1: %% ****** Start of file template.aps ****** %
2: %%
3: %%
4: %% This file is part of the APS files in the REVTeX 4 distribution.
5: %% Version 4.0 of REVTeX, August 2001
6: %%
7: %%
8: %% Copyright (c) 2001 The American Physical Society.
9: %%
10: %% See the REVTeX 4 README file for restrictions and more information.
11: %%
12: %
13: % This is a template for producing manuscripts for use with REVTEX 4.0
14: % Copy this file to another name and then work on that file.
15: % That way, you always have this original template file to use.
16: %
17: % Group addresses by affiliation; use superscriptaddress for long
18: % author lists, or if there are many overlapping affiliations.
19: % For Phys. Rev. appearance, change preprint to twocolumn.
20: % Choose pra, prb, prc, prd, pre, prl, prstab, or rmp for journal
21: % Add 'draft' option to mark overfull boxes with black boxes
22: % Add 'showpacs' option to make PACS codes appear
23: % Add 'showkeys' option to make keywords appear
24: %
25: %\documentclass[aps,pra,twocolumn,groupedaddress]{revtex4}
26: \documentclass[aps,pra,preprint,groupedaddress]{revtex4}
27: %\documentclass[aps,pra,preprint,superscriptaddress]{revtex4}
28: %\documentclass[aps,pra,onecolumn,groupedaddress]{revtex4}
29:
30: \usepackage{graphicx}% Include figure files
31: \usepackage{dcolumn}% Align table columns on decimal point
32: \usepackage{bm}% bold math
33:
34: % You should use BibTeX and apsrev.bst for references
35: % Choosing a journal automatically selects the correct APS
36: % BibTeX style file (bst file), so only uncomment the line
37: % below if necessary.
38: %\bibliographystyle{apsrev}
39:
40: \begin{document}
41:
42: % Use the \preprint command to place your local institutional report
43: % number in the upper righthand corner of the title page in preprint mode.
44: % Multiple \preprint commands are allowed.
45: % Use the 'preprintnumbers' class option to override journal defaults
46: % to display numbers if necessary
47: %\preprint{}
48:
49: %Title of paper
50: \title{Second Order Correlation Function of a Phase Fluctuating Bose-Einstein
51: Condensate}
52:
53: % repeat the \author .. \affiliation etc. as needed
54: % \email, \thanks, \homepage, \altaffiliation all apply to the current
55: % author. Explanatory text should go in the []'s, actual e-mail
56: % address or url should go in the {}'s for \email and \homepage.
57: % Please use the appropriate macro foreach each type of information
58:
59: % \affiliation command applies to all authors since the last
60: % \affiliation command. The \affiliation command should follow the
61: % other information
62: % \affiliation can be followed by \email, \homepage, \thanks as well.
63: \author{L. Cacciapuoti}
64: \altaffiliation[Present address: ]{BNM-SYRTE, Observatoire de
65: Paris, 61 avenue de l'Observatoire, 75014 Paris-France; e-mail:
66: Luigi.Cacciapuoti@obspm.fr.}
67: %\email{Luigi.Cacciapuoti@obspm.fr.}
68: \author{D. Hellweg}
69: \author{M. Kottke}
70: \author{T. Schulte}
71: \author{K. Sengstock,$^1$ W. Ertmer}
72: \author{J.J. Arlt}
73: \affiliation{Institut f\"{u}r Quantenoptik, Universit\"{a}t
74: Hannover, Welfengarten 1, 30167 Hannover, Germany \\
75: $^1$Institut f\"{u}r Laserphysik, Universit\"{a}t Hamburg, Luruper
76: Chaussee 149, 22761 Hamburg, Germany}
77: %\homepage[]{Your web page}
78: %\thanks{}
79: %\altaffiliation{}
80: \author{L. Santos}
81: \author{M. Lewenstein}
82: \affiliation{Institut f\"{u}r Theoretische Physik, Universit\"{a}t
83: Hannover, Appelstra{\ss}e 2, 30167 Hannover, Germany}
84: %Collaboration name if desired (requires use of superscriptaddress
85: %option in \documentclass). \noaffiliation is required (may also be
86: %used with the \author command).
87: %\collaboration can be followed by \email, \homepage, \thanks as well.
88: %\collaboration{}
89: %\noaffiliation
90:
91: \date{\today}
92:
93: \begin{abstract}
94: The coherence properties of phase fluctuating Bose-Einstein
95: condensates are studied both theoretically and experimentally. We
96: derive a general expression for the $N$-particle correlation
97: function of a condensed Bose gas in a highly elongated trapping
98: potential. The second order correlation function is analyzed in
99: detail and an interferometric method to directly measure it is
100: discussed and experimentally implemented. Using a Bragg
101: diffraction interferometer, we measure intensity correlations in
102: the interference pattern generated by two spatially displaced
103: copies of a parent condensate. Our experiment demonstrates how to
104: characterize the second order correlation function of a highly
105: elongated condensate and to measure its phase coherence length.
106: %We present theoretical and experimental studies on the coherence
107: %properties of phase fluctuating Bose-Einstein condensates. A
108: %general expression for the $N$-particle correlation function of a
109: %highly elongated Bose gas is explicitly calculated. We analyze in
110: %detail the behavior of the second order correlation function and
111: %discuss an interferometric technique to directly measure it. An
112: %analytic theory based on the linearization of density and phase
113: %around the self-similar solution of the Gross-Pitaevskii equation
114: %is developed to model the ballistic expansion of the phase
115: %fluctuating condensate during the measurement process. We
116: %experimentally demonstrate the method by using a Bragg diffraction
117: %interferometer to detect intensity correlations in the
118: %interference pattern. From an averaging process, the second order
119: %correlation function of the condensate can be extracted. This
120: %method reveals the possibility to measure the phase coherence
121: %length of the condensate.
122: \end{abstract}
123:
124: % insert suggested PACS numbers in braces on next line
125: \pacs{}
126: % insert suggested keywords - APS authors don't need to do this
127: %\keywords{}
128:
129: %\maketitle must follow title, authors, abstract, \pacs, and \keywords
130: \maketitle
131:
132: % body of paper here - Use proper section commands
133: % References should be done using the \cite, \ref, and \label commands
134: %\section{}
135: % Put \label in argument of \section for cross-referencing
136: %\section{\label{}}
137: % BEC coherence
138:
139: \section{Introduction}
140: Among the various topics related to the exciting field of
141: Bose-Einstein condensation (BEC) \cite{Dalfovo99}, the analysis of
142: coherence properties of degenerate Bose gases has attracted major
143: interest. Coherence plays a key role in the understanding of the
144: fundamentals of BEC, and has a crucial importance for many
145: promising BEC applications, such as matter wave interferometry,
146: guided atomic beams, and atom lasers. The coherent character of
147: trapped 3D condensates well below the BEC transition temperature
148: $T_{c}$ has been confirmed by several experiments, using
149: interferometric \cite{Andrews97,Hagley99} and spectroscopic
150: methods \cite{Stenger99}.
151:
152: % QuasiBEC: Theoretical
153: However, recent theoretical and experimental developments have
154: shown that phase coherence is far from being an obvious property
155: of BEC. In particular, a phase fluctuating BEC at equilibrium has
156: been theoretically predicted in one-dimensional \cite{Petrov00a},
157: two-dimensional \cite{Petrov00,Kagan00}, and even in highly
158: elongated, but still three-dimensional \cite{Petrov01} trapped
159: Bose gases. Interestingly, in these cases the density distribution
160: does not differ from the usual BEC profile, since density
161: fluctuations are largely suppressed by the repulsive mean-field
162: potential. These systems are commonly called quasicondensates.
163: Phase fluctuations can be induced either by quantum \cite{Ho99} or
164: by thermal fluctuations \cite{Kane67}. For typical experimental
165: temperatures quantum phase fluctuations can safely be neglected as
166: long as the system remains in the weakly-interacting regime
167: \cite{Gangardt03}. The amplitude of phase fluctuations, therefore,
168: depends strongly on temperature and trapping geometry. In this
169: sense, a nearly phase coherent BEC in a highly elongated trap can
170: only be achieved far below $T_c$, imposing severe limitations on
171: experiments in constrained geometries. Phase fluctuating BECs have
172: been the subject of recent theoretical efforts, including the
173: development of a modified mean-field theory valid in all
174: dimensions and all temperatures below the critical point
175: \cite{Andersen02,Khawaja03}, the analysis of dynamic correlation
176: functions \cite{Luxat02}, and the extension of Bogoliubov theory
177: to low-dimensional degenerate Bose gases \cite{Mora02}.
178:
179: % QUASIBEC: experimental
180: The phase fluctuating nature of highly elongated BECs was first
181: experimentally demonstrated in Ref.~\cite{Dettmer01}. During the
182: ballistic expansion, phase fluctuations transform into density
183: modulations. The appearance of phase fluctuations and their
184: statistic nature were studied and the dependence of their average
185: value on experimental parameters was characterized
186: \cite{Dettmer01,Hellweg01}. Moreover, the results obtained from
187: measurements of the energy released during the expansion confirmed
188: the absence of density fluctuations in the trapped cloud
189: \cite{Kreutzmann03,Richard03}. Recently, the physics of
190: quasicondensates has been studied by means of Bragg spectroscopy,
191: showing that the existence of phase fluctuations leads to an
192: observable broadening of the momentum distribution
193: \cite{Gerbier02a,Richard03}. A further experiment has analyzed the
194: phase coherence length of non-equilibrium BECs by means of a
195: condensate-focusing technique \cite{Shvarchuck02}.
196:
197: % This paper
198: In this paper, we present the theoretical foundation of our
199: studies on coherence properties of phase fluctuating condensates.
200: We analyze the behavior of the second order correlation function
201: for our experimental conditions and provide a detailed discussion
202: of the experimental technique used in Ref.~\cite{Hellweg03} to
203: measure it. This technique is based on the analysis of the density
204: correlations in the interference pattern generated by a matter
205: wave Bragg interferometer. In analogy to the original
206: Hanbury-Brown and Twiss experiment \cite{Hanbury56,Hanbury56a},
207: our method is used to extract the phase coherence length of the
208: degenerate Bose gas from density correlation measurements.
209:
210: % Organization of the paper
211: This paper is organized as follows: In Sec.~II, we briefly review
212: the theory of phase fluctuating Bose-Einstein condensates in 3D
213: elongated traps \cite{Petrov01} and analyze the evolution of the
214: phase pattern during the ballistic expansion. The knowledge of the
215: free dynamics of the phase is important to closely model the BEC
216: evolution during the measurement process. In Sec.~III, we study
217: the coherence properties of the condensate and derive a general
218: expression for the $N$-particle correlation function of highly
219: elongated 3D BECs. In Sec.~IV, the experimental technique used to
220: measure the second-order correlation function and the phase
221: coherence length of the condensate is reviewed in detail.
222: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
223: %
224: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
225: \section{Phase fluctuating condensates}
226: In this section, we present the phase operator of a highly
227: elongated condensate \cite{Petrov01} and develop an analytic
228: description of the ballistic expansion of the fluctuating phase
229: pattern. These results, when combined with the free evolution of
230: density modulations presented in \cite{Dettmer01,Hellweg01},
231: provide a full understanding of the order parameter dynamics
232: during the time-of-flight.
233: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
234: \subsection{Phase operator}
235: In the following, we consider a cylindrically symmetric condensate
236: in the Thomas-Fermi regime, where the repulsive mean-field
237: interaction exceeds the radial ($\hbar \omega_{\rho}$) and the
238: axial ($\hbar \omega_z$) trap energies. At $T=0$, the density
239: profile has the well-known shape $n_0(\rho,z)=n_{0\mbox{\small m}}
240: (1-\rho^2/R^2-z^2/L^2)$, where $n_{0\mbox{\small m}}=\mu/g$
241: denotes the maximum density of the condensate, $\mu$ is the
242: chemical potential, $g=4\pi\hbar^2a/m$ the interaction constant,
243: $m$ the atomic mass, and $a>0$ the scattering length. Under the
244: condition $\omega_{\rho}\gg\omega_z$, the radial size of the
245: condensate, given by the Thomas-Fermi radius
246: $R=(2\mu/m\omega_{\rho}^2)^{1/2}$, is much smaller than the axial
247: size, which corresponds to the Thomas-Fermi length
248: $L=(2\mu/m\omega_z^2)^{1/2}$.
249:
250: Due to the repulsive mean-field energy, density fluctuations are
251: strongly suppressed in a trapped BEC. Therefore, the field
252: operator describing the condensate can be written in the form
253: $\hat\psi({\bf r})=\sqrt{n_0({\bf r})}\exp(i\hat\phi({\bf r}))$,
254: where the phase operator is defined by (see e.g.
255: Ref.~\cite{Shevchenko92})
256: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
257: \begin{equation}
258: \hat\phi({\bf r})=[4n_0({\bf
259: r})]^{-1/2}\sum_{j=1}^{\infty}f_j^{+}({\bf r})\hat a_j
260: +\mbox{h.c.}. \label{operphi}
261: \end{equation}
262: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
263: Here $\hat{a}_j$ represents the annihilation operator of the
264: quasiparticle excitation with quantum number $j$ and energy
265: $\epsilon_j$; $f_j^{+}= u_j + v_j$ is the sum of the excitation
266: wavefunctions $u_j$ and $v_j$, obtained from the corresponding
267: Bogoliubov-de Gennes (BdG) equations. The low-energy axial modes,
268: which are responsible for the long wavelength axial phase
269: fluctuations, have the energy spectrum
270: $\epsilon_j=\hbar\omega_z\sqrt{j(j+3)/4}$ \cite{Stringari98}. The
271: wavefunctions $f_j^+$ of these quasiparticle modes have the form
272: \cite{Petrov01}
273: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
274: \begin{equation}
275: f_j^{+}({\bf r})=\sqrt{\frac{(j+2)(2j+3)gn_0({\bf r})} {4\pi
276: (j+1)R^2 L\epsilon_j}} P_j^{(1,1)}\left(\frac{z}{L}\right),
277: \label{fpm}
278: \end{equation}
279: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
280: where $P_j^{(1,1)}$ are Jacobi polynomials.
281: Equations~(\ref{operphi}) and (\ref{fpm}) show that the phase
282: operator only depends on the axial coordinate $z$. In sec.~III, we
283: analyze the coherence properties of the condensate by studying the
284: correlation functions of the operator $\hat\psi({\bf r})$.
285: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
286: \subsection{Evolution of the phase fluctuating pattern}
287: Starting from the results presented in
288: Refs.~\cite{Dettmer01,Hellweg01}, we analyze the evolution of
289: phase fluctuations during the free expansion of the degenerate
290: Bose gas. Since the trap is highly elongated, we can assume the
291: condensate as an infinite cylinder, and use the local density
292: approximation. The time-of-flight dynamics of the order parameter
293: is described by the scaling law \cite{Kagan96,Castin96}
294: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
295: \begin{equation}
296: \psi(\rho,z,t)= \frac{\kappa(\tilde{\rho},z,t)}{\lambda_\rho(t)}
297: e^{i\frac{m\dot\lambda_\rho}{2\hbar\lambda_\rho}\rho^2}
298: e^{-i\frac{\mu\tilde{t}}{\hbar}}, \label{sca}
299: \end{equation}
300: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
301: where $(m\dot\lambda_\rho/2\hbar\lambda_\rho)\rho^2$ is the
302: quadratic phase associated with the expansion dynamics,
303: $\lambda_\rho^2(t)=1+\omega_\rho^2 t^2$ is the scaling
304: coefficient, $\tilde{t}=\int^t dt'/\lambda_{\rho}(t')^2$ is the
305: re-scaled time, and $\tilde{\rho}=\rho/\lambda_\rho(t)$ is the
306: re-scaled radial coordinate. Let $\kappa_0=\sqrt{n_0}$ be the
307: solution of the following equation
308: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
309: \begin{equation}
310: \left [ -\frac{\hbar^2}{2m}\nabla_{\tilde{\rho}}^2+
311: \frac{m\omega_\rho^2}{2}\tilde{\rho}^2 +g|\kappa_0|^2 -
312: \mu\right]\kappa_0=0.
313: \end{equation}
314: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
315: If we define $\kappa=\sqrt{n}\exp(i\phi)$, with $n=n_0+\delta n$,
316: and substitute the scaling law of Eq.~(\ref{sca}) into the
317: corresponding Gross-Pitaevskii equation (GPE), after linearizing
318: in $\delta n$ and $\phi$ we obtain:
319: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
320: \begin{eqnarray}
321: \frac{\partial (\delta n)}{\partial t}&=&
322: \frac{\hat\xi\phi}{\lambda_\rho^2(t)}-
323: \frac{\hbar}{m}\frac{\partial^2}{\partial z^2}(n_0\phi),
324: \label{BdG1} \\
325: \frac{\partial(n_0\phi)}{\partial t}&=&-\frac{\hat\xi(\delta
326: n/n_0)}{4\lambda_\rho^2(t)}+
327: \frac{\hbar}{4m}\frac{\partial^2}{\partial z^2}(\delta n) -
328: \frac{gn_0}{\hbar \lambda_\rho^2(t)}(\delta n), \label{BdG2}
329: \end{eqnarray}
330: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
331: where $\hat\xi=-(\hbar/m)[ n_0 \nabla_{\tilde{\rho}}^2
332: +\nabla_{\tilde{\rho}}n_0\nabla_{\tilde{\rho}}]$. The first term
333: on the right hand side of Eq.~(\ref{BdG2}) can be neglected in the
334: Thomas-Fermi regime. Following Ref.~\cite{Stringari98}, we average
335: over the radial coordinates. Let $n_I$ be the radially-integrated
336: unperturbed density, and $\delta n_I$ the radially-integrated
337: density fluctuations. From Eq.~(\ref{BdG2}) we obtain:
338: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
339: \begin{eqnarray}
340: \phi(\tilde{z},\tau)&=&\phi(\tilde{z},0)+\frac{1}{8\lambda^2\zeta}
341: \frac{\partial^2}{\partial \tilde{z}^2} \left [ \int_0^\tau
342: \frac{\delta n_I(\tilde{z},\tau')} {n_I(\tilde{z},\tau')} d\tau' \right ] \nonumber \\
343: &-& \frac{\zeta}{2}\int_0^\tau \frac{1}{\lambda_\rho^2(\tau')}
344: \frac{\delta n_I(\tilde{z},\tau')}{n_I(\tilde{z},\tau')} d\tau',
345: \label{phit}
346: \end{eqnarray}
347: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
348: with $\tau=\omega_\rho t$, $\tilde{z}=z/L$,
349: $\zeta=\mu/\hbar\omega_\rho$, and $\lambda=\omega_\rho/\omega_z$.
350: Equation~(\ref{phit}) can be evaluated from the known expression
351: \cite{Dettmer01}
352: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
353: \begin{equation}
354: \frac{\delta n_I(\tilde{z},\tau)}{n_I(\tilde{z},\tau)}=\sum_j c_j
355: P_j^{(1,1)}(\tilde{z})\sin\left(
356: \frac{a_j\tau}{1-\tilde{z}^2}\right)\tau^{-b_j},
357: \label{dn}
358: \end{equation}
359: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
360: where $b_j=(\epsilon_j/\hbar\omega_\rho)^2$, $a_j=b_j/\zeta$ and
361: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
362: \begin{equation}
363: c_j=\left [ \frac{(j+2)(2j+3)g}{4\pi R^2 L \epsilon_j (j+1)}
364: \right ]^{1/2} \frac{(\alpha_j+\alpha_j^{\ast})}{2}.
365: \label{ccoeff}
366: \end{equation}
367: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
368: $\alpha_j$ and $\alpha_j^{\ast}$ are random variables with a zero
369: mean value and $\langle |\alpha_j|^2 \rangle =N_j$, $N_j$ being
370: the occupation of the quasiparticle mode $j$. Near the trap
371: center, $\delta n_I/n_I\simeq\sum_j c_j P_j^{(1,1)}(\tilde{z})
372: \sin (a_j\tau)\tau^{-b_j}$, and hence
373: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
374: \begin{eqnarray}
375: \phi(\tilde{z},\tau)=&&\phi(\tilde{z},0) \nonumber \\
376: &&+\sum_j c_j \left \{
377: \frac{(j+3)(j+4)}{32\zeta\lambda^2}P_{j-2}^{(3,3)}(\tilde{z})
378: \int_0^\tau d\tau' \sin(a_j\tau')(\tau')^{-b_j} \right\delimiter 0 \nonumber \\
379: &&- \left\delimiter 0 \frac{\zeta}{2}P_j^{(1,1)}(\tilde{z})
380: \int_0^\tau d\tau' \frac{\sin(a_j\tau')(\tau')^{-b_j}}{1+\tau'^2}
381: \right \}.
382: \label{fullphase}
383: \end{eqnarray}
384: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
385: For large $\lambda$ and sufficiently short times-of-flight, the
386: significant contribution to the phase fluctuations is due to the
387: modes $j$ such that $\tau<<\lambda^2\zeta/[j(j+3)/4]$, and
388: $b_j=j(j+3)/4\lambda^2 \ll 1$. Then, using Eq.~(\ref{operphi}) for
389: $\phi(\tilde{z},0)$, we obtain:
390: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
391: \begin{equation}
392: \phi(\tilde{z},\tau)\simeq\sum_j c_j \left
393: \{1-\frac{1}{2}\arctan(\tau) \frac{j(j+3)}{4\lambda^2} \right
394: \}P_j^{(1,1)}(\tilde{z}).
395: \end{equation}
396: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
397: The second term in the brackets is the correction to the phase
398: contribution of the $j$-th mode due to the ballistic expansion.
399: For typical times-of-flight (tens of milliseconds), this
400: correction term is very small ($\simeq10^{-5}$) and the phase
401: pattern can be assumed as completely frozen. Using
402: Eq.~(\ref{fullphase}), we have verified that, for our typical
403: experimental parameters (see Sec.~IV-C), the phase change due to
404: the free evolution of the condensate is less than $\pi/10$.
405: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
406: %
407: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
408: \section{Correlation functions of a phase fluctuating condensate}
409: The coherence properties of a condensate are described by the
410: correlation functions of the field operator $\hat{\psi}$. The
411: importance of correlation functions becomes clear if we consider
412: that most experimental signals can be modelled by using this
413: formalism. For example, the first and second order correlation
414: functions, describing the single-particle and two-particle
415: correlation properties of the system, are connected to the
416: visibility of fringes in an interference experiment and to the
417: two-body collision rate in the condensate, respectively.
418:
419: As discussed in Ref.~\cite{Petrov01}, the single-particle
420: correlation function of a highly elongated degenerate Bose gas can
421: be expressed in terms of the mean square fluctuations of the
422: phase:
423: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
424: \begin{equation}
425: \langle\hat{\psi}^\dagger({\bf r}_1)\hat{\psi}({\bf r}_2)\rangle=
426: \sqrt{n_0({\bf r}_1)n_0({\bf r}_2)}
427: \exp{\{-\langle[\delta\hat{\phi}({\bf r}_1,{\bf
428: r}_2)]^2\rangle/2\}},
429: \label{C_Eq_1}
430: \end{equation}
431: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
432: where $\delta\hat{\phi}({\bf r}_1,{\bf r}_2)=\hat{\phi}({\bf
433: r}_1)- \hat{\phi}({\bf r}_2)$ depends directly on the phase
434: operator $\hat{\phi}$ given in Eqs.~(\ref{operphi}). At
435: equilibrium, the population of the $j$-th quasiparticle mode,
436: $\langle\hat{a}_j^\dagger\hat{a}_j\rangle$, is a random variable
437: with mean value $N_j$, given by the Bose-Einstein distribution
438: function. The appearance of phase fluctuations is a stochastic
439: process governed by the temperature $T$ of the system. Since
440: individual realizations are not predictable, we average over an
441: ensemble of identically prepared condensates in thermal
442: equilibrium at temperature $T$. This average is indicated by
443: $\langle\dots\rangle_T$. When $k_BT\gg\hbar\omega_z$ ($k_B$ is the
444: Boltzmann constant), the population of the $j$-th mode is
445: $N_j\simeq k_BT/\epsilon_j$, and the thermal average of the mean
446: square fluctuations of the phase becomes
447: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
448: \begin{equation}
449: \langle[\delta\hat{\phi}(z_1,z_2)]^2\rangle_T=
450: \delta_L^2(T)f^{(1)}(z_1/L,z_2/L),
451: \label{C_Eq_2}
452: \end{equation}
453: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
454: where
455: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
456: \begin{equation}
457: \delta_L^2(T)=\frac{32\mu k_BT}{15N_0(\hbar\omega_z)^2}
458: \label{C_Eq_3}
459: \end{equation}
460: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
461: and
462: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
463: \begin{equation}
464: f^{(1)}(z_1/L,z_2/L)=
465: \frac{1}{8}\sum_{j=1}^{\infty}\frac{(j+2)(2j+3)}
466: {j(j+1)(j+3)}\left[P_j^{(1,1)}\left(\frac{z_1}{L}\right)-
467: P_j^{(1,1)}\left(\frac{z_2}{L}\right)\right]^2, \label{C_Eq_4}
468: \end{equation}
469: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
470: $N_0$ indicating the number of atoms in the condensate fraction.
471: The first order correlation function of the degenerate Bose gas is
472: defined by (see e.g. \cite{Scully97})
473: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
474: \begin{equation}
475: g^{(1)}_T({\bf r}_1,{\bf r}_2)=
476: \frac{\langle\hat{\psi}^\dagger({\bf r}_1)\hat{\psi}({\bf
477: r}_2)\rangle_T} {(\langle\hat{\psi}^\dagger({\bf
478: r}_1)\hat{\psi}({\bf r}_1)\rangle_T \langle\hat{\psi}^\dagger({\bf
479: r}_2) \hat{\psi}({\bf r}_2)\rangle_T)^{1/2}}.
480: \label{C_Eq_6}
481: \end{equation}
482: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
483: According to Eqs.~(\ref{C_Eq_1}) and (\ref{C_Eq_2}), this results
484: in
485: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
486: \begin{equation}
487: g^{(1)}_T(z_1,z_2)=
488: \exp\{-\delta_L^2(T)f^{(1)}(z_1/L,z_2/L)/2\}.
489: \label{C_Eq_7}
490: \end{equation}
491: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
492: For $|z_1|,|z_2|\ll L$, using the asymptotic expression of the
493: Jacobi Polynomials \cite{Gradsteyn65}, and summing over the
494: different modes in the continuous limit, one obtains an
495: approximated formula for the $f^{(1)}$ function valid around the
496: center of the condensate \cite{Petrov01}:
497: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
498: \begin{equation}
499: f^{(1)}(z_1/L,z_2/L)=|z_1-z_2|/L.
500: \label{C_Eq_5}
501: \end{equation}
502: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
503: In that case,
504: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
505: \begin{equation}
506: g^{(1)}_T(z_1,z_2)=
507: \exp\{-\delta_L^2(T)|z_1-z_2|/2L\}.
508: \label{C_Eq_8}
509: \end{equation}
510: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
511: This result suggests the introduction of the phase coherence
512: length of the condensate
513: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
514: \begin{equation}
515: L_{\phi}=\frac{L}{\delta_L^2(T)}, \label{C_Eq_9}
516: \end{equation}
517: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
518: defined as the distance at which the first order correlation
519: function decreases to $1/\sqrt{e}$. The approximate formula shown
520: in Eq.~(\ref{C_Eq_5}) can be extended to describe the behavior of
521: the $f^{(1)}$ function far from the center of the condensate. For
522: $\delta_L^2(T)\gg1$, the coherence length $L_{\phi}$ is small
523: compared to the axial size $L$, and the system is well described
524: by means of the local density approximation
525: \cite{Dettmer01,Hellweg01,Gerbier02a}. As pointed out in Ref.\
526: \cite{Gerbier02a}, this limit is equivalent to the use of the
527: approximate formula for the Jacobi polynomials with large $j$
528: \cite{Gradsteyn65}. Equation~(\ref{C_Eq_4}) can thus be written in
529: the form
530: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
531: \begin{equation}
532: f^{(1)}(z_1/L,z_2/L)=\frac{|z_1-z_2|/L}{[1-(z_1+z_2)^2/(2L)^2]^2},
533: \label{C_Eq_10}
534: \end{equation}
535: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
536: generalizing the result obtained in Eq.~(\ref{C_Eq_5}).
537:
538: We use a similar approach to calculate the two-particle
539: correlation function of the condensate. Introducing the operator
540: $\delta^{(2)}\hat{\phi}({\bf r}_1,{\bf r}_2, {\bf r}_3,{\bf r}_4)=
541: \hat{\phi}({\bf r}_1)+\hat{\phi}({\bf r}_2)- \hat{\phi}({\bf
542: r}_3)-\hat{\phi}({\bf r}_4)$, we obtain
543: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
544: \begin{equation}
545: \langle\hat{\psi}^\dagger({\bf r}_1) \hat{\psi}^\dagger({\bf r}_2)
546: \hat{\psi}({\bf r}_3)\hat{\psi}({\bf r}_4)\rangle=
547: \prod_{i=1}^4\sqrt{n_0({\bf r}_i)} \exp{\{-\langle[
548: \delta^{(2)}\hat{\phi}({\bf r}_1,{\bf r}_2,{\bf r}_3, {\bf
549: r}_4)]^2\rangle/2\}}.
550: \label{C_Eq_11}
551: \end{equation}
552: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
553: Using Eq.~(\ref{operphi}) for the phase operator, a
554: straightforward calculation yields
555: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
556: \begin{eqnarray}
557: \langle[\delta^{(2)}\hat{\phi}(z_1,z_2,z_3,z_4)]^2\rangle=&&
558: \sum_{j=1}^{\infty}
559: \frac{(j+2)(2j+3)\mu}{15(j+1)\epsilon_jN_0}N_j \nonumber\\
560: &&\times\left[P_j^{(1,1)}\left(\frac{z_1}{L}\right)
561: +P_j^{(1,1)}\left(\frac{z_2}{L}\right)\right. \nonumber\\
562: &&\left.-P_j^{(1,1)}\left(\frac{z_3}{L}\right)-
563: P_j^{(1,1)}\left(\frac{z_4}{L}\right)\right]^2.
564: \label{C_Eq_12}
565: \end{eqnarray}
566: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
567: In the limit $k_BT\gg\hbar\omega_z$, the thermal average of
568: Eq.~(\ref{C_Eq_12}) gives
569: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
570: \begin{equation}
571: \langle[\delta^{(2)}\hat{\phi}(z_1,z_2,z_3,z_4)]^2\rangle_T=
572: \delta_L^2(T)f^{(2)}(z_1/L,z_2/L,z_3/L,z_4/L),
573: \label{C_Eq_13}
574: \end{equation}
575: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
576: where
577: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
578: \begin{eqnarray}
579: f^{(2)}(z_1/L,z_2/L,z_3/L,z_4/L)=&&
580: f^{(1)}(z_1/L,z_3/L)+f^{(1)}(z_2/L,z_4/L) \nonumber \\
581: &&-f^{(1)}(z_1/L,z_2/L)-f^{(1)}(z_3/L,z_4/L) \nonumber \\
582: &&+f^{(1)}(z_1/L,z_4/L)+f^{(1)}(z_2/L,z_3/L).
583: \label{C_Eq_15}
584: \end{eqnarray}
585: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
586: Thus, the two-particle correlation function can be expressed as a
587: product of one-particle correlation functions.
588: Equations~(\ref{C_Eq_5}) and (\ref{C_Eq_10}) can be used to derive
589: simplified expressions for the $f^{(2)}$ function, valid in the
590: limit $|z_i|\ll L$ ($i=1,\ldots,4$) and in the local density
591: approximation. Figure~\ref{C_Fig_1} shows the dependence of
592: $f^{(2)}$ calculated in
593: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
594: \begin{equation}
595: \bar{z}_1=\frac{d+s}{2},\quad \bar{z}_2= \frac{-d-s}{2},\quad
596: \bar{z}_3=\frac{-d+s}{2},\quad \bar{z}_4= \frac{d-s}{2}
597: \label{C_Eq_16}
598: \end{equation}
599: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
600: as a function of $s>0$.
601: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
602: \begin{figure}
603: \includegraphics{Fig1}
604: \caption{\label{C_Fig_1}$f^{(2)}(\bar{z}_1/L,\bar{z}_2/L,\bar{z}_3/L,\bar{z}_4/L)$
605: as a function of $s>0$. The complete expression in
606: Eq.~(\ref{C_Eq_15}) (solid line) is compared with the approximated
607: formulas derived from Eqs.~(\ref{C_Eq_5}) and (\ref{C_Eq_10}),
608: valid in the condensate center (dotted line) and in the local
609: density approximation (dashed line). The inset shows $f^{(2)}$ for
610: different values of $d>0$.}
611: \end{figure}
612: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
613: The full expression of $f^{(2)}$ can be compared with the two
614: approximated formulas, the first valid in the condensate center,
615: the second valid in the local density approximation. The inset of
616: Fig.~\ref{C_Fig_1} shows the same curves for different values of
617: $d>0$. This choice of variables follows the particular
618: experimental realization. In Sec.~IV, we demonstrate how these
619: curves can be measured in a matter wave interferometry experiment.
620: There, $d$ is the displacement between the two interfering
621: condensate copies, and $s$ is the separation between the positions
622: in the interference pattern at which the particle densities are
623: evaluated.
624:
625: A qualitative understanding of the behavior shown in
626: Fig.~\ref{C_Fig_1} is possible if we consider that
627: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
628: \begin{eqnarray}
629: \langle[\delta^{(2)}\hat{\phi}(\bar{z}_1,\bar{z}_2,
630: \bar{z}_3,\bar{z}_4)]^2\rangle_T=&&
631: \langle[\delta\hat{\phi}(\bar{z}_1,\bar{z}_3)]^2\rangle_T+
632: \langle[\delta\hat{\phi}(\bar{z}_2,\bar{z}_4)]^2\rangle_T \nonumber \\
633: &&+2\langle\delta\hat{\phi}(\bar{z}_1,\bar{z}_3)
634: \delta\hat{\phi}(\bar{z}_2,\bar{z}_4)\rangle_T.
635: \label{C_Eq_17}
636: \end{eqnarray}
637: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
638: The first and the second term are the thermal averages of the
639: operator $(\delta\hat{\phi})^2$ calculated in
640: $(\bar{z}_1,\bar{z}_3)$ and in $(\bar{z}_2,\bar{z}_4)$; the last
641: term is proportional to the correlation function of
642: $\delta\hat{\phi}$ at the same coordinates. For a fixed
643: displacement $d$, when the examined positions are close to the
644: condensate center ($d,s \ll L$), the first two terms of
645: Eq.~(\ref{C_Eq_17}) do not depend on the separation $s$. However,
646: as $s$ rises from $0$ to $d$, the third term increases from
647: $-2\langle[\delta\hat{\phi}(\bar{z}_1,\bar{z}_3)]^2\rangle_T$
648: (complete anticorrelation) to its maximum value $0$, resulting in
649: an uncorrelated phase difference for every $s\geq d$. In the
650: interval $0\leq s\leq d$, the $f^{(2)}$ function depends linearly
651: on $s$ with slope 2.
652:
653: The second order correlation function is defined as
654: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
655: \begin{equation}
656: g^{(2)}_T({\bf r}_1,{\bf r}_2,{\bf r}_3,{\bf r}_4)=
657: \frac{\langle\hat{\psi}^\dagger({\bf r}_1) \hat{\psi}^\dagger({\bf
658: r}_2) \hat{\psi}({\bf r}_3)\hat{\psi}({\bf r}_4)\rangle_T}
659: {(\langle\hat{\psi}^\dagger({\bf r}_1) \hat{\psi}({\bf
660: r}_1)\rangle_T\ldots \langle\hat{\psi}^\dagger({\bf r}_4)
661: \hat{\psi}({\bf r}_4)\rangle_T)^{1/2}}.
662: \label{C_Eq_18}
663: \end{equation}
664: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
665: Substituting Eqs.~(\ref{C_Eq_11}) and (\ref{C_Eq_13}) in
666: Eq.~(\ref{C_Eq_18}), we obtain:
667: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
668: \begin{equation}
669: g^{(2)}_T(z_1,z_2,z_3,z_4)=
670: \exp\{-\delta_L^2(T)f^{(2)}(z_1/L,z_2/L,z_3/L,z_4/L)/2\}.
671: \label{C_Eq_19}
672: \end{equation}
673: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
674: Note that, due to the suppression of density modulations, the
675: normalized density correlation function of the trapped condensate
676: is constant: $g^{(2)}_T(z_1,z_2,z_2,z_1)=1$.
677:
678: The calculation we have described for the second order correlation
679: function can be extended to obtain a general expression for the
680: $N$-th order correlation function. Defining the operator
681: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
682: \begin{equation}
683: \delta^{(N)}\hat{\phi}(\{{\bf r}_i\}_{i=1,\ldots,2N})=
684: \hat{\phi}({\bf r}_1)+\ldots+\hat{\phi}({\bf r}_N)-
685: \hat{\phi}({\bf r}_{N+1})-\ldots-\hat{\phi}({\bf r}_{2N}),
686: \label{C_Eq_20}
687: \end{equation}
688: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
689: the $N$-particle correlation function is given by
690: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
691: \begin{equation}
692: \langle\hat{\psi}^\dagger({\bf r}_1)\ldots
693: \hat{\psi}^\dagger({\bf r}_N)
694: \hat{\psi}({\bf r}_{N+1})\ldots
695: \hat{\psi}({\bf r}_{2N})\rangle=
696: \prod_{i=1}^N\sqrt{n_0({\bf r}_i)}
697: \exp{\{-\langle[\delta^{(N)}\hat{\phi}
698: (\{{\bf r}_i\}_{i=1,\ldots,2N})]^2
699: \rangle/2\}}.
700: \label{C_Eq_21}
701: \end{equation}
702: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
703: In general, the thermal average of the operator
704: $(\delta^{(N)}\hat{\phi})^2$ can be written in the form
705: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
706: \begin{equation}
707: \langle[\delta^{(N)}\hat{\phi}
708: (\{{\bf r}_i\}_{i=1,\ldots,2N})]^2
709: \rangle_T=\delta_L^2(T)f^{(N)}
710: (\{z_i/L\}_{i=1,\ldots,2N}).
711: \label{C_Eq_22}
712: \end{equation}
713: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
714: The $f^{(N)}$ function, depending on the Jacobi polynomials
715: $P_j^{(1,1)}$, can be expressed as a combination of $f^{(1)}$
716: functions:
717: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
718: \begin{equation}
719: f^{(N)}(\{z_i/L\}_{i=1,\ldots,2N})=\sum_{1\leq l<m\leq2N}
720: \mathcal{P}^{\{l,m\}}
721: f^{(1)}\left(\frac{z_l}{L},\frac{z_m}{L}\right), \label{C_Eq_23}
722: \end{equation}
723: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
724: where the coefficient $\mathcal{P}^{\{l,m\}}$ is defined as
725: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
726: \begin{equation}
727: \mathcal{P}^{\{l,m\}}=\left\{
728: \begin{array}{l}
729: +1 \quad \textrm{if $l\leq N<m$} \\
730: -1 \quad \textrm{if $l,m\leq N$ or $l,m> N$}
731: \end{array}
732: \right..
733: \label{C_Eq_24}
734: \end{equation}
735: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
736: The $N$-th order correlation function is given by
737: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
738: \begin{equation}
739: g^{(N)}_T(\{{\bf r}_i\}_{i=1,\ldots,2N})=
740: \frac{\langle\hat{\psi}^\dagger({\bf r}_1)\ldots
741: \hat{\psi}^\dagger({\bf r}_N) \hat{\psi}({\bf r}_{N+1})\ldots
742: \hat{\psi}({\bf r}_{2N})\rangle_T}
743: {(\langle\hat{\psi}^\dagger({\bf r}_1) \hat{\psi}({\bf
744: r}_1)\rangle_T\ldots \langle\hat{\psi}^\dagger({\bf r}_{2N})
745: \hat{\psi}({\bf r}_{2N})\rangle_T)^{1/2}}
746: \label{C_Eq_25}
747: \end{equation}
748: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
749: and, from Eqs.~(\ref{C_Eq_21}) and (\ref{C_Eq_22}),
750: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
751: \begin{equation}
752: g^{(N)}_T(\{z_i\}_{i=1,\ldots,2N})=
753: \exp\{-\delta_L^2(T)f^{(N)}(\{z_i/L\}_{i=1,\ldots,2N})/2\}.
754: \label{C_Eq_26}
755: \end{equation}
756: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
757: This general result shows that the spatial correlation function of
758: phase fluctuating condensates is completely characterized by the
759: parameter $\delta^2_L(T)$ and, therefore, by the phase coherence
760: length $L_{\phi}$.
761: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
762: %
763: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
764: \section{Interferometric measurement of the second order correlation function}
765: The coherence of a matter wave can be studied by using
766: interferometric methods. However, as standard interference
767: experiments measure the first order correlation function of the
768: field operator $\hat{\psi}$, they are very sensitive to phase
769: noise introduced by the experimental apparatus. The method
770: presented here is analogous to the original Hanbury-Brown and
771: Twiss experiment \cite{Hanbury56,Hanbury56a} in which the
772: spatially resolved second order correlation function $g^{(2)}({\bf
773: r}_1,{\bf r}_2,{\bf r}_2,{\bf r}_1)$ of a light source is obtained
774: from intensity correlation measurements. As discussed before, for
775: a highly elongated BEC $g^{(2)}_T(z_1,z_2,z_2,z_1)=1$. This result
776: suggests that a simple measurement of density correlations in the
777: condensate is not sufficient to describe the coherence properties
778: of the sample. Nevertheless, by measuring density correlations in
779: the interference pattern generated by two spatially displaced
780: copies of a parent BEC, it is possible to correlate the field
781: operator $\hat{\psi}$ at four different positions and extract
782: $g^{(2)}_T(z_1,z_2,z_3,z_4)$ directly. Compared to standard
783: interference experiments, the main advantage of this technique is
784: the intrinsic stability of the density correlation measurement
785: against variations of the global phase between the interfering
786: condensates.
787:
788: In this section, we show how a matter wave Bragg interferometer
789: can be used to characterize the second order correlation function
790: of the condensate and measure its phase coherence length.
791: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
792: %
793: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
794: \subsection{Interferometric scheme}
795: Our interferometric sequence is shown in Fig.~\ref{I_Fig_1}.
796: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
797: \begin{figure}
798: \includegraphics[width=0.60\textwidth]{Fig2.eps}
799: \caption{\label{I_Fig_1}a) Matter wave Bragg interferometer. The
800: condensate is released from the magnetic trap and evolves freely
801: for $2\,\textrm{ms}$. The sample is interrogated by the first
802: $\pi/2$ Bragg pulse which splits the parent BEC in two copies with
803: momenta $0$ and $2\hbar k$. After a time $\Delta t$, the second
804: $\pi/2$ Bragg pulse splits the condensates again and allows them
805: to interfere. The time interval $\Delta t$ defines the
806: displacement $d$ between the two interfering condensates. b) A
807: typical line density profile at the output ports of the
808: interferometer. The distance between the two autocorrelated copies
809: ($d=46\,\mu\textrm{m}$) is comparable to the phase coherence
810: length of the parent condensate ($L_{\phi}=43\,\mu\textrm{m}$).In
811: the schematic of the matter wave Bragg interferometer, the
812: distance $d$ has been exaggerated for clarity.}
813: \end{figure}
814: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
815: The condensate is released from the magnetic trap and expands
816: freely for $2\,\textrm{ms}$. This short time-of-flight is
817: important to lower the density, thus reducing $s$-wave scattering
818: processes occurring during the Bragg diffraction of the condensate
819: \cite{Chikkatur00}. The interrogation sequence consists of two
820: $\pi/2$ Bragg pulses. Each pulse is composed of two
821: counterpropagating laser beams of wave number $k$, detuned from
822: the atomic transition. The first Bragg pulse splits the condensate
823: in the two momentum eigenstates $|2\hbar k\rangle$ and $|0\rangle$
824: along the axial direction ($z$). After a time $\Delta t$, a second
825: $\pi/2$ pulse splits the condensates again, creating two
826: interfering copies in each momentum state. The time interval
827: $\Delta t$ between the two pulses sets the spatial overlap,
828: $d=2\hbar k\Delta t/m$, between the interfering BECs at the output
829: ports of the interferometer. The relative phase of the two
830: counterpropagating Bragg beams is externally controlled by an
831: electro-optic modulator ($\textrm{EOM}$) and can be changed
832: between the two pulses. This allows us to imprint an extra phase
833: ${\varphi}$ which can be precisely tuned. Control of the
834: $\textrm{EOM}$ phase is crucial for our method, as described in
835: Sec.~IV-B.
836:
837: Using the results derived in Sec.~II, the atoms detected in the
838: output port A (Fig.~\ref{I_Fig_1}), after a total time-of-flight
839: $t$, are described by the order parameter
840: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
841: \begin{equation}
842: \psi({\bf r},d,t)=\frac{1}{2}\sqrt{\eta({\bf
843: r}',t)}+\frac{1}{2}\sqrt{\eta({\bf r},t)}
844: \exp\{i[\delta\phi(z,z',t)+\alpha(z,z',t)+\beta(z,z')+\gamma(d)]\},
845: \label{I_Eq_1}
846: \end{equation}
847: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
848: where ${\bf r}'={\bf r}-d\,{\bf\hat{z}}$ and $\eta({\bf r},t)$ is
849: the time-evolved density profile normalized to the total number of
850: atoms in the parent condensate. The relative phase between the
851: interfering condensates contains several contributions.
852: $\delta\phi(z,z',t)=\phi(z,t)-\phi(z',t)$ describes the phase
853: difference between $z$ and $z'$ that evolves from the phase
854: fluctuations in the parent condensate. The term
855: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
856: \begin{equation}
857: \alpha(z,z',t)=\frac{m\dot{\lambda}_z}{2\hbar\lambda_z}
858: (z^2-z'\,^2) \label{I_Eq_2}
859: \end{equation}
860: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
861: represents the non-uniform spatial phase profile developed during
862: the mean-field-driven expansion. The mean-field gradient between
863: the interfering BECs is responsible for a force repelling the
864: centers of mass of the two clouds. This effect is described by the
865: phase term
866: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
867: \begin{equation}
868: \beta(z,z')=\frac{m\delta v}{2\hbar}(z+z'),
869: \label{I_Eq_3}
870: \end{equation}
871: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
872: proportional to the relative repulsion velocity $\delta v$ between
873: the interfering condensates \cite{Simsarian00}. After the first
874: Bragg pulse the relative phase of the atoms in the $|2\hbar
875: k\rangle$ momentum state evolves with a characteristic frequency
876: $\delta_{\mbox{\small Bragg}}$, given by the detuning of the
877: lasers from the resonance of the two-photon transition
878: \cite{Weiss94}. Therefore, the last term,
879: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
880: \begin{equation}
881: \gamma(d)=\delta_{\mbox{\small Bragg}}\Delta t+\varphi=
882: \delta_{\mbox{\small Bragg}}\frac{m d}{2\hbar k}+\varphi,
883: \label{I_Eq_4}
884: \end{equation}
885: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
886: represents a global phase depending on the detuning from the Bragg
887: transition and the externally controlled phase $\varphi$.
888:
889: The density of atoms at the output port A of the interferometer is
890: given by
891: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
892: \begin{eqnarray}
893: I({\bf r},d,t)=&&\frac{1}{4}\eta({\bf r},t)+
894: \frac{1}{4}\eta({\bf r}',t) \nonumber\\
895: &&+\frac{1}{2}\sqrt{\eta({\bf r},t)\eta({\bf r}',t)}
896: \cos[\delta\phi(z,z',t)+\alpha(z,z',t)+\beta(z,z')+\gamma(d)].
897: \label{I_Eq_5}
898: \end{eqnarray}
899: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
900: The presence of strong phase fluctuations alters the interference
901: pattern generated by the two autocorrelated condensates. In fact,
902: when $d\simeq L_{\phi}$ the phase term $\delta\phi$ can be
903: comparable to $\pi$, modifying drastically and in an unpredictable
904: way the position and the spacings of the interference fringes.
905: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
906: %
907: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
908: \subsection{Method}
909: Starting from Eq.~(\ref{I_Eq_5}), we want to calculate the density
910: correlation function of the interference pattern, for an ensemble
911: of identically prepared condensates at a given temperature $T$,
912: averaged over all the global phase values $\varphi$. This
913: averaging process is indicated by the symbol
914: $\langle\ldots\rangle_{T,\,\varphi}$. It is therefore important
915: that the phase delay $\varphi$ induced by the $\textrm{EOM}$ is
916: uniformly changed between $0$ and $2\pi$. In Sec.~II, we have
917: shown that, for typical times-of-flight (tens of milliseconds),
918: the evolution of the fluctuating phase of the condensate is
919: basically frozen. This allows us to neglect the time-dependence of
920: $\delta\phi(z,z',t)$. We also neglect the contribution of density
921: modulations induced by the initial phase pattern on the
922: Thomas-Fermi profile of the condensate. The validity of this
923: approximation is verified below. Under these assumptions, we
924: calculate the normalized density correlation function
925: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
926: \begin{equation}
927: \gamma^{(2)}({\bf r}_1,{\bf r}_2,d,t)= \frac{\langle(I_1-\langle
928: I_1\rangle_{T,\,\varphi})(I_2-\langle
929: I_2\rangle_{T,\,\varphi})\rangle_{T,\,\varphi}}
930: {\sqrt{\langle(I_1-\langle
931: I_1\rangle_{T,\,\varphi})^2\rangle_{T,\,\varphi}\langle(I_2-
932: \langle I_2\rangle_{T,\,\varphi})^2\rangle_{T,\,\varphi}}},
933: \label{M_Eq_1}
934: \end{equation}
935: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
936: where $I_{1,2}=I({\bf r}_{1,2},d,t)$. After a lengthy but
937: straightforward calculation, the averaging process gives
938: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
939: \begin{eqnarray}
940: \gamma^{(2)}(z_1,z_2,d,t)=&&\cos\left[\frac{m}{\hbar}\left(
941: \frac{\dot{\lambda}_z}{\lambda_z}d+\delta
942: v\right)(z_1-z_2)\right] \nonumber \\
943: &&\times\exp[-\delta_L^2(T)f^{(2)}(z_1/L,(z_2-d)/L,(z_1-d)/L,z_2/L)/2].
944: \label{M_Eq_2}
945: \end{eqnarray}
946: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
947: $\gamma^{(2)}(z_1,z_2,d,t)$ results from the product of two
948: different terms: the first is a periodic function, whose argument
949: is the contribution of the mean-field energy to the phase profile
950: (ballistic expansion and relative repulsion between the
951: interfering condensates); the second is an exponential term which
952: corresponds to the $g^{(2)}_T$ function of the parent phase
953: fluctuating condensate. The decay constant of this function is
954: given by the phase coherence length of the condensate (see
955: Eq.~\ref{C_Eq_9}).
956:
957: From the experimental point of view, the averaging process
958: described above is equivalent to the following procedure: The
959: radially integrated density profile $I=I(z,d,t)$ at the output
960: port A of the interferometer is measured for different values of
961: the global phase, uniformly distributed in the range
962: $0\leq\varphi<2\pi$; then the average value $\langle
963: I\rangle_{T,\varphi}$ is calculated and used to determine
964: $I-\langle I\rangle_{T,\varphi}$ for each experimental
965: realization. These profiles, averaged according to
966: Eq.~(\ref{M_Eq_1}), give a measurement of
967: $\gamma^{(2)}(z_1,z_2,d,t)$. We evaluate the density correlations
968: as a function of the separation $s=z_2-z_1$. For simplicity, we
969: choose symmetric positions around the center ($z=d/2$) of the
970: interference pattern in the output port A. The positions in
971: Eq.~(\ref{C_Eq_16}) are defined such that $z_1=\bar{z}_1$ and
972: $z_2=\bar{z}_4$ (see Fig.~\ref{I_Fig_1}). The method described
973: here allows us to characterize the dependence of the correlation
974: function
975: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
976: \begin{eqnarray}
977: \gamma^{(2)}(s,d,t)=&&\cos\left[\frac{m}{\hbar}\left(
978: \frac{\dot{\lambda}_z}{\lambda_z}d+\delta
979: v\right)s\right] \nonumber \\
980: &&\times\exp[-\delta_L^2(T)
981: f^{(2)}(\bar{z}_1/L,\bar{z}_2/L,\bar{z}_3/L,\bar{z}_4/L)/2]
982: \label{M_Eq_3}
983: \end{eqnarray}
984: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
985: on the separation $s$ for any fixed displacement $d$ between the
986: interfering condensates.
987: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
988: %
989: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
990: \subsection{Experimental results and numerical simulations}
991: We perform the experiment with $^{87}\textrm{Rb}$ condensates in
992: the $F=1$, $m_F=-1$ state. The atoms are confined in a highly
993: elongated magnetic trap with cylindrical symmetry, the long axis
994: lying in the horizontal plane. The confining potential has an
995: axial frequency $\omega_z=2\pi\times3.4\,\textrm{Hz}$ and a radial
996: frequency $\omega_{\rho}$ which is varied between
997: $2\pi\times300\,\textrm{Hz}$ and $2\pi\times380\,\textrm{Hz}$.
998: Further details on the experimental apparatus can be found in
999: \cite{Kreutzmann03}. After the BEC formation, we let the system
1000: thermalize in the magnetic trap for typically $4\,\textrm{s}$ in
1001: presence of radio frequency shielding \footnote{We measure the
1002: typical thermalization times of our condensates by performing
1003: experiments of BEC growth. After times of few hundreds of
1004: milliseconds the system reaches an equilibrium condition.}. That
1005: time is important to reach an equilibrium condition in which any
1006: quadrupole oscillation has been damped down. As shown in
1007: Fig.~\ref{I_Fig_1}, our matter wave interferometer consists of two
1008: $\pi/2$ Bragg diffraction pulses. Each of them is composed of two
1009: counterpropagating laser beams, detuned by about $3\,\textrm{GHz}$
1010: from the atomic transition. This detuning suppresses spontaneous
1011: scattering of photons during the interrogation time. The Bragg
1012: pulse duration of $100\,\mu\textrm{s}$ is sufficiently short not
1013: to resolve the momentum distribution of the atoms in the
1014: condensate and long enough to avoid higher order Bragg diffraction
1015: processes. A fixed frequency difference is set between the two
1016: counterpropagating beams to match the Bragg condition. The
1017: condensate is released from the magnetic trap and after
1018: $2\,\textrm{ms}$ of time-of-flight is probed by the two-pulse
1019: sequence of the interferometer. The atomic cloud is detected after
1020: the ballistic expansion by resonant absorption imaging.
1021:
1022: Figure~\ref{I_Fig_1}b shows a typical line density profile of an
1023: interference pattern where the distance between the two
1024: autocorrelated copies ($d=46\,\mu\textrm{m}$) is comparable to the
1025: phase coherence length of the parent condensate
1026: ($L_{\phi}=43\,\mu\textrm{m}$). Because of the stochastic nature
1027: of phase fluctuations, the fringe spacing is not regular and
1028: differs in each experimental realization. This experimentally
1029: demonstrates that the fluctuating phase of the condensate can
1030: significantly change on distances comparable with the phase
1031: coherence length of the sample. Even if each single image shows
1032: high contrast, the interference pattern is completely washed out
1033: when we average a significant number of realizations.
1034:
1035: The results of standard interference experiments are related to
1036: the correlations of the wavefunction and therefore are very
1037: sensitive to phase instabilities. Figure~\ref{E_Fig_1} shows the
1038: interference signal obtained by measuring the number of atoms in
1039: an interval of width $0.2\times L$ around the center of the
1040: interference pattern ($z=d/2$) at the output port $A$, as a
1041: function of the global phase $\varphi$ controlled by the
1042: $\textrm{EOM}$. This signal is normalized to the corresponding
1043: number of atoms in the parent condensate.
1044: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1045: \begin{figure}
1046: \includegraphics{Fig3}
1047: \caption{\label{E_Fig_1}The number of atoms measured in the
1048: interval $d/2-0.1\times L<z<d/2+0.1\times L$, around the center of
1049: the interference pattern detected at the output port A, is plotted
1050: as a function of the phase $\varphi$ controlled by the
1051: electro-optic modulator. The signal is normalized to the
1052: corresponding number of atoms in the parent condensate. The two
1053: sets of data correspond to different displacements $d$ between the
1054: overlapping condensates. The solid line is obtained by fitting the
1055: experimental data with a sinusoidal function. The measurements
1056: refer to condensates with about $3\times10^5$ atoms, a typical
1057: axial size of $L=180\,\mu\textrm{m}$ and a temperature
1058: $T=170\,\textrm{nK}$ .}
1059: \end{figure}
1060: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1061: The two plots correspond to different displacements $d$ between
1062: the interfering condensates. A small displacement is related to a
1063: short time interval between the two interrogation Bragg pulses. In
1064: that case, the contribution of phase fluctuations and the effect
1065: of technical phase noise introduced by the experimental apparatus
1066: are both negligible. Therefore, according to Eq.~(\ref{I_Eq_5}),
1067: when $d\ll L_{\phi}$ and $\Delta t$ is small compared to the
1068: characteristic time stability of our Bragg pulses, the normalized
1069: signal oscillates sinusoidally with high contrast. For $d$
1070: approaching $L_{\phi}$, the random phase introduced by the phase
1071: fluctuations washes out the oscillation. If external disturbances
1072: can be neglected, the contrast of the oscillations is directly
1073: related to the first order correlation function $g^{(1)}$ at a
1074: given displacement $d$. However, as $d$ increases, the external
1075: disturbances \footnote{For example, a change in the release
1076: velocity of $5.5\times10^{-3}v_{rec}$, where $v_{rec}=\hbar
1077: k/m=5.9\,\mu \textrm{m}/\textrm{ms}$ is the recoil velocity on the
1078: $D_2$ line, leads to a phase change of $\pi/2$ during a time
1079: $\Delta t=3\,\textrm{ms}$ between the Bragg pulses. This value
1080: approximately corresponds to the stability of the release velocity
1081: in our experimental apparatus.} also increase and produce a random
1082: phase noise which destroys the oscillating behavior and hides the
1083: effect of phase fluctuations on the detected signal.
1084:
1085: This problem can be solved by using the method described in
1086: Sec.~IV-B. The measurement of intensity correlations, in
1087: combination with the subsequent averaging process, has the major
1088: advantage of being insensitive to technical phase noise introduced
1089: by the experimental apparatus. Figure~\ref{E_Fig_2} shows the
1090: correlation function $\gamma^{(2)}(s,d,t)$ extracted from a set of
1091: 29 line density profiles corresponding to $5.0\times10^5$
1092: condensed atoms at a temperature $T=216\,\textrm{nK}$, detected
1093: after a total time-of-flight $t=37\,\textrm{ms}$. The displacement
1094: between the interfering BECs is $d=35\,\mu\textrm{m}$.
1095: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1096: \begin{figure}
1097: \includegraphics{Fig4}
1098: \caption{\label{E_Fig_2}Circles: Correlation function
1099: $\gamma^2(s,d,t)$ extracted from a set of 29 line density
1100: profiles. The data correspond to samples with $5.0\times10^5$
1101: condensed atoms at a temperature $T=216\,\textrm{nK}$, detected
1102: after a total time-of-flight $t=37\,\textrm{ms}$. The displacement
1103: between the interfering BECs is $d=35\,\mu\textrm{m}$. The bars on
1104: the experimental points represent the statistical errors. Crosses:
1105: Numerical simulation which takes into account the time dependence
1106: of the fluctuating phase and of density modulations, modelled on
1107: the experimental parameters. Solid line: Fit to the experimental
1108: data using the model function of Eq.~(\ref{E_Eq_1}). Dashed line:
1109: Second order correlation function $g^{(2)}_T(s,d)=
1110: g^{(2)}_T(\overline{z}_1,\overline{z}_2,\overline{z}_3,\overline{z}_4)$
1111: extracted from the fit to the experimental data. The phase
1112: coherence length of the sample is graphically indicated on the
1113: plot.}
1114: \end{figure}
1115: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1116: The experimental data is compared with a numerical simulation
1117: which produces random phase patterns according to the experimental
1118: conditions and uses Eq.~(\ref{I_Eq_1}) to describe the evolution
1119: of the order parameter. The numerically calculated points shown in
1120: Fig.~\ref{E_Fig_2} are obtained by following the same averaging
1121: procedure we have applied to the experimental data. This kind of
1122: analysis includes the time dependence of the fluctuating phase and
1123: of the density modulations induced by the initial phase pattern.
1124: The solid line is the result of a fit to the experimental data.
1125: According to Eqs.~(\ref{M_Eq_3}) and (\ref{C_Eq_16}), the model
1126: function
1127: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1128: \begin{equation}
1129: \cos(a\cdot s)\exp[-b\cdot
1130: f^{(2)}(\overline{z}_1/L,\overline{z}_2/L,
1131: \overline{z}_3/L,\overline{z}_4/L)/2] \label{E_Eq_1}
1132: \end{equation}
1133: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1134: contains only two free parameters. The curves clearly show the
1135: damped oscillating behavior. The oscillation frequency strictly
1136: defines the parameter $a$, while the damping coefficient gives a
1137: measurement of $\delta_L^2(T)$. From the fitting function it is
1138: possible to extract the spatial dependence of the second order
1139: correlation function
1140: $g^{(2)}_T(s,d)=g^{(2)}_T(\overline{z}_1,\overline{z}_2,\overline{z}_3,\overline{z}_4)$
1141: (see Eqs.~(\ref{C_Eq_19}) and (\ref{M_Eq_3})). The fit on the
1142: experimental data gives a phase coherence length
1143: $L_{\varphi}^{\mbox{\footnotesize exp}}=(57\pm10)\,\mu\textrm{m}$,
1144: compatible with the expected value
1145: $L_{\varphi}^{\mbox{\footnotesize th}}=(58\pm2)\,\mu\textrm{m}$.
1146: The good agreement between the experimental data, the numerical
1147: simulation and the model function of Eq.~(\ref{E_Eq_1})
1148: demonstrates that the free evolution of density modulations and of
1149: the fluctuating phase pattern do not influence the measurement of
1150: the second order correlation function. This result justifies the
1151: use of Eq.~(\ref{M_Eq_2}) to model the experimental data and to
1152: extract the coherence properties of the condensate.
1153: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1154: In Fig.~\ref{E_Fig_3}, we show a direct comparison between the
1155: measured phase coherence lengths in the center of the BEC and the
1156: theoretical values calculated according to Eq.~(\ref{C_Eq_9}), by
1157: using the measured numbers of atoms, temperatures and trapping
1158: frequencies. The bars indicate the statistical errors both on the
1159: measured values and on the theoretical predictions.
1160: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1161: \begin{figure}
1162: \includegraphics{Fig5}
1163: \caption{\label{E_Fig_3}Direct comparison between the measured
1164: phase coherence lengths and the theoretical values, calculated
1165: according to Eq.~(\ref{C_Eq_9}) by using the measured numbers of
1166: atoms, temperatures and trapping frequencies. The dotted line with
1167: slope 1 is used to compare experiment and theory. The bars on the
1168: plotted points indicate statistical errors. The relative
1169: systematic uncertainties on the calculated and measured phase
1170: coherence length are $26\%$ and $15\%$, respectively. This figure
1171: has previously been shown in \cite{Hellweg03}.}
1172: \end{figure}
1173: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1174: The dotted line with slope 1 highlights the good quantitative
1175: agreement between experiments and theory.
1176: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1177: %
1178: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1179: \section{Conclusion}
1180: In this paper, we have studied the coherence properties of phase
1181: fluctuating Bose-Einstein condensates. In highly elongated BECs
1182: the thermal excitation of quasiparticle modes can significantly
1183: reduce the coherence length of the system. Starting from the
1184: results of Petrov et al. \cite{Petrov01}, we have derived a
1185: general formula for the $N$-particle correlation function. The
1186: second order correlation function has been studied in detail and
1187: its limits both around the center of the condensate and in the
1188: local density approximation have been analyzed. In particular, we
1189: have discussed a method to directly characterize the second order
1190: correlation properties of the system. An analytic theory that
1191: describes the free evolution of the condensate phase has been
1192: developed to closely model the measurement process. Using a Bragg
1193: diffraction interferometer, we have measured the density
1194: correlations of the interference pattern generated by two
1195: spatially displaced copies of a parent BEC. This kind of
1196: measurement allows to correlate the field operator $\hat{\psi}$ of
1197: the parent condensate in four different $z$ positions. The
1198: averaging process directly gives the second order correlation
1199: function. The experiment confirms our theoretical predictions and
1200: demonstrates a method to measure the phase coherence length of the
1201: condensate. Compared to usual interference experiments this
1202: technique has the advantage of being insensitive to the global
1203: phase noise introduced by the experimental apparatus. The method
1204: presented here is in direct analogy to the original Hanbury-Brown
1205: and Twiss experiment and demonstrates the possibility of using
1206: density correlation measurements to study the coherence properties
1207: of Bose-Einstein condensates.
1208: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1209: %
1210: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1211: %\subsubsection{}
1212:
1213: % If in two-column mode, this environment will change to single-column
1214: % format so that long equations can be displayed. Use
1215: % sparingly.
1216: %\begin{widetext}
1217: % put long equation here
1218: %\end{widetext}
1219:
1220: % figures should be put into the text as floats.
1221: % Use the graphics or graphicx packages (distributed with LaTeX2e)
1222: % and the \includegraphics macro defined in those packages.
1223: % See the LaTeX Graphics Companion by Michel Goosens, Sebastian Rahtz,
1224: % and Frank Mittelbach for instance.
1225: %
1226: % Here is an example of the general form of a figure:
1227: % Fill in the caption in the braces of the \caption{} command. Put the label
1228: % that you will use with \ref{} command in the braces of the \label{} command.
1229: % Use the figure* environment if the figure should span across the
1230: % entire page. There is no need to do explicit centering.
1231:
1232: % \begin{figure}
1233: % \includegraphics{}%
1234: % \caption{\label{}}
1235: % \end{figure}
1236:
1237: % Surround figure environment with turnpage environment for landscape
1238: % figure
1239: % \begin{turnpage}
1240: % \begin{figure}
1241: % \includegraphics{}%
1242: % \caption{\label{}}
1243: % \end{figure}
1244: % \end{turnpage}
1245:
1246: % tables should appear as floats within the text
1247: %
1248: % Here is an example of the general form of a table:
1249: % Fill in the caption in the braces of the \caption{} command. Put the label
1250: % that you will use with \ref{} command in the braces of the \label{} command.
1251: % Insert the column specifiers (l, r, c, d, etc.) in the empty braces of the
1252: % \begin{tabular}{} command.
1253: % The ruledtabular enviroment adds doubled rules to table and sets a
1254: % reasonable default table settings.
1255: % Use the table* environment to get a full-width table in two-column
1256: % Add \usepackage{longtable} and the longtable (or longtable*}
1257: % environment for nicely formatted long tables. Or use the the [H]
1258: % placement option to break a long table (with less control than
1259: % in longtable).
1260: % \begin{table}%[H] add [H] placement to break table across pages
1261: % \caption{\label{}}
1262: % \begin{ruledtabular}
1263: % \begin{tabular}{}
1264: % Lines of table here ending with \\
1265: % \end{tabular}
1266: % \end{ruledtabular}
1267: % \end{table}
1268:
1269: % Surround table environment with turnpage environment for landscape
1270: % table
1271: % \begin{turnpage}
1272: % \begin{table}
1273: % \caption{\label{}}
1274: % \begin{ruledtabular}
1275: % \begin{tabular}{}
1276: % \end{tabular}
1277: % \end{ruledtabular}
1278: % \end{table}
1279: % \end{turnpage}
1280:
1281: % Specify following sections are appendices. Use \appendix* if there
1282: % only one appendix.
1283: %\appendix
1284: %\section{}
1285:
1286: % If you have acknowledgments, this puts in the proper section head.
1287: \begin{acknowledgments}
1288: We gratefully thank DFG for the support in the
1289: Sonderforschungsbereich 407 as well as the European Union for
1290: support in the RTN network "Preparation and application of
1291: quantum-degenerate cold atomic/molecular gases", contract
1292: HPRN-CT-2000-00125. L.S.~thanks the Alexander von Humboldt
1293: foundation and the ZIP program of the German government for
1294: support.
1295: \end{acknowledgments}
1296:
1297: % Create the reference section using BibTeX:
1298: \bibliography{g2}
1299:
1300: \end{document}
1301: %
1302: % ****** End of file template.aps ******
1303: