1:
2: \documentclass[aps,pra,twocolumn,showpacs,floatfix,twoside,superscriptaddress]{revtex4}
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: \usepackage{amsfonts}
5: \usepackage{amsmath}
6: \usepackage{amssymb}
7: \usepackage{graphicx}
8:
9:
10: \begin{document}
11:
12: \title{Interference of Bose-Einstein condensates and entangled single-atom
13: state in a spin-dependent optical lattice}
14: \author{Linghua Wen}
15: \email{wenlinghua@wipm.ac.cn}
16: \author{Min Liu}
17: \author{Hongwei Xiong}
18: \email{xionghongwei@wipm.ac.cn}
19: \affiliation{State Key Laboratory
20: of Magnetic Resonance and Atomic and Molecular Physics, Wuhan
21: Institute of Physics and Mathematics, Chinese Academy of Sciences,
22: Wuhan 430071, P. R. China} \affiliation{Center for Cold Atom
23: Physics, Chinese Academy of Sciences, Wuhan 430071, P. R. China}
24: \affiliation{Graduate School, Chinese Academy of Sciences, Beijing
25: 100080, P. R. China}
26: \author{Mingsheng Zhan}
27: \email{mszhan@wipm.ac.cn}
28: \affiliation{State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics,
29: Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences,
30: Wuhan 430071, P. R. China}
31: \affiliation{Center for Cold Atom Physics, Chinese Academy of Sciences, Wuhan 430071, P.
32: R. China}
33: \date{\today }
34:
35: \begin{abstract}
36: We present a theoretical model to investigate the interference of
37: an array of Bose-Einstein condensates loaded in a one-dimensional
38: spin-dependent optical lattice, which is based on an assumption
39: that for the atoms in the entangled single-atom state between the
40: internal and the external degrees of freedom each atom interferes
41: only with itself. Our theoretical results agree well with the
42: interference patterns observed in a recent experiment by Mandel
43: \textit{et al}. [Phys. Rev. Lett. \textbf{91}, 010407 (2003)]. In
44: addition, an experimental suggestion of nonuniform phase
45: distribution is proposed to test further our theoretical model and
46: prediction. The present work shows that the entanglement of a
47: single atom is sufficient for the interference of the condensates
48: confined in a spin-dependent optical lattice and this interference
49: is irrelevant with the phases of individual condensates, i.e.,
50: this interference arises only between each condensate and itself
51: and there is no interference effect between two arbitrary
52: different condensates.
53: \end{abstract}
54:
55: \pacs{03.75.Lm, 03.75.Gg, 05.60.Gg} \maketitle
56:
57: \section{Introduction}
58:
59: Since the first interference measurement \cite{R1} on two
60: expanding condensates there has been a growing interest in the
61: experimental and theoretical study on the interference of
62: Bose-Einstein condensates (BECs). In particular, optical lattices
63: created by retroreflected laser beams provide a unique tool for
64: testing at a fundamental level the quantum properties of BECs in a
65: periodic potential \cite{R2}. The interference patterns obtained
66: from the expansion of an array of condensates trapped in an
67: optical lattice are commonly used as a probe of the phase
68: properties of this system \cite{R3,R4,R5,R6}. Current
69: understanding of the interference of BECs is largely based on the
70: concept of phase coherence which reveals the superfluidity and the
71: matter wave nature of the condensates.
72:
73: For the interference of two condensates, it is shown that an
74: interference pattern arises whether they are initially in phase
75: state (with locked relative phase) \cite{R1} or in Fock state
76: (with definite particle number) \cite{R7,R8,R9,R10}. In the latter
77: case, the interference effect is still originated from a
78: well-defined relative phase which is "built up" during the
79: sequences of measurement, i.e., the definite phase is derived from
80: the dynamic evolution of this system (initially with random
81: relative phase). For a fully coherent array of condensates in
82: optical lattices, the interference pattern obtained from the free
83: expansion is a natural result of the fixed relative phases between
84: different condensates belonging to consecutive wells
85: \cite{R3,R4,R5,R11,R12,R13,R14,R15,R16}. When
86: the coherent array of condensates enter the Mott insulating phase (MIP) \cite%
87: {R17,R18} in which phase coherence is lost, things become
88: complicated. The pioneering experiment \cite{R17} of the
89: superfluid to Mott-insulator transition demonstrated that when the
90: weakly interacting gas entered the MIP the interference pattern
91: became blurry even disappeared completely. Whereas for a strong
92: interacting gas it has been pointed out that a good measure for
93: this Mott transition was excitation spectrum rather than
94: interference pattern \cite{R19}. Besides, relevant theoretical
95: works \cite{R20,R21} based on a correlation function method imply
96: that even in the MIP interference pattern should also be observed
97: in a single measurement. On the other hand, a recent interference
98: experiment by Hadzibabic \textit{et al.} \cite{R22} states that
99: the periodicity of the optical lattice is sufficient for the
100: interference of an array of independent BECs even with no phase
101: coherence. A similar discussion is also found in a theoretical
102: Ref. \cite{R23}. It is therefore important to explore the physics
103: of interference patterns produced following the releasing and
104: expansion of BECs.
105:
106: In this paper, we present a theoretical model to investigate the
107: interference of an array of BECs confined in a spin-dependent
108: optical lattice, which is motivated by a recent experiment by
109: Mandel \textit{et al.} \cite{R24}. The interference patterns
110: obtained from our theoretical model and numerical calculations
111: agree well with those observed in the experiment \cite{R24}. Our
112: conclusion is that the interference of an array of Bose
113: condensates trapped in a spin-dependent optical lattice results
114: from the entanglement of a single atom and the interference is
115: irrelevant with the phases of individual condensates.
116:
117: This paper is organized as follows: In Sec.~2, after introducing
118: the basic spirit of the experiment \cite{R24} a theoretical model
119: is presented. In Sec.~3, we calculate numerically the density
120: distributions of the wave packets in spin states $\left\vert
121: 1\right\rangle $ and $\left\vert 0\right\rangle $, respectively.
122: Then we compare the theoretical interference patterns with those
123: observed in the experiment. The following Sec.~4 deals with an
124: experimental suggestion of nonuniform phase distribution, which
125: can be used to test further the theoretical model and prediction.
126: Finally, discussion and conclusion is given in Sec.~5.
127:
128:
129: \section{Theoretical model}
130:
131: For a Bose-condensed gas in a harmonic magnetic trap and a
132: three-dimensional (3D) optical lattice, the atoms are localized on
133: individual lattice sites when the system enters the MIP. After
134: switching off the magnetic trap and the lattice potentials along
135: the $y$ and $z$ directions there only exists a 1D spin-dependent
136: optical lattice along the $x$ direction which is formed by two
137: counterpropagating laser beams with linear polarization vectors
138: enclosing an angle $\theta $. Then each atom is prepared into a
139: coherent superposition of two spin states $\left\vert
140: 0\right\rangle \equiv \left\vert F=1,m_{F}=-1\right\rangle $ and
141: $\left\vert 1\right\rangle \equiv \left\vert
142: F=2,m_{F}=-2\right\rangle $ using a microwave pulse. By changing
143: the polarization angle $\theta $ one can realize the splitting and
144: transport of atomic wave packets such that the
145: wave packets of an atom in spin states $\left\vert 0\right\rangle $ and $%
146: \left\vert 1\right\rangle $ respectively are transported in
147: opposite directions, which is the so-called spin-dependent
148: transport. Finally, the optical lattice is turned off and the
149: emerging interference patterns can be used as a diagnosis signal
150: for the coherence of the spin-dependent transport. In a word, the
151: spin-dependent transport is the principal idea of the experiment
152: by Mandel \textit{et al. }\cite{R24}.
153:
154: Our starting point is that there is an array of Bose condensates
155: formed in the 1D spin-dependent optical lattice along the
156: horizontal $x$ direction when the magnetic trap and the lattice
157: potentials along the $y$ and $z$ directions are turned off. Since
158: the system experienced a Mott transition in advance these
159: condensates do not have any phase coherence relative to each other
160: any more. In this situation, the tunnelling between neighboring
161: lattice sites is suppressed and the effects of the atomic
162: interactions during the expansion of the condensates can be
163: neglected, which holds in the experiment. Since each condensate
164: confined in the lattice potential is fully coherent, in the frame
165: of single particle theory it can be described by a single order
166: parameter $\Psi _{k}=(N/(2k_{M}+1))^{1/2}\exp [i\theta _{k}]\phi
167: _{k}$ according to the Hartree or mean-field approximation
168: \cite{R25}, where $\phi _{k}$ is the single-particle wave function
169: in the $k$th lattice site and $\theta _{k}$ is the initially
170: random relative phase of the $k$th condensate. The coefficient
171: $N/(2k_{M}+1)$ represents the average particle number of each
172: condensate, with $N$ denoting the total particle number of the
173: whole condensate, and $2k_{M}+1$ being the total number of lattice
174: sites.
175:
176: Following the experimental manipulation sequence in \cite{R24}, we
177: now
178: consider an atom with two spin states $\left\vert 0\right\rangle $ and $%
179: \left\vert 1\right\rangle $ forming its two logical basis-vectors.
180: Initially, the atom lies in spin state $\left\vert 0\right\rangle
181: _{k}$. Without loss of generality, by using an initial arbitrary
182: $\alpha $ microwave pulse to drive Rabi oscillations between the
183: two spin states, the atom can be placed into a coherent
184: superposition of the two spin states $\left\vert 0\right\rangle
185: _{k} $ and $\left\vert 1\right\rangle _{k}$,
186:
187: \begin{equation} \label{evolution rule}
188: \left\{%
189: \begin{array}{ll}
190: \left\vert 0\right\rangle _{k}\to\cos [\frac{\alpha
191: }{2}]\left\vert 0\right\rangle _{k}+i\sin [\frac{\alpha
192: }{2}]\left\vert 1\right\rangle _{k},
193: & \\
194: \left\vert 1\right\rangle _{k}\to i\sin [\frac{\alpha
195: }{2}]\left\vert 0\right\rangle _{k}+\cos [\frac{\alpha
196: }{2}]\left\vert 1\right\rangle _{k}.
197: & \\
198: \end{array}%
199: \right.
200: \end{equation}
201: After a spin-dependent transport, the spin state\ of the atom is given by $%
202: \cos [\alpha /2]\left\vert 0\right\rangle _{k}+i\exp [i\beta ]\sin
203: [\alpha /2]\left\vert 1\right\rangle _{k+r}$, where the spatial
204: wave packet of the atom is split into two components in states
205: $\left\vert 0\right\rangle _{k}$ and $\left\vert 1\right\rangle
206: _{k+r}$, respectively, i.e., the atomic wave packet is delocalized
207: over the $k$th and the $(k+r)$th lattice site. In the above
208: notation, the wave packet in state $\left\vert 0\right\rangle $
209: has retained the original lattice site index. Here, $r$ denotes
210: the separation between two wave packets which are originated from
211: the same $k$th lattice site. The relative phase $\beta $ between
212: the two wave packets, being independent of the number of
213: particles, is determined by the accumulated kinetic and potential
214: energy phase in the transport process. With the choice of
215: parameters in the experiment, the phase $\beta $ is almost
216: constant throughout the cloud of atoms and its absolute value is
217: small. Consequently, the atomic wave function can be described by
218: an entangled single-atom state, i.e., an entangled quantum state
219: between the internal degree of freedom (spin) and the external
220: degree of freedom (spatial wave packet)
221:
222: \begin{equation} \label{entangled state}
223: \psi _{k}=\cos [\frac{\alpha }{2}]\left\vert 0\right\rangle
224: _{k}\varphi _{k}+i\exp [i\beta ]\sin [\frac{\alpha }{2}]\left\vert
225: 1\right\rangle _{k+r}\varphi _{k+r}.
226: \end{equation}
227:
228: We assume that the spatial wave packet has a form of Gaussian
229: distribution in coordinate space, i.e., $\varphi _{k}=A\exp
230: [-(x-kd)^{2}/2\sigma ^{2}]$, where $d=\lambda /2$ is the period of
231: the optical lattice and $\lambda $ is the wavelength of the
232: retroreflected laser beams. $A=1/\sigma ^{1/2}\pi ^{1/4}$ is a
233: normalization constant, and $\sigma $ denotes the width of the
234: condensate in each optical well. By applying a final $\pi /2$
235: microwave pulse whose transform rule is given by
236: Eq.(\ref{evolution rule}), one has
237:
238: \begin{equation}
239: \phi _{k}=\left\vert 0\right\rangle \Xi _{0,k}+\left\vert
240: 1\right\rangle \Xi _{1,k}, \label{erasing which way information}
241: \end{equation}%
242: where $\Xi _{0,k}$ and $\Xi _{1,k}$ are respectively given by
243:
244: \begin{eqnarray}
245: \Xi _{0,k} &=&\frac{A}{\sqrt{2}}\{\cos [\frac{\alpha }{2}]\exp [-\frac{%
246: (x-kd)^{2}}{2\sigma ^{2}}] \nonumber \\
247: &&-\sin [\frac{\alpha }{2}]\exp [i\beta -\frac{(x-(k+r)d)^{2}}{2\sigma ^{2}}%
248: ]\}, \label{wave function of spin 0} \\
249: \Xi _{1,k} &=&\frac{iA}{\sqrt{2}}\{\cos [\frac{\alpha }{2}]\exp [-\frac{%
250: (x-kd)^{2}}{2\sigma ^{2}}] \nonumber \\
251: &&+\sin [\frac{\alpha }{2}]\exp [i\beta -\frac{(x-(k+r)d)^{2}}{2\sigma ^{2}}%
252: ]\}. \label{wave function of spin 1}
253: \end{eqnarray}%
254: In Eq.(\ref{erasing which way information}), the indices of spin states $%
255: \left\vert 0\right\rangle $ and $\left\vert 1\right\rangle $ are
256: removed in view of the bosonic identity.
257:
258: Once the spin-dependent optical lattice is switched off, the
259: evolution of the spatial components $\Xi _{j,k}(x,t)$ $(j=0,1)$ of
260: the atomic wave function can be derived by the propagator method
261: \cite{R15,R16,R25}
262:
263: \begin{equation}
264: \Xi _{j,k}(x,t)=\int\nolimits_{-\infty }^{\infty }K(x,t;y,t=0)\Xi
265: _{j,k}(y,t=0)dy, \label{integral equation}
266: \end{equation}%
267: where $\Xi _{j,k}(y,t=0)$ $(j=0,1)$ are the spatial components at
268: the initial time $t=0$ which are given by Eqs.(\ref{wave function
269: of spin 0}) and (\ref{wave function of spin 1}), and
270: $K(x,t;y,t=0)$ is the propagator in free space expressed as
271: \cite{R26}
272:
273: \begin{equation}
274: K(x,t;y,t=0)=[\frac{m}{2\pi i\hbar t}]^{\frac{1}{2}}\exp [\frac{im}{2\hbar t}%
275: (x-y)^{2}]. \label{propagator}
276: \end{equation}%
277: By combining the formulae (\ref{wave function of spin 0})-(\ref{propagator}%
278: ), one can obtain the following analytical results of the spatial
279: components after a straightforward calculation:
280:
281: \begin{eqnarray}
282: \Xi _{0,k}(x,t) &=&\frac{A}{\sqrt{2(1+i\gamma t)}}\{\cos [\frac{\alpha }{2}%
283: ]\exp [-\frac{(x-kd)^{2}}{2\sigma ^{2}(1+i\gamma t)}] \nonumber \\
284: &&-\sin [\frac{\alpha }{2}]\exp [i\beta
285: -\frac{(x-(k+r)d)^{2}}{2\sigma
286: ^{2}(1+i\gamma t)}]\}, \label{analytical result 0} \\
287: \Xi _{1,k}(x,t) &=&\frac{iA}{\sqrt{2(1+i\gamma t)}}\{\cos [\frac{\alpha }{2}%
288: ]\exp [-\frac{(x-kd)^{2}}{2\sigma ^{2}(1+i\gamma t)}] \nonumber \\
289: &&+\sin [\frac{\alpha }{2}]\exp [i\beta
290: -\frac{(x-(k+r)d)^{2}}{2\sigma ^{2}(1+i\gamma t)}]\},
291: \label{analytical result 1}
292: \end{eqnarray}%
293: where the parameter $\gamma =\hbar /m\sigma ^{2}$ denotes the
294: trapping frequency within a single well of the optical lattice.
295:
296: We now consider the density distribution of the overall
297: condensates after switching off the spin-dependent optical
298: lattice. The wave function of the whole sample at time $t$ can be
299: expressed as
300:
301: \begin{equation}
302: \Psi (x,t)=\sum_{k=-k_{M}}^{k_{M}}\sqrt{\frac{N}{2k_{M}+1}}\exp
303: [i\theta _{k}]\phi _{k}(x,t), \label{total wave function}
304: \end{equation}%
305: where the time-dependent atomic wave function is given by $\phi
306: _{k}(x,t)=\left\vert 0\right\rangle \Xi _{0,k}(x,t)+\left\vert
307: 1\right\rangle \Xi _{1,k}(x,t)$, and $\theta _{k}$ denotes the
308: random phase of the $k$th condensate at time $t$. In
309: Eq.(\ref{total wave function}), we have neglected the phase
310: diffusion of each condensate possibly induced by quantum and
311: thermal fluctuations, which won't affect the essential of our
312: present problem.
313:
314: For a Bose-condensed gas confined in a trap, the phase fluctuation
315: is
316: characterized by the fluctuations in the chemical potential \cite%
317: {R27,R28,R29}. In the presence of a 1D optical lattice with
318: sufficiently strong intensity, the phase fluctuations for
319: different condensates in individual wells are independent from
320: each other. Two dominating physical ingredients are responsible
321: for the creation of phase diffusion: one is the collision between
322: condensed atoms and background hot cloud (thermal fluctuation),
323: and the second is spontaneously collective excitation due to
324: quantum fluctuation \cite{R29}. In real experiments, the phase
325: diffusion effect is small and can be omitted safely. Actually,
326: when taking into account the phase diffusion effect, the holistic
327: characters of interference pattern will not change except that the
328: central peak of the interference pattern will decrease a little
329: relatively \cite{R29}.
330:
331: Obviously, there is no interference between the wave packets in
332: different spin states as the two logical basis-vectors $\left\vert
333: 0\right\rangle $ and $\left\vert 1\right\rangle $\ are orthogonal.
334: The following model is based on an assumption that for the atoms
335: in the entangled single-atom state between the internal and
336: external degrees of freedom each atom interferes only with itself.
337: Concretely, the density distributions of the wave packets in spin
338: states $\left\vert 0\right\rangle $ or $\left\vert 1\right\rangle
339: $, i.e., the density distributions of the overall condensates
340: confined in the total occupied lattice sites, are not expressed by
341: Eqs.(\ref{density of spin 0}) and (\ref{density of spin 1})
342:
343: \begin{eqnarray}
344: n_{0}(x,t) &=&\left\vert \sum_{k=-k_{M}}^{k_{M}}\sqrt{\frac{N}{2k_{M}+1}}%
345: \exp [i\theta _{k}]\Xi _{0,k}(x,t)\right\vert ^{2},
346: \label{density of spin 0} \\
347: n_{1}(x,t) &=&\left\vert \sum_{k=-k_{M}}^{k_{M}}\sqrt{\frac{N}{2k_{M}+1}}%
348: \exp [i\theta _{k}]\Xi _{1,k}(x,t)\right\vert ^{2}, \label{density
349: of spin 1}
350: \end{eqnarray}%
351: but given by Eqs.(\ref{simplified density of spin 0}) and
352: (\ref{simplified density of spin 1})
353:
354: \begin{eqnarray}
355: n_{0}(x,t) &=&\frac{N}{2k_{M}+1}\sum_{k=-k_{M}}^{k_{M}}\left\vert
356: \Xi
357: _{0,k}(x,t)\right\vert ^{2}, \label{simplified density of spin 0} \\
358: n_{1}(x,t) &=&\frac{N}{2k_{M}+1}\sum_{k=-k_{M}}^{k_{M}}\left\vert
359: \Xi _{1,k}(x,t)\right\vert ^{2}. \label{simplified density of
360: spin 1}
361: \end{eqnarray}
362:
363: The test criterion of this model depends on whether its
364: theoretical prediction accords with the experimental results,
365: i.e., whether this model can interpret well the experiment. Then
366: we perform a Monte-Carlo analysis of $n_{1}(x,t)$ by assigning
367: sets of random numbers to the phase $\{\theta _{k}\}$. Our
368: simulation results show that the interference patterns based on
369: Eq.(\ref{density of spin 1}) don't agree with those observed in
370: the experiment \cite{R24} at all. In addition, provide that there
371: are locked phases for individual condensates, i.e., the phase
372: $\theta _{k}$ is not random (a simplest case is that the phase of
373: each condensate is the same), the calculation also shows that the
374: interference patterns derived from Eq.(\ref{density of spin 1})
375: are not in agreement with the experimental results. Thus it is
376: implied that Eqs.(\ref{density of spin 0}) and (\ref{density of
377: spin 1}) are invalid in explaining the experiment. As expected,
378: however, we find that the theoretical interference patterns based
379: on Eq.(\ref{simplified density of spin 1}) agree well with the
380: observed interference patterns in the experiment (see Fig.1-Fig.2
381: in section 3), which indicates that this model, i.e.
382: Eqs.(\ref{simplified density of spin 0}) and (\ref{simplified
383: density of spin 1}), can be employed to describe the real physics
384: of the emerging interference patterns in the experiment
385: \cite{R24}.
386:
387: The physical essence of the density distributions expressed by Eqs.(\ref%
388: {simplified density of spin 0}) and (\ref{simplified density of
389: spin 1}) is that due to the entanglement of a single atom each
390: atom interferes only with itself (or each condensate interferes
391: only with itself), i.e., there is no interference effect between
392: two arbitrary different condensates. In other words, the
393: entanglement of a single atom is sufficient for the interference
394: of the overall condensates and this interference will be
395: irrelevant with the phases of individual condensates.
396:
397: \section{Density distributions and evolution}
398:
399: By using the experimental parameters in \cite{R24}, we plot the
400: density distributions of the atomic wave packets in states
401: $\left\vert
402: 0\right\rangle $ and $\left\vert 1\right\rangle $ respectively based on Eqs.(%
403: \ref{simplified density of spin 0}) and (\ref{simplified density
404: of spin 1}).
405:
406:
407: \subsection{Parameters}
408:
409: In the following calculations, the relevant parameters are
410: consistent with
411: those in the experiment, where $\alpha =\pi /2$, $N=3\times 10^{5}$, $%
412: \lambda =785$ nm, and $d=392.5$ nm. For simplicity, we treat the
413: relative phase $\beta $ as zero. Nevertheless, we'll also take
414: into account the effect of it on the interference patterns to
415: compare with the omitted case. Since the value of $\sigma $, which
416: characterizes the width of condensate in
417: each lattice site, is chiefly determined by the optical confinement \cite%
418: {R11}, one can evaluate it in terms of a variational calculation.
419: As a result, the ratio $\sigma /d=0.173$ is obtained. The total
420: number of the lattice sites $2k_{M}+1$ can be determined theoretically by the formula $%
421: k_{M}^{2}=2\hbar \varpi (15Nad/8\pi ^{1/2}a_{ho}\sigma
422: )^{2/5}/(m\omega _{x}^{2}d^{2})$ (see Eq.(10) in \cite{R11}),
423: where the geometric average of the magnetic frequencies $\varpi
424: _{x}=2\pi \times 16$ Hz, $m$ is the mass of
425: $^{87}$Rb atom, the oscillator atom, the oscillator length $a_{ho}=\sqrt{%
426: \hbar /m\varpi }$, and the s-wave scattering length for $^{87}$Rb atom is $%
427: a\sim 50$ \r{A}. Thus $k_{M}\sim 50$ is obtained from the above
428: equation.
429:
430: \subsection{Density distributions in state $\left\vert 1\right\rangle $}
431:
432: In order to compare with the experiment, we consider firstly the
433: density distribution of the wave packets in state $\left\vert
434: 1\right\rangle $ after the optical lattice is switched off with a
435: time of flight being 14 ms. The
436: analytical result of $n_{1}(x,t)$ at time $t$ is given by Eq.(\ref%
437: {simplified density of spin 1}). Shown in Fig.1(a) is the density
438: distribution (in units of $H=NA^{2}/(2k_{M}+1)$) in state
439: $\left\vert 1\right\rangle $ at $t=14$ ms after initially
440: localized atoms have been delocalized over two lattice sites. Note
441: that in all the figures plotted in this paper the horizontal
442: coordinate $x$ is in units of $\mu $m and the vertical coordinate
443: is in units of $H=NA^{2}/(2k_{M}+1)$. The density distributions in
444: the cases that initially localized atoms have been delocalized
445: over three (b), four (c), five (d), six (e), and seven (f) lattice
446: sites are given in Fig.1(b)--(f), respectively, where the
447: delocalized extension\ is denoted by $r$. With the separation $r$
448: increasing, we see that the fringe spacing of interference
449: patterns decreases remarkably and the visibility of the
450: interference patterns reduces distinctively (see Fig.1), which is
451: in agreement with the experimental results (see Fig.4 in
452: \cite{R24}).
453:
454: \begin{figure}[h]
455: \centerline{\includegraphics[width=6cm,angle=-90]{Figure1.eps}}
456: \caption{Density distributions in state $\left\vert 1\right\rangle
457: $ after switching off the spin-dependent optical lattice in the
458: cases that initially localized atoms have been delocalized over
459: two (a), three (b), four (c), five (d), six (e), and seven (f)
460: lattice sites by the interferometer sequence (see Fig.3 in
461: \protect\cite{R24}). The time of flight period is $14$
462: ms. The vertical coordinate $n_{1}(x,t)$ is in units of $H$ ($%
463: H=NA^{2}/(2k_{M}+1)$) and the horizontal coordinate $x$ is in units of $%
464: \protect\mu $m. $r$\ denotes the separation between the two wave
465: packets originated from the same lattice site.}
466: \end{figure}
467:
468: In Fig.2, we show the evolution of the density distribution in state $%
469: \left\vert 1\right\rangle $ after initially localized atoms have
470: been delocalized over three lattice sites. Displayed in Fig.2(b)
471: is the density distribution at $t=15$ ms, which agrees well with
472: the observed interference pattern in the experiment (see Fig.5 in
473: \cite{R24}).
474:
475: \begin{figure}[h]
476: \centerline{\includegraphics[width=2cm,angle=-90]{Figure2.eps}}
477: \caption{Evolution of the density distribution in state
478: $\left\vert 1\right\rangle $ with time $t$ after switching off the
479: spin-dependent optical lattice in the case that initially
480: localized atoms have been delocalized over three lattice sites.
481: The density distributions are shown at $t=2$ ms (a) and $t=15$ ms
482: (b).}
483: \end{figure}
484:
485: In the calculations mentioned above, we have neglected the effect
486: of the relative phase $\beta $ on the density distribution by
487: treating it as zero.
488: Displayed in Fig.3 are the density distributions for the relative phase $%
489: \beta $ with $-\pi /12$ and $-\pi /3$ respectively. When taking
490: into account the relative phase $\beta $ between the two wave
491: packets the right-hand side peaks of the density distributions
492: become higher than the left-hand side ones, which breaks the
493: symmetry of the interference patterns to a certain extent. In
494: addition, the larger the absolute value of the phase $\beta $ is,
495: the weaker the symmetry of the interference pattern becomes.
496: According to the observed interference patterns, we can conclude
497: that the absolute value of the phase $\beta $ is possibly close to
498: zero in the experiment \cite{R24}.
499:
500: \begin{figure}[h]
501: \centerline{\includegraphics[width=2cm,angle=-90]{Figure3.eps}}
502: \caption{The effect of the phase $\protect\beta $ on the density
503: distribution in state $\left\vert 1\right\rangle $ after switching
504: off the spin-dependent optical lattice in the case that initially
505: localized atoms have been delocalized over three lattice sites.
506: The time of flight period is
507: $15$ ms. The density distributions are shown at $\protect\beta =-\protect\pi %
508: /12$ (a) and $\protect\beta =-\protect\pi /3$ (b), respectively.}
509: \end{figure}
510:
511: \subsection{Density distributions in state $\left\vert 0\right\rangle $}
512:
513: Now, we discuss the density distributions in state $\left\vert
514: 0\right\rangle $ which were not observed in \cite{R24}. Similarly
515: to the foregoing analysis, the analytical result of the density
516: distributions in state $\left\vert 0\right\rangle $ is given by
517: Eq.(\ref{simplified density
518: of spin 0}). Shown in Fig.4 are the density distributions in state $%
519: \left\vert 0\right\rangle $ at time $t=14$ms. In contrast with the
520: density distributions in state $\left\vert 1\right\rangle $, the
521: positions of the sharp peaks in Fig.4(a)--(f) just become those of
522: local minimum densities in Fig.1(a)--(f) and vice versa, which can
523: be interpreted by the conservation of energy and particle number.
524:
525: \begin{figure}[h]
526: \centerline{\includegraphics[width=6cm,angle=-90]{Figure4.eps}}
527: \caption{Density distributions in state $\left\vert 0\right\rangle
528: $ after switching off the spin-dependent optical lattice in the
529: cases that initially localized atoms have been delocalized over
530: two (a), three (b), four (c), five (d), six (e), and seven (f)
531: lattice sites. The time of flight period is $14$ ms.}
532: \end{figure}
533:
534: \section{Experimental suggestion}
535:
536: As mentioned above, the Bose condensates confined in the 1D
537: spin-dependent optical lattice have no phase coherence relative to
538: each other as the system experienced a Mott transition beforehand.
539: In this situation, the density distributions of the atomic wave
540: packets based on our theoretical model are in good agreement with
541: the observed interference patterns in \cite{R24}. From the
542: theoretical model (see Eqs.(\ref{simplified density of spin 0})
543: and (\ref{simplified density of spin 1})), the density
544: distributions of the atomic wave packets in different spin states
545: are irrelevant with the phases of individual condensates in this
546: system. Hence, when there is a locked phase distribution for the
547: array of condensates, the density distributions will not change.
548:
549: To test further the validity of this model, we propose an
550: experimental suggestion of nonuniform phase distribution.
551: Concretely, we design a fixed linear phase distribution with a
552: total width of $2\pi $ for the array of condensates, in which the
553: phase difference between two neighboring condensates is $\delta
554: \theta =2\pi /(2k_{M}+1)$ ($k_{M}\sim 50$), i.e., the phase of the
555: $k$th condensate can
556: be expressed by $\theta _{k}=2\pi (k_{M}+k)/(2k_{M}+1)$ ($k=-k_{M},...,k_{M}$%
557: ). This goal can be achieved by using the techniques of phase
558: redistributions such as phase imprinting \cite{R30,R31} and phase
559: engineering \cite{R32,R33}. Once a nonuniform phase distribution
560: is
561: performed successfully on the array of condensates, one applies an initial $%
562: \alpha =\pi /2$ microwave pulse to drive Rabi oscillations between
563: the two spin states $\left\vert 0\right\rangle $ and $\left\vert
564: 1\right\rangle $, respectively. Thus all the atoms initially in
565: spin state $\left\vert 0\right\rangle $ are placed in a coherent
566: superposition of the two spin states, where the transform rule is
567: given by Eq.(\ref{evolution rule}).
568:
569: The following deduction is similar to that in the preceding
570: sections. After a spin-dependent transport and applying a final
571: $\pi /2$ microwave pulse as well as a releasing of the optical
572: lattice, the density distributions of the wave packets in states
573: $\left\vert 0\right\rangle $ and $\left\vert
574: 1\right\rangle $ respectively would be given by Eqs.(\ref{density of spin 0}%
575: ) and (\ref{density of spin 1}) if there were interference effects
576: between different condensates. In this case, the density
577: distributions would be quite different from Fig.1-Fig.4. Shown in
578: Fig.5 would be the density distribution of the wave packets at
579: time $t=15$ ms with $r=2d$ and $\delta \theta =2\pi /(2k_{M}+1)$.
580: From Fig.5, we can see that there exists a strong decay and
581: revival of the density oscillation, and there is even no legible
582: interference fringe (see Fig.5(b)).
583:
584: \begin{figure}[h]
585: \centerline{\includegraphics[width=2cm,angle=-90]{Figure5.eps}}
586: \caption{Density distributions in states $\left\vert
587: 1\right\rangle $ (a) and $\left\vert 0\right\rangle $ (b)
588: respectively provide that there is a fixed phase distribution
589: among the array of condensates confined in the spin-dependent
590: optical lattice and there exists interference effects between
591: different condensates. The time of flight period is $15$ ms. The
592: phase of
593: the $k$th condensate is given by $\protect\theta _{k}=2\protect\pi %
594: (k_{M}+k)/(2k_{M}+1)$ ($k=-k_{M},...,k_{M}$). Here $r=2$ denotes
595: that initially localized atoms have been delocalized over three
596: lattice sites.}
597: \end{figure}
598:
599: Due to the entanglement of a single atom, however, we predict that
600: after the nonuniform phase distribution the density distributions
601: of the wave packets in states $\left\vert 0\right\rangle $ and
602: $\left\vert 1\right\rangle $ respectively are still given by
603: Eqs.(\ref{simplified density of spin 0}) and (\ref{simplified
604: density of spin 1}), i.e., the density distributions will not
605: change. In other words, for the atoms in the entangled single-atom
606: state each atom interferes only\ with itself whether the
607: condensates are full coherent (with fixed relative phases) or
608: completely independent (with random relative phases), which
609: implies that the entanglement of a single atom is sufficient for
610: the interference of BECs in a spin-dependent optical lattice and
611: the interference effect is irrelevant with the phases of the
612: individual condensates. This experimental proposal provides a
613: straight way to test further our theoretical model and prediction.
614:
615: \section{Discussion and conclusion}
616:
617: To summarize, we have developed a theoretical model to investigate
618: the interference of an array of BECs confined in a 1D
619: spin-dependent optical lattice by calculating the density
620: distributions and evolution of the atomic wave packets. In such a
621: system as experienced beforehand a Mott transition and a
622: spin-dependent transport, each atom can be described by an
623: entangled single-atom state between the internal (spin) and the
624: external (spatial wave packet) degrees of freedom . Our
625: theoretical model is based on an assumption that for the atoms in
626: the entangled single-atom state each atom interferes only\ with
627: itself. The results obtained from this model agree well with the
628: interference patterns observed in a recent experiment \cite{R24},
629: which in turn verifies the validity of this model and assumption.
630: In addition, when taking into account the relative phase $\beta $
631: between the two wave packets of an atom which is obtained during
632: the transport process, it is found that the symmetry of the
633: density distributions is broken to a certain extent.
634:
635: From the present work, it has been shown that due to the
636: entanglement of a single atom each atom interferes only with
637: itself (or each condensate interferes only with itself in this
638: system), i.e., there is no interference effect between two
639: arbitrary different condensates. In other words, the entanglement
640: of a single atom is sufficient for the interference of BECs
641: confined in a spin-dependent optical lattice and the interference
642: shows no relevancy with the phases of individual condensates.
643: Finally, an experimental suggestion of nonuniform phase
644: distribution is proposed to test further our theoretical model and
645: prediction. The theoretical model presented here can be also
646: applied to describe the dynamics of BECs trapped in a combined
647: harmonic and optical lattice potential, wherein the number of
648: atoms in individual lattice sites is different. Possibly, the
649: method even can be extended to consider the case of non-perfect
650: Mott-insulator state.
651:
652: \begin{acknowledgments}
653: The authors would like to thank I. Bloch, J. Wang, K. L. Gao, and
654: Y. Wu for useful discussions. Pertinent comments and suggestions
655: from the editor and anonymous reviewers are acknowledged. This
656: work was supported by the National Natural Science Foundation of
657: China under Grant Nos.10474119,10205011 and 10474117, by the
658: National Fundamental Research Program of China under Grant
659: No.001CB309309, and also by funds from the Chinese Academy of
660: Sciences.
661: \end{acknowledgments}
662:
663: \bigskip
664:
665: \begin{thebibliography}{99}
666: \bibitem{R1} M. R. Andrews \emph{et al.}, Science \textbf{275}, 637 (1997).
667:
668: \bibitem{R2} I. Bloch, Physics world, April 2004, pp.25-29.
669:
670: \bibitem{R3} C. Orzel \emph{et al.}, Science \textbf{291}, 2386 (2001).
671:
672: \bibitem{R4} F. S. Cataliotti \emph{et al.}, Science \textbf{293},
673: 843 (2001).
674:
675: \bibitem{R5} M. Greiner \emph{et al.}, Phys. Rev. Lett. \textbf{87}, 160405 (2001).
676:
677: \bibitem{R6} H. Ellmann, J. Jersblad, and A. Kastberg, Phys. Rev. Lett.
678: \textbf{90}, 053001(2003).
679:
680: \bibitem{R7} J. Javanainen and S. M. Yoo, Phys. Rev. Lett. \textbf{76}, 161
681: (1996).
682:
683: \bibitem{R8} M. Naraschewski \emph{et al.}, Phys. Rev. A \textbf{54}, 2185 (1996).
684:
685: \bibitem{R9} J. I. Cirac, C. W. Gardiner, M. Naraschewski, and P. Zoller,
686: Phys. Rev. A \textbf{54}, R3714 (1996).
687:
688: \bibitem{R10} Y. Castin and J. Dalibard, Phys. Rev. A \textbf{55}, 4330
689: (1997).
690:
691: \bibitem{R11} P. Pedri \emph{et al.}, Phys.
692: Rev. Lett. \textbf{87}, 220401 (2001).
693:
694: \bibitem{R12} M. Greiner, O. Mandel, T. W. H\"{a}nsch, and I. Bloch, Nature
695: \textbf{419}, 51(2002).
696:
697: \bibitem{R13} O. Morsch \emph{et al.}, Phys. Rev. A \textbf{66}, 021601(R) (2002).
698:
699: \bibitem{R14} R. Roth and K. Burnett, Phys. Rev. A \textbf{67}, 031602(R)
700: (2003).
701:
702: \bibitem{R15} H. Xiong, S. Liu, G. Huang, and Z. Xu, J. Phys. B: At. Mol.
703: Opt. Phys. \textbf{35}, 4863 (2002).
704:
705: \bibitem{R16} S. Liu, H. Xiong, Z. Xu, and G. Huang, J. Phys. B: At. Mol.
706: Opt. Phys. \textbf{36}, 2083 (2003).
707:
708: \bibitem{R17} M. Greiner \emph{et al.}, Nature \textbf{415}, 39 (2002).
709:
710: \bibitem{R18} Y. Wu and X. Yang, Phys. Rev. A \textbf{68}, 013608 (2003).
711:
712: \bibitem{R19} T. St\"{o}ferle \emph{et al.}, Phy. Rev. Lett. \textbf{92}, 130403 (2004).
713:
714: \bibitem{R20} R. Bach and K. Rza\.{z}ewski, Phys. Rev. Lett. \textbf{92},
715: 200401 (2004).
716:
717: \bibitem{R21} R. Bach and K. Rza\.{z}ewski, Preprint cond-mat/0407022.
718:
719: \bibitem{R22} Z. Hadzibabic \emph{et al.}, Phys. Rev. Lett. \textbf{93}, 180403 (2004).
720:
721: \bibitem{R23} S. Ashhab, Preprint cond-mat/0407414.
722:
723: \bibitem{R24} O. Mandel \emph{et al.}, Phys. Rev. Lett. \textbf{91}, 010407 (2003).
724:
725: \bibitem{R25} C. J. Pethick and H. Smith, \textit{Bose-Einstein Condensation
726: in Dilute Gases }(Cambridge Univ. Press, Cambridge, 2002).
727:
728: \bibitem{R26} R. P. Feynman and A. R. Hibbs, \textit{Quantum Mechanics and
729: Path Integrals} (McGraw-Hill, New York, 1965).
730:
731: \bibitem{R27} M. Lewenstein and L. You, Phys. Rev. Lett. \textbf{77}, 3489
732: (1996).
733:
734: \bibitem{R28} A. Imamoglu, M. Lewenstein, and L. You, Phys. Rev. Lett.
735: \textbf{78}, 2511 (1997).
736:
737: \bibitem{R29} H. Xiong, S. Liu, G. Huang, and L. Wang, J. Phys. B: At. Mol.
738: Opt. Phys. \textbf{36}, 3315 (2003).
739:
740: \bibitem{R30} \L . Dobrek \emph{et al.}, Phys. Rev. A \textbf{60}, 3381(R) (1999).
741:
742: \bibitem{R31} S. Burger \emph{et al.}, Phys. Rev. Lett.
743: \textbf{83}, 5198 (1999).
744:
745: \bibitem{R32} J. Denschlag \emph{et al.}, Science \textbf{287}, 97 (2000).
746:
747: \bibitem{R33} L. D. Carr, J. Brand, S. Burger, and A. Sanpera, Phys. Rev. A
748: \textbf{63}, 051601(R) (2001).
749: \end{thebibliography}
750:
751: \end{document}
752: