1: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2:
3: \usepackage{graphicx}% Include figure files
4: \usepackage{dcolumn}% Align table columns on decimal point
5: \usepackage{bm}% bold math
6:
7:
8: \newcommand\foreign[1]{{\it #1\spacefactor=1000}}
9: \newcommand\eg{\foreign{e.g.}}
10: \newcommand\ie{\foreign{i.e.}}
11: \newcommand\ibid{\foreign{ibid}}
12:
13: \newcommand\Denv{{D_{\mbox{\scriptsize env}}}}
14: \newcommand\Dcl {{D_{\mbox{\scriptsize cl }}}}
15:
16: \begin{document}
17:
18: \title{An Atom Laser is not monochromatic}
19: \author{S. Choi$^{1}$, D. Str\"{o}mberg$^{1,2}$,
20: and B. Sundaram$^{1}$} \affiliation{$^{1.}$ Department of Physics,
21: University of Massachusetts, Boston, MA 02125, USA \\ $^{2.}$
22: Department of Information Technology, Uppsala University, 751 05
23: Uppsala, Sweden}
24:
25:
26: \begin{abstract}
27: We study both numerically and analytically the possibility of using
28: an adiabatic passage control method to construct a Mach-Zehnder
29: interferometer (MZI) for Bose-Einstein condensates (BECs) in the
30: time domain, in exact one-to-one correspondence with the
31: traditional optical MZI that involves two beam splitters and two
32: mirrors. The interference fringes one obtains from such a
33: minimum-disturbance set up clearly demonstrates that, fundamentally,
34: an atom laser is not monochromatic due to interatomic interactions.
35: We also consider how the amount of entanglement in the system
36: correlates to the interference fringes.
37: \end{abstract}
38:
39: \pacs{03.75.Dg,03.75.Kk,03.75.Lm}
40:
41: \maketitle
42:
43:
44: Atomic Bose-Einstein condensates (BECs) are often referred to as the
45: ``atom laser,'' the matter-wave equivalent of a laser. Unlike
46: massless photons from a laser which do not interact, atoms from an
47: atom laser do interact with each other, and so far it has not been
48: clearly established how and to what extent the monochromaticity of
49: an atom laser is compromised by such interactions. The term ``atom
50: laser'' is not yet defined universally other than in a broad general
51: sense, although there have been attempts to be more
52: specific about what an atom laser should be, including the
53: directionality of the beam in a close analogy with a
54: laser~\cite{Wiseman}. However, given fundamental
55: differences such as the fact that
56: the atoms in a BEC are intrinsically stationary, here we shall use
57: the term to denote an {\it intense} source of {\it
58: coherent} atoms {\it for atom optical transformations} such as
59: reflection, refraction and beam splitting.
60:
61: In general, the non-zero
62: mass of the atoms necessitates rather elaborate and experimentally
63: demanding schemes to coherently accelerate and guide the
64: matter-wave. In order to mitigate this problem, we propose the use
65: of atom optics in the {\it time-domain} based on adiabatic passage.
66: An important reason for this is that time domain atom optics is less
67: prone to decoherence due to, say, atom losses as compared to
68: techniques such as guiding a BEC in space. Further, the
69: adiabatic passage method is a well-known technique in quantum
70: mechanics that provides a way to ``delicately'' control a quantum
71: system via slow passage along the energy landscape, while preserving
72: the quantum system in one of the energy eigenstates, typically the
73: ground state. By minimizing any disturbance, and maintaining the
74: condensate in the ground state, one can draw conclusions about the
75: intrinsic properties of an atom laser unmodified by such additional
76: actions as ballistic expansion.
77:
78: \begin{figure}
79: \begin{center}
80: \centerline{\includegraphics[height=10.5cm]{MZI_fig1}}
81: \caption{(a) Schematic diagram of the
82: transformations that the potential undergoes over time, starting
83: with beam splitting, addition of a potential step, recombination and
84: final beam splitting. (b) Spatio-temporal probability density obtained by solving the
85: Gross-Pitaevkii equation describing the condensate passing through
86: the time-domain MZI.
87: } \label{fig1}
88: \end{center}
89: \vspace{-0.5in}
90: \end{figure}
91:
92: To address the question of the monochromaticity of an atom laser, we
93: model both numerically and analytically an ideal Mach-Zehnder
94: Interferometer (MZI) for BECs in the time domain and deduce the
95: spectral composition of an atom laser from the interference pattern
96: generated. The construction of a MZI that uses BEC as the source of
97: coherent atoms is one of the goals in atom optics owing to
98: the promise of, for instance, high precision matter-wave based
99: metrology. In a typical optical MZI scheme, coherent light from a
100: laser is passed through a beam splitter which sends the light beam
101: into two paths; one of the beams passes through a phase shifter and
102: the two beams are then recombined by reflecting off mirrors onto a
103: second beam splitter. The phase shift experienced in one arm is
104: reflected in the intensity variation at the two detectors. We have
105: simulated the time-domain MZI by adding an optical potential to the
106: usual quadratic trapping potential and adiabatically varying its
107: amplitude to simulate exactly the action of the optical elements in
108: an usual MZI, namely the two beam splitters, mirrors and wave
109: guides. The phase shift is provided by adding an additional
110: potential step to one ``arm'' of the interferometer. Given the
111: adiabatic nature of the method, the two arms do not interfere unless
112: they are explicitly recombined by way of the second beam splitter.
113: The population difference $\Delta P$ between the two arms is finally
114: measured as a function of the step height. A schematic diagram of
115: the required operations is provided in Fig. \ref{fig1}(a). To the
116: best of our knowledge, such a scheme has not yet been experimentally
117: implemented, although there have been several experiments on
118: interference of BEC in optical potentials~\cite{DoubleWell} and in
119: atom chips~\cite{AtomChip,AtomChip2, AtomChip3}, with the important
120: difference that all of these release the condensates to expand
121: ballistically after the first beam splitter. The time-domain MZI
122: proposed here keeps the condensate in the ground state, better
123: revealing the nature of the atom laser.
124:
125:
126: \begin{figure}
127: \begin{center}
128: \centerline{\includegraphics[height=6cm]{MZI_fig2}}
129: %\centerline{\includegraphics[height=6cm]{fringes}}
130: \caption{Interference fringes obtained by solving the
131: Gross-Pitaevskii Equation (GPE) to model the time-domain MZI for
132: various values of the nonlinearity constant $g$ {\bf (a)} $g = 0$
133: {\bf (b)} $g = 0.5$ {\bf (c)} $g = 5$ {\bf (d)} $g = 10$ The
134: sinusoidal fringe for the linear $g=0$ case is clearly modified for
135: $g
136: > 0$ i.e. in the presence of interatomic interactions, showing that the atom laser is
137: no longer monochromatic.} \label{fig2}
138: \end{center}
139: \vspace{-0.4in}
140: \end{figure}
141:
142:
143: The Gross-Pitaevskii Equation (GPE) was used to simulate the
144: dynamics of the condensate in the presence of the spatio-temporal
145: potential. The GPE has been extremely
146: successful in describing the dynamics of BECs under a variety of
147: experimental conditions including optical lattices\cite{Review}.
148: Specifically, we consider the 1-D GPE:
149: \begin{equation}
150: i\hbar \frac{\partial \psi}{\partial t} = \left [
151: -\frac{\hbar^2}{2m} \frac{d^2}{dx^2} + V(x,t) + g|\psi(x,t)|^2
152: \right ] \psi(x,t)
153: \end{equation}
154: where $\psi(x,t)$ denotes the condensate mean field and $V(x,t)$
155: denotes the adiabatically changing time-dependent potential. As
156: shown later in our analysis, the dimensionality of
157: the GPE simulation does not affect the final results. We always
158: retain the parabolic trapping potential, so $V(x,t) =
159: \frac{1}{2} m \omega_{t} x^2 + Af(t) \cos (k x) + V_{step}(x,
160: T_{1/4}<t<2T_{1/4})$ where $k = 2\pi/ \lambda$, $A$ is the amplitude
161: of the optical lattice potential
162: and $T_{1/4}$ denotes $1/4$ of the
163: total duration of the simulation $T$. $f(t)$ provides the time
164: dependence for the adiabatically changing amplitude of the optical
165: lattice:
166: \begin{equation}
167: f(t) = \left\{ \begin{array}{c l}
168: t/T_{1/4} & \mbox{ $0 < t < T_{1/4}$, first beam splitter} ;\\
169: 1 & \mbox{ $T_{1/4} < t < 2T_{1/4}$, first beam splitter} ; \\
170: (3 - t/T_{1/4}) & \mbox{ $2T_{1/4} < t < 3T_{1/4}$, recombination}; \\
171: (t/T_{1/4} - 3) & \mbox{ $3T_{1/4} < t < 4T_{1/4}$, final beam splitter}.\end{array}
172: \right.
173: \end{equation}
174: The potential step was introduced during $T_{1/4} < t < 2T_{1/4}$
175: using a super-Gaussian in one arm $V_{step}(x, t) = h e^{ - [(t -
176: T_{1/4})/\sigma_{t}]^{10}} e^{ - [(x -
177: \lambda/2)/\sigma_{x}]^{30}}$, with the spatial and temporal width
178: of $\sigma_{x} = \lambda/3$ and $\sigma_{t} = T_{1/4}/2$
179: respectively. The probability density of the condensate obtained
180: from GPE dynamics with $g = 5$, $A = 25$, $T_{1/4} = 400$, and $\lambda = 15$
181: in harmonic oscillator units of the initial trapping potential is
182: shown in Fig. \ref{fig1}(b) as a spatio-temporal plot. As is clear
183: from the figure, the simulation provides exact one-to-one
184: correspondence to the MZI, and the adiabatic condition ensures that the
185: wave function evolves extremely cleanly. One of the advantages of
186: this scheme of interferometry is that the area enclosed by the two
187: arms of the interferometer is readily controllable by changing
188: $\lambda$; in particular, with a larger enclosed area, higher
189: sensitivity is possible for use in, for instance, interferometric
190: rotational sensors used in space navigation.
191:
192: We plot in Fig. \ref{fig2} the
193: final intensity difference as a function of step height $h$ for
194: different values of nonlinearity $g$. It is noted that a relatively
195: large phase shift is recorded even for small variations in step
196: height, indicating high sensitivity of the condensate to changes in
197: potential height. In the linear case (non BEC) $g = 0$ one obtains
198: perfect sinusoidal variation as a function of $h$, while for $g >
199: 0$, the pattern clearly deviates from the sinusoidal, implying that
200: an atom laser loses monochromaticity due to the presence of
201: interatomic interactions. This is confirmed by a Fourier transform
202: which reveals four closely spaced frequency components.
203:
204: In order to better understand this result and to ensure that the
205: adiabatic assumptions are not violated, a detailed analysis of this
206: system can be given as follows. First, we assume that the adiabatic
207: passage we used above preserves the single mode approximation. The
208: condensate after beam splitting and the potential step (which
209: imparts relative phase of $\Theta$) may then be written as:
210: \begin{equation}
211: \hat{\psi}(x, t, \Theta) = \phi_{L}(x,d) \hat{a}_{L}(t) +
212: \phi_{R}(x,d) \exp(i \Theta) \hat{a}_{R} (t) \label{singlemode}
213: \end{equation}
214: where $\hat{a}_{L(R)}(t)$ are bosonic annihilation operators for the
215: modes of the matter wave field in the left and right arms and
216: $\phi_{L(R)}(x,d)$ are the corresponding spatial mode functions.
217: These may be approximated by spatial functions such as shifted
218: Gaussians $\phi_{L(R)}(x,d) = (\pi \sigma^2)^{-1/4} e^{-(x \pm
219: d)^2/4 \sigma^2}$ that, under adiabatic evolution, do not change
220: shape. The phase shift has been modeled by the explicit insertion of
221: the $\exp(i \Theta)$ term, which is clearly an approximation to the
222: full GPE simulation where the phase shift results from the passage
223: of the condensate over a potential step in one arm. The analysis is
224: therefore best suited to the cases with lower values of $g$ for
225: which interplay between the potential step and repulsive atomic
226: interactions do not complicate the phase relationship given in Eq.
227: (\ref{singlemode}).
228:
229: Since we use the adiabatic passage method, there are no additional
230: net physical effects on the BEC undergoing the remaining two operations that
231: constitute the MZI (recombination and then second beam splitting),
232: other than those of the interatomic scattering
233: and the natural tunnelling flow between the left and right arms. This
234: is due to the Josephson effect and the geometry of the system though
235: the tunnelling would be minimized when the arms are well separated.
236: The effective interference fringes are then given by the
237: difference in the number of atoms in the two arms after the system
238: has evolved a certain time $\tau$ through the MZI. With the field
239: decomposition of Eq. (\ref{singlemode}) the bosonic Hamiltonian is given
240: by
241: \begin{eqnarray}
242: \hat{H} & = & \hbar \omega (\hat{a}^{\dagger}_{L}\hat{a}_{L} +
243: \hat{a}^{\dagger}_{R}\hat{a}_{R}) + \frac{\Delta E(t)}{2}
244: [\hat{a}^{\dagger}_{L}\hat{a}_{R} +
245: \hat{a}^{\dagger}_{R}\hat{a}_{L}] \nonumber \\
246: & & + g(\hat{a}^{\dagger 2}_{L}\hat{a}^{2}_{L} + \hat{a}^{\dagger
247: 2}_{R}\hat{a}^{2}_{R})
248: \end{eqnarray}
249: where $\Delta E(t) = \hbar \omega \exp [-d^{2}(t)/\sigma^2]$ gives
250: the overlap of the two spatial modes and is the time-dependent
251: tunneling energy. This overlap clearly depends on the
252: dimensionality of the modes. However, change in dimensionality
253: merely alters this single scaling parameter and not the behavior we
254: describe. This Hamiltonian is also well-known as the Hamiltonian
255: that describes two-component BECs when ``L'' and ``R'' denote two
256: different species instead of the left and right arms. Introducing
257: the Schwinger angular
258: momentum operators, $\hat{J}_{+(-)} = \hat{a}_{L(R)}^{\dagger}\hat{a}_{R(L)}$ and $\hat{J}%
259: _z = \frac{1}{2} ( \hat{a}_{L}^{\dagger}\hat{a}_{L} -
260: \hat{a}_{R}^{\dagger}\hat{a}_{R})$, the Hamiltonian is equivalent to
261: a ``Lipkin-Meshkov-Glick'' type Hamiltonian~\cite{LMG},
262: \begin{eqnarray}
263: \hat{H} & = & \frac{\Delta E(t)}{2} \left ( \hat{J}_{+} e^{i
264: \Theta} + \hat{J}_{-} e^{-i \Theta} \right ) + 2g \hat{J}_{z}^2 .
265: \end{eqnarray}
266: The time evolution operator corresponding
267: to this nonlinear Hamiltonian has been studied
268: previously~\cite{ChoiEntangle}. The measured intensity difference
269: is given by $\langle J_{z} (\tau) \rangle$ at time $\tau$,
270: $\langle \hat{J}_{z} (\tau, \Theta) \rangle = \langle
271: \hat{R}^{\dagger}e^{iH^{\prime} \tau }\hat{R}
272: \hat{J}_{z} \hat{R}^{\dagger}e^{-iH^{\prime} \tau }\hat{R} \rangle$,
273: where $\hat{R} =
274: \exp \left [ -\frac{\pi}{4} (\hat{J}_{-} e^{-i \Theta} - \hat{J}_{+}
275: e^{i \Theta}) \right ]$ is the well-known rotation operator and $H'$
276: is the transformed Hamiltonian diagonal in $\hat{J}_{z}$:
277: $\hat{H}^{\prime} =\hat{R}\hat{H} \hat{R}^{\dagger} - i \hat{R} \frac{%
278: \partial}{\partial t} \hat{R}^{\dagger} \equiv \Delta E \hat{J}_{z} - g
279: \hat{J}_{z}^{2}$.
280:
281: Writing the quantum state in a general form $| \Psi \rangle =
282: \sum_{m} c_m |m \rangle_{L} | N- m \rangle_{R}$ where $N$ is the
283: total number of atoms in the system, one obtains,
284: \begin{eqnarray}
285: \langle \hat{J}_{z} (\tau, \Theta) \rangle
286: & = & - \frac{e^{i g\tau }}{2} \left ( \sum_{m'} \beta^{(+)}_{m'} \gamma^{(+)*}_{m'} \right ) e^{i (\Theta + \Delta \omega \tau)} \nonumber \\
287: & & - \frac{e^{i g\tau }}{2} \left (\sum_{m'} \beta^{(-)}_{m'}
288: \gamma^{(-)*}_{m'} \right ) e^{-i(\Theta + \Delta \omega \tau)} \nonumber
289: \end{eqnarray}
290: where $\Delta \omega = \Delta E /\hbar$ and
291: \begin{eqnarray}
292: \beta^{(\pm)}_{m'}
293: %& = & \langle \Psi | \hat{R}^{\dagger} | m' \rangle \sqrt{j(j + 1) - m'(m' \pm 1)} \nonumber \\
294: & \equiv & \sum_{m} c^{*}_{m} {\cal R}_{m,m'} \sqrt{j(j + 1) - m'(m' \pm 1)} \nonumber \\
295: \gamma^{(\pm)}_{m'}
296: %& = & \langle \Psi | \hat{R} \hat{J}_{-} | m \pm 1 \rangle e^{\mp i 2 m'
297: %g \tau } \nonumber \\
298: & \equiv & \sum_{m} c^{*}_{m} {\cal R}_{m,m' \pm 1} e^{\pm i 2 m'
299: g \tau }
300: \end{eqnarray}
301: with the matrix elements ${\cal R}_{m,m'} \equiv \langle m |
302: \hat{R}^{\dagger} | m' \rangle$ provided by an analytical
303: expression~\cite{bloch}.
304:
305:
306: \begin{figure}
307: \begin{center}
308: \centerline{\includegraphics[height=5cm]{MZI_fig3}} \caption{ The
309: interference fringe given by the fractional number difference in the
310: two arms of the interferometer, $\langle \hat{J}_{z} \rangle/N$,
311: obtained using the analytical methods under the adiabatic
312: assumption. The figure, which shows remarkable agreement with the
313: simulation results obtained above using the GPE, again clearly
314: indicates the presence of several frequency components in the atom
315: laser from the way it deviates from the sinusoidal pattern.}
316: \label{fig3}
317: \end{center}
318: \vspace{-0.4in}
319: \end{figure}
320:
321: A natural choice of quantum state for the two component BECs with a
322: relative phase $\Theta$ is a coherent spin state or the Bloch state
323: with equal number of atoms in both arms such that $|\Psi \rangle =
324: |\theta = \pi/2, \phi = \Theta \rangle$. In this case, the
325: coefficients $c_m = (C^{2j}_{j + m})^{1/2} e^{i(j-m)\Theta}/2^j$
326: where $C^{n}_{m}$ denotes the combination, $C^{n}_{m} =
327: n!/[(n-m)!m!]$. The resulting interference fringe with $(\Delta
328: \omega) \tau$ chosen to be $(2n + 2)\pi$ and $g\tau = (2n +
329: \frac{3}{2N})\pi$, where $n$ is an integer, is plotted in Fig.
330: \ref{fig3} which shows remarkable agreement with the result from the
331: GPE simulation. With $n \approx 127$, we are simulating the $g=0.5$
332: case in the GPE simulation above, and the effective magnitude of $g$
333: of the order $\frac{3}{2N}$ places the system in the Josephson
334: regime. The amplitudes are different from the GPE simulation simply
335: due to our choice of $\langle \hat{J}_{z} \rangle$ to be half the
336: atom number difference.
337:
338: The results confirm that the coherent spin state is a natural state
339: for BECs with relative phase arising from the adiabatic passage
340: method. With the current system, the initial state prepared at the
341: first beam splitter can be controlled so that the number of atoms in
342: each arm are not equally distributed. For instance, this can be accomplished by
343: adiabatically deforming the potential wells asymmetrically to collect
344: more atoms in one arm and then continue adiabatically to model
345: subsequent actions such as recombination. In such cases, a coherent spin state with
346: $\theta \neq \pi/2$ can be generated, and consequently the amount of
347: entanglement between the left and right arms is directly
348: controllable. The state with $\theta = \pi/2$ has the largest entanglement
349: while the state with $\theta = 0$ has the least amount of entanglement~\cite{ChoiEntangle}.
350:
351:
352: \begin{figure}
353: \begin{center}
354: \centerline{\includegraphics[height=6cm]{MZI_fig4}}
355: \caption{Entanglement parameterized by the von Neumann entropy of
356: the input state plotted against the visibility of the final
357: inference fringes. Inset: Visibility vs. fractional atom number
358: difference between the two arms $\Delta P$ of the input state.
359: Solid, dashed, and dotted lines: equivalent nonlinearity in GPE of
360: $g = 0$, $0.25$ and $0.5$ respectively in both the main figure and
361: the inset. } \label{entangle}
362: \end{center}
363: \vspace{-0.4in}
364: \end{figure}
365:
366: Finally, how the presence of interatomic interaction modifies the
367: connection between the fringe visibility and entanglement is of
368: interest from the point of possible experimental determination of
369: parameters such as the interatomic scattering length using this
370: method. The degree of entanglement between the left and right arms
371: can be quantified using the von Neumann entropy
372: $E(t) = -\frac{1}{\log_2(N+1)} \sum_{m = -j}^{j} |c_m|^2 \log_2
373: |c_m|^2\;,$
374: which is clearly connected to the interference pattern at time
375: $\tau$ given by $\langle \hat{J}_z (\tau, \Theta) \rangle$. We plot
376: in Fig. \ref{entangle} the von Neumann entropy of the input state
377: against visibility ${\cal V}$ of the fringes ${\cal V} = (I_{max} -
378: I_{min})/(I_{max} + I_{min})$ where $I= |\langle \hat{J}_{z}(\tau)
379: \rangle|^2$ denotes the intensity of the fringe pattern. The
380: visibility of the interference fringes is obviously reduced with an
381: asymmetric beam splitting that corresponds to $\theta \neq \pi/2$.
382: The different line shapes correspond to different values of
383: nonlinearity. For a given value of visibility one obtains lower
384: entanglement with higher nonlinearity. This initially
385: counterintuitive result can be understood from the inset of Fig.
386: \ref{entangle}, which provides the relationship between the
387: fractional number difference between the two arms of the initial
388: input state $\Delta P$ and the visibility of the fringes. As
389: expected ${\cal V} \rightarrow 0$ as $\Delta P \rightarrow 1$. For
390: $\Delta P < 0.75$, ${\cal V}$ already reaches 1 as one starts
391: finding atoms in the other arm such that $\langle \hat{J}_{z}
392: \rangle$ starts to take on negative values as a function of
393: $\Theta$. With higher nonlinearity one finds ${\cal V} \rightarrow
394: 1$ even for larger values of $\Delta P$. This may be understood as
395: the result of higher entanglement due to collisions. With higher
396: entanglement, the ``wave-like'' nature of the wave function is
397: enhanced (changing one arm affects the other arm instantaneously)
398: which encourages tunneling between the two arms, and subsequently
399: higher visibility.
400:
401: In conclusion, we have studied a potentially realizable scheme for
402: a MZI using BECs. The resulting intensity difference between the
403: two arms as a function of phase shift is found to demonstrate that,
404: fundamentally, an atom laser is not monochromatic but rather
405: comprises of a number of frequency components. The adiabatic
406: passage method ensured that the only ``alteration'' made to the
407: condensate is that of coherent splitting and the establishment of a
408: relative phase between the two arms which naturally puts the trapped
409: condensate onto a coherent spin state. The dynamics of the split
410: condensate was then given simply by a standard many-body Hamiltonian
411: over a finite time interval. The fact that monochromaticity is lost
412: even in this idealized model implies that one should exercise
413: caution when discussing an atom laser as simply a matter-wave
414: equivalent of a laser, since additional effects such as decoherence
415: due to atom losses is expected to degrade the quality even further.
416: Finally we observed that with higher nonlinearity, higher fringe
417: visibility can be obtained with very uneven beam splitting owing to
418: the enhanced entanglement between the two arms. This work suggests
419: that there are many new aspects to be explored in the time domain
420: quantum control of BEC using adiabatic passage.
421:
422:
423: \begin{thebibliography}{99}
424:
425: \bibitem{Wiseman} H. M. Wiseman, Phys. Rev. A {\bf 56}, 2068 (1997)
426:
427: \bibitem{DoubleWell} Y. Shin, M. Saba, T. A. Pasquini, W. Ketterle, D.E. Pritchard, and A.E. Leanhardt, Phys. Rev. Lett. {\bf 92}, 050405 (2004)
428:
429: \bibitem{AtomChip} Y. Shin, C. Sanner, G.-B. Jo, T. A. Pasquini, M. Saba, W. Ketterle, D. E. Pritchard, M. Vengalattore and M. Prentiss, Phys. Rev. A {\bf 72}, 021604(R) (2005)
430:
431: \bibitem{AtomChip2} G.-B. Jo, Y. Shin, S. Will, T. A. Pasquini, M. Saba, W. Ketterle , D. E. Pritchard, M. Vengalattore, and M. Prentiss, Phys. Rev. Lett. {\bf 98}, 030407 (2007)
432:
433: \bibitem{AtomChip3} G.-B. Jo, J.-H. Choi, C. A. Christensen, T. A. Pasquini, Y.-R. Lee, W. Ketterle, and D. E. Pritchard, Phys. Rev. Lett. {\bf 98}, 180401 (2007)
434:
435: \bibitem{Review} O. Morsch and M. Oberthaler, Rev. Mod. Phys. {\bf 78}, 179 (2006)
436:
437: \bibitem{LMG} H. J. Lipkin, N. Meshkov, and N. Glick, Nucl. Phys. A {\bf 62}
438: 188 (1965)
439:
440: \bibitem{ChoiEntangle} S. Choi, and N. P. Bigelow, Phys. Rev. A {\bf 72} 033612
441: (2005)
442:
443: \bibitem{bloch} F. T. Arrechi, E. Courtens, R. Gilmore, and
444: H. Thomas, Phys. Rev. A {\bf 6}, 2211 (1972)
445:
446: \end{thebibliography}
447:
448: \end{document}
449: