begin{references}
383:
384: \bibitem{ph}{\it Theory of Interacting Fermi Systems},
385: Ph. Norzi\'eres, (Addison-Wesley, Reading, Massachusetts, 1997).
386:
387: \bibitem{castin}Y. Castin, in {\it Coherent atomic matter waves},
388: Lecture Notes of Les Houches Summer School, p.1-136,
389: edited by R. Kaiser, C. Westbrook, and F. David,
390: (EDP Sciences and Springer-Verlag, 2001).
391:
392: %\bibitem{gibble}R. Legere and K. Gibble, Phys. Rev. Lett. {\bf 81}, 5780 (1998).
393:
394: %\bibitem{jin}B. DeMarco, J. L. Bohn, J. P. Burke, Jr.,
395: %M. Holland, and D. S. Jin, Phys. Rev. Lett. {\bf 82}, 4208 (1999).
396:
397: \bibitem{e2}S. L. Cornish {\it et al.},
398: Phys. Rev. Lett. {\bf 85}, 1795 (2000).
399:
400: \bibitem{e3}E. A. Donley, N. R. Claussen, S. L. Cornish, J. L.
401: Roberts, E. A. Cornell, and C. E. Wieman, Nature {\bf 412}, 295
402: (2001).
403:
404: \bibitem{mircea}M. Marinescu and L. You, Phys. Rev. Lett. {\bf 81}, 4596
405: (1998).
406:
407: \bibitem{yi1}S. Yi and L. You, Phys. Rev. A {\bf 61}, 041604(R)
408: (2000).
409:
410: \bibitem{goral}K. Goral, K. Rzazewski, and Tilman Pfau,
411: Phys. Rev. A {\bf 61}, 051601(R) (2000);
412: J.-P. Martikainen, Matt Mackie, and K.-A. Suominen,
413: Phys. Rev. A {\bf 64}, 037601 (2001).
414:
415: \bibitem{santos} L. Santos, G. V. Shlyapnikov, P. Zoller, and M.
416: Lewenstein, Phys. Rev. Lett. {\bf 85}, 1791 (2000).
417:
418: \bibitem{yi2}S. Yi and L. You, Phys. Rev. A {\bf 63}, 053607 (2001).
419:
420: \bibitem{stringari}
421: F. Dalfovo, S. Giorgini, L. P. Pitaevskii, and S. Stringari,
422: Rev. Mod. Phys. {\bf 71}, 463 (1999).
423:
424: \bibitem{keith}P. A. Ruprecht {\it et al.},
425: Phys. Rev. A{\bf 51}, 4704 (1995).
426:
427: \bibitem{jila}D. S. Jin {\it et al.},
428: Phys. Rev. Lett. {\bf 77}, 420 (1996).
429:
430: \bibitem{mit}M. O. Mewes {\it et al.}, Phys. Rev. Lett. {\bf 77}, 988 (1996).
431:
432: \bibitem{zoller}V. M. Perez-Garcia {\it et al.},
433: Phys. Rev. Lett. {\bf 77}, 5320 (1996).
434:
435: \bibitem{stoof}M. J. Bijlsma and H. T. C. Stoof,
436: Phys. Rev. A {\bf 60}, 3973 (1999).
437:
438: \bibitem{pri}C. E. Wieman and E. Cornell, (private
439: communications).
440:
441: \bibitem{keith2}P. A. Ruprecht, M. Edwards, K. Burnett, and C. W. Clark,
442: Phys. Rev. A {\bf 54}, 4178 (1996).
443:
444: \bibitem{mit2}S. Inouye, M. R. Andrews, J. Stenger, H.-J. Miesner, D. M.
445: Stamper-Kurn, and W. Ketterle, Nature (London) {\bf 392}, 151
446: (1998).
447:
448: \end{references}
449:
450: \end{document}
451: