1: % ****** Start of file apssamp.tex ******
2: %
3: % This file is part of the APS files in the REVTeX 4 distribution.
4: % Version 4.0 of REVTeX, August 2001
5: %
6: % Copyright (c) 2001 The American Physical Society.
7: %
8: % See the REVTeX 4 README file for restrictions and more information.
9: %
10: % TeX'ing this file requires that you have AMS-LaTeX 2.0 installed
11: % as well as the rest of the prerequisites for REVTeX 4.0
12: %
13: % See the REVTeX 4 README file
14: % It also requires running BibTeX. The commands are as follows:
15: %
16: % 1) latex apssamp.tex
17: % 2) bibtex apssamp
18: % 3) latex apssamp.tex
19: % 4) latex apssamp.tex
20: %
21: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
22: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
23:
24: % Some other (several out of many) possibilities
25: %\documentclass[preprint,aps]{revtex4}
26: %\documentclass[preprint,aps,draft]{revtex4}
27: %\documentclass[prb]{revtex4}% Physical Review B
28:
29: \usepackage{graphicx}% Include figure files
30: \usepackage{dcolumn}% Align table columns on decimal point
31: \usepackage{bm}% bold math
32:
33: %\nofiles
34:
35: \begin{document}
36:
37: \title{Demonstration of three-photon de Broglie wavelength
38: by projection measurement}
39:
40: \author{B. H. Liu$^1$, F. W. Sun$^{1}$, Y. X. Gong$^{1}$, Y. F. Huang$^{1}$, Z. Y. Ou$^{1,2}\footnote{E-mail:
41: zou@iupui.edu}$, and G. C. Guo$^1$}
42: \affiliation{$^1$Key Laboratory of Quantum Information,
43: University of Science and Technology of China, \\CAS, Hefei, 230026, the People's Republic of China
44: \\$^2$Department of Physics, Indiana
45: University-Purdue University Indianapolis, 402 N. Blackford
46: Street, Indianapolis, IN 46202}
47:
48: \begin{abstract}
49: Two schemes of projection measurement are realized experimentally
50: to demonstrate the de Broglie wavelength of three photons without
51: the need for a maximally entangled three-photon state (the NOON
52: state). The first scheme is based on the proposal by Wang and
53: Kobayashi (Phys. Rev. A {\bf 71}, 021802) that utilizes a couple
54: of asymmetric beam splitters while the second one applies the
55: general method of NOON state projection measurement to
56: three-photon case. Quantum interference of three photons is
57: responsible for projecting out the unwanted states, leaving only
58: the NOON state contribution in these schemes of projection
59: measurement.
60: \end{abstract}
61:
62: \pacs{42.50.Dv, 42.25.Hz, 42.50.St, 03.65.Ta}
63:
64: \maketitle
65:
66: \section{Introduction}
67:
68: Photonic de Broglie wavelength of a multi-photon state is the
69: equivalent wavelength of the whole system when all the photons in
70: the system act as one entity. Early work by Jacobson {\it et al}
71: \cite{Jac} utilized a special beam splitter that sends a whole
72: incident coherent state to either one of the outputs thus creating
73: a Sch\"odinger-cat like state. The equivalent de Broglie
74: wavelength in this case was shown to be $\lambda/\langle n\rangle$
75: with $\langle n\rangle$ as the average photon number of the
76: coherent state. Such a scheme can be used in precision phase
77: measurement to achieve the so called Heisenberg limit
78: \cite{hei,hol,bol,ou} of $1/\langle n\rangle$ in phase
79: uncertainty.
80:
81: Perhaps, the easiest way to demonstrate the de Broglie wavelength
82: is to use the maximally entangled photon number state or the so
83: called NOON state of the form \cite{bol,ou,kok}
84: \begin{eqnarray}
85: |NOON\rangle = {1\over \sqrt{2}} \Big(|N\rangle_1|0\rangle_2 +
86: |0\rangle_1|N\rangle_2\Big), \label{1}
87: \end{eqnarray}
88: where 1,2 denote two different modes of an optical field. The $N$
89: photons in this state stick together either all in mode 1 or in
90: mode 2. Indeed, if we recombine modes 1 and 2 and make an N-photon
91: coincidence measurement, the coincidence rate is proportional to
92: \begin{eqnarray}
93: R_N \propto 1+\cos (2\pi N\Delta/\lambda),\label{2}
94: \end{eqnarray}
95: where $\Delta$ is the path difference between the two modes and
96: $\lambda$ is the single-photon wavelength. Eq.(\ref{2}) shows an
97: equivalent de Broglie wavelength of $\lambda/N$ for $N$ photons.
98:
99: NOON state of the form in Eq.(\ref{1}) for $N=2$ case was realized
100: with two photons from parametric down-conversion, which led to the
101: demonstrations of two-photon de Broglie wavelength \cite{ou2,rar}.
102: For $N>2$, however, it is not easy to generate the NOON state. The
103: difficulty lies in the cancellation of all the unwanted terms of
104: $|k, N-k\rangle$ with $k\ne 0, N$ in an arbitrary N-photon
105: entangled state of
106: \begin{eqnarray}
107: |\Phi_N\rangle = \sum_{k=0}^N c_k|k, N-k\rangle. \label{3}
108: \end{eqnarray}
109: A number of schemes have been proposed \cite{hof,sha,liub} and
110: demonstrated \cite{wal,mit} which were based on some sort of
111: multi-photon interference scheme for the cancellation.
112:
113: Without exceptions, the proposed and demonstrated schemes
114: \cite{hof,sha,liub,wal,mit} for the NOON state generation rely on
115: multi-photon coincidence measurement for revealing the phase
116: dependent relation in Eq.(\ref{2}). Since coincidence measurement
117: is a projective measurement, it may not respond to all the terms
118: in Eq.(\ref{3}). Indeed, Wang and Kobayashi \cite{wan} applied
119: this idea to a three-photon state and found that only the NOON
120: state part of Eq.(\ref{3}) contribute to a specially designed
121: coincidence measurement with asymmetric beam splitters. The
122: coincidence rate shows the signature dependence in the form of
123: Eq.(\ref{2}) on the path difference for the three-photon de
124: Broglie wavelength. Another projective scheme was recently
125: proposed by Sun {\it et al} \cite{sun1} and realized
126: experimentally by Resch {\it et al} \cite{res} for six photons and
127: by Sun {\it et al} \cite{sun2} for four photons. This scheme
128: directly projects an arbitrary N-photon state of the form in
129: Eq.(\ref{3}) onto an N-photon NOON state and thus can be scaled up
130: to an arbitrary N-photon case.
131:
132: In this paper, we will apply the two projective schemes to the
133: three-photon case. The three-photon state is produced from two
134: pairs of photons in parametric down-conversion by gating on the
135: detection of one photon among them \cite{san}. We find that
136: because of the asymmetric beam splitters, the scheme by Wang and
137: Kabayashi \cite{wan} has some residual single-photon effect under
138: less perfect situation while the NOON state projection scheme
139: cancels all lower order effects regardless the situation.
140:
141: The paper is organized as follows: in Sect.II, we will discuss the
142: scheme by Wang and Kobayashi \cite{wan} and its experimental
143: realization. In Sect.III, we will investigate the NOON state
144: projection scheme for three-photon case and implement it
145: experimentally. In both sections, we will deal with a more
146: realistic multi-mode model to cover the imperfect situations. We
147: conclude with a discussion.
148:
149: \section{Projection by asymmetric beam splitters}
150:
151: This scheme for three-photon case was first proposed by Wang and
152: Kobayashi \cite{wan} to use asymmetric beam splitter to cancel the
153: unwanted $|2,1\rangle$ or $|1,2\rangle$ term and is a
154: generalization of the Hong-Ou-Mandel interferometer
155: \cite{ou2,rar,hon} for two-photon case. But different from the
156: two-photon case, the state for phase sensing is not a three-photon
157: NOON state since only one unwanted term can be cancelled and there
158: is still another one left there. So a special arrangement has to
159: be made in the second beam splitter to cancel the contribution
160: from the other term. The following is the detail of the scheme.
161:
162: \subsection{Principle of Experiment}
163:
164: We first start with a single mode argument by Wang and Kobayashi.
165: The input state is a three-photon state of
166: $|2\rangle_a|1\rangle_b$. The three photons are incident on an
167: asymmetric beam splitter (BS1) with $T\ne R$ from two sides as
168: shown in Fig.1. The output state can be easily found from the
169: quantum theory of a beam splitter as \cite{ou3,cam}
170: \begin{widetext}
171: \begin{eqnarray}
172: |\mathrm{BS}1\rangle_{out} = \sqrt{3T^2R}~|3_c, 0_d\rangle +
173: \sqrt{3TR^2}~|0_c, 3_d\rangle + \sqrt{T}(T-2R)|2_c, 1_d\rangle +
174: \sqrt{R}(R-2T)|1_c, 2_d\rangle.~~~~~~ \label{4}
175: \end{eqnarray}
176: \end{widetext}
177: When $R=2T=2/3$, the $|1_c, 2_d\rangle$ term disappears from
178: Eq.(\ref{4}) due to three-photon interference and Eq.(\ref{4})
179: becomes
180: \begin{eqnarray}
181: |\mathrm{BS}1\rangle_{out} = {\sqrt{2}\over 3} |3_c, 0_d\rangle +
182: {2\over 3}|0_c, 3_d\rangle - {\sqrt{3}\over 3}|2_c, 1_d\rangle .
183: \label{5}
184: \end{eqnarray}
185: But unlike the two-photon case, the $|2_c, 1_d\rangle$ term is
186: still in Eq.(\ref{5}) so that the output state is not a NOON state
187: of the form in Eq.(\ref{1}).
188:
189: Now we can arrange a projection measurement to take out the $|2_c,
190: 1_d\rangle$ term in Eq.(\ref{5}). Let us combine $A$ and $B$ with
191: another beam splitter (BS2 in Fig.1) that has same transmissivity
192: and reflectivity ($R=2T=2/3$) as the first BS (BS1). According to
193: Eq.(\ref{5}), $|2_c, 1_d\rangle$ will not contribute to the
194: probability $P_3(1_e,2_f)$. So only $|3_c, 0_d\rangle$ and $|0_c,
195: 3_d\rangle$ in Eq.(\ref{5}) will contribute. The projection
196: measurement of $P_3(1_e,2_f)$ will cancel the unwanted middle
197: terms like $|2_c,1_d\rangle$ from Eq.(\ref{5}). Although the
198: coefficients of $|3_c, 0_d\rangle$ and $|0_c, 3_d\rangle$ in
199: Eq.(\ref{5}) are not equal, their contributions to $P_3(1_e,2_f)$
200: are the same after considering the unequal $T$ and $R$ in BS2. So
201: the projection measurement of $P_3(1_e,2_f)$ is responsive only to
202: the three-photon NOON state. Use of an asymmetric beam splitter
203: for the cancellation of $|2_c, 1_d\rangle$ was discussed by Sanaka
204: {\it et al} in Fock state filtering \cite{san}.
205:
206: The above argument can be confirmed by calculating the
207: three-photon coincidence rate $P_3(1_e,2_f)$ directly for the
208: scheme in Fig.1, which is proportional to \cite{m-w8}:
209: \begin{eqnarray}
210: P_3(1_e,2_f) = \langle 2_a,1_b|\hat e^{\dag}\hat f^{\dag 2} \hat
211: f^2 \hat e|2_a,1_b\rangle, \label{6}
212: \end{eqnarray}
213: with
214: \begin{equation}
215: \begin{cases}
216: \hat e = (\hat c + e^{i\varphi}\sqrt{2}\hat d)/\sqrt{3},\cr \hat
217: f = (e^{i\varphi} \hat d - \sqrt{2}\hat c)/\sqrt{3},
218: \end{cases}
219: \label{7}
220: \end{equation}
221: where we introduce a phase $\varphi$ between $A$ and $B$. But for
222: the first BS, we have
223: \begin{eqnarray}
224: \begin{cases}\hat c = (\hat a + \sqrt{2}\hat b)/\sqrt{3},\cr \hat d =
225: (\hat b - \sqrt{2}\hat a)/\sqrt{3}.
226: \end{cases}
227: \label{8}
228: \end{eqnarray}
229: Substituting Eq.(\ref{7}) into Eq.(\ref{6}) with Eq.(\ref{8}), we
230: obtain
231: \begin{eqnarray}
232: P_3(1_e,2_f) &=& \langle 2_a,1_b|\hat e^{\dag}\hat f^{\dag2}\hat
233: f^2\hat e|2_a,1_b\rangle \cr &=& {32\over 81} (1+\cos
234: 3\varphi),\label{9}
235: \end{eqnarray}
236: which has a dependence on the path difference $\Delta = \varphi
237: \lambda/2\pi$ that is same as in Eq.(\ref{2}) but with $N=3$.
238:
239: \begin{figure}[tbp]
240: \includegraphics[width=8cm]{Fig1.eps}
241: \caption{Arrangement of asymmetric beam splitters of a
242: three-photon interferometer for the demonstration of the
243: three-photon de Broglie wavelength.}
244: \end{figure}
245:
246: \subsection{Experiment}
247:
248: Experimentally, asymmetric beam splitters are realized via
249: polarization projections as shown in Fig.2 and its inset (a),
250: where a three-photon state of $|2_H,1_V\rangle$ is incident on a
251: combination of two half wave plates (HWP1, HWP2) and a phase
252: retarder (PS). The first half wave plate (HWP1) rotates the state
253: $|2_H,1_V\rangle$ by an angle $\alpha $ to $|2_a,1_b\rangle$ with
254: \begin{eqnarray}
255: \begin{cases}
256: \hat a_H = \hat a \cos \alpha + \hat b \sin\alpha ,\cr \hat a_V =
257: \hat b \cos \alpha - \hat a \sin\alpha,
258: \end{cases}\label{10}
259: \end{eqnarray}
260: where $\cos \alpha = \sqrt{T} = 1/\sqrt{3}$ is the amplitude
261: transmissivity of the asymmetric beam splitter. Eq.(\ref{10}) is
262: equivalent to Eq.(\ref{7}). The phase retarder (PS) introduces the
263: phase shift $\varphi$ between the H and V polarization. The second
264: half wave plate (HWP2) makes another rotation of the same angle
265: $\alpha$ for the two phase-shifted polarizations and the
266: polarization beam splitter finishes the projection required by
267: Eq.(\ref{8}).
268:
269: \begin{figure}
270: \includegraphics[width=3in]{Fig2}
271: \caption{Experimental setup. HWP1 and HWP2 is set for different
272: measurement, PS is the phase shifter between H-photon and
273: V-photon. Insets: (a) arrangement with asymmetric beam splitters
274: by polarization beam splitter (PBS) and (b) the three-photon NOON
275: state projection.}
276: \end{figure}
277:
278: The three-photon polarization state of $|2_H,1_V\rangle$ is
279: prepared by using two type-II parametric down conversion processes
280: shown in Fig.2. This scheme was first constructed by Liu {\it et
281: al} \cite{liu} to demonstrate controllable temporal
282: distinguishability of three photons. When the delay between the
283: two H-photons is zero, we have the state of $|2_H,1_V\rangle$. In
284: this scheme, two $\beta$-Barium Borate (BBO) crystals are pumped
285: by two UV pulses from a common source of frequency doubled
286: Ti:sapphire laser operating at 780 nm. The H-photon from BBO1 is
287: coupled to the H-polarization mode of BBO2 while the other
288: V-photon is detected by detector D and serves as a trigger. The H-
289: and V-photons from BBO2 are coupled to single-mode fibers and then
290: are combined by a polarization beam splitter (PBS). The combined
291: fields pass through an interference filter with 3 nm bandwidth and
292: then go to the assembly of HWP1, PS, and HWP2 to form a
293: three-photon polarization interferometer. There are two schemes of
294: projection measurement. In this section, we deal with the first
295: scheme in inset (a) of Fig.2, which consists of a PBS for
296: projection and three detectors (A, B, C) for measuring the
297: quantity $P_3(1_e, 2_f)$ in Eq.(\ref{6}) by three-photon
298: coincidence. To realize the transformation in Eqs.(\ref{7},
299: \ref{8}), HWP1 and HWP2 are set to rotate the polarization by
300: $\alpha = \cos^{-1}(1/\sqrt{3}) = 54.7^{\circ}$. In order to
301: obtain an input state of $|2_H,1_V\rangle$ to the interferometer,
302: we need to gate the three-photon coincidence measurement on the
303: detection at detector D. In this way, we ensure that the two
304: H-photons come from two crystals separately. Otherwise, we will
305: have an input state of $|2_H,2_V\rangle$. The delay ($\Delta T_H$)
306: between the two H-photons from BBO1 and BBO2 as well as the delay
307: ($\Delta T_V$) between the H- and V-photons are adjusted to insure
308: the three photons are indistinguishable in time. This is confirmed
309: by the photon bunching effect of the two H-photons \cite{liu} and
310: a generalized Hong-Ou-Mandel effect for three photons \cite{san}.
311:
312: \begin{figure}[tbp]
313: \includegraphics[width=8cm]{Fig3.eps}
314: \caption{Experimental result for projection measurement with
315: asymmetric beam splitters. The data is least-square-fitted to
316: $P_{40}(1+{\cal V}_{3}\cos 3\protect\varphi+{\cal V}_{1}\cos
317: \protect\varphi)$ with ${\cal V}_{3}=85\%$ and ${\cal V}_{1}=5\%$
318: after background subtraction. }
319: \end{figure}
320:
321: Four-photon coincidence count among ABCD detectors is measured as
322: a function of the phase shift $\varphi$. The experimental result
323: after background subtraction is shown in Fig.3. The data is
324: gathered in 100 sec for each point and the error bars are one
325: standard deviation. Backgrounds due to three and more pairs of
326: photons are estimated from single and two-photon rates to
327: contribute 1.2/sec on average to the raw data and are subtracted.
328: The data clearly show a $\cos 3\varphi$ dependence except the
329: unbalanced minima and maxima, which indicates an extra
330: $\cos\varphi$ dependence. Indeed, the data fit well to the
331: function
332: \begin{eqnarray}
333: P_4 = P_{40}( 1+{\cal V}_3\cos 3\varphi +{\cal V}_1 \cos
334: \varphi)\label{11}
335: \end{eqnarray}
336: with $P_{40}=184$, ${\cal V}_3 = 85\%$ and ${\cal V}_1=5\%$. The
337: $\chi^2$ of the fit is 30 and is comparable to the number of data
338: of 25, indicating a mostly statistical cause for the error.
339:
340: The appearance of the $\cos\varphi$ term in Eq.(\ref{11}) is an
341: indication that the cancellation of the $|2_c,1_d\rangle$ and
342: $|1_c,2_d\rangle$ is not complete in Eqs.(\ref{4}, \ref{9}) and
343: the residuals mix with the $|3_c,0_d\rangle$ and $|0_c,3_d\rangle$
344: terms to produce the $\cos\varphi$ term. This imperfect
345: cancellation is not a result of the wrong $T, R$ values but is due
346: to temporal mode mismatch among the three photons in the input
347: state of $|2_H,1_V\rangle$. To account for this mode mismatch, we
348: resort to a multi-mode model of the parametric down-conversion
349: process.
350:
351:
352: \subsection{Multi-mode analysis of three-photon interferometer with
353: asymmetric beam splitters}
354:
355: We start by finding the multi-mode description of the quantum
356: state from two parametric down-conversion processes. Since the
357: first one serves as the input to the second one, we need the
358: evolution operator for the process, which was first dealt with by
359: Ghosh {\it et al} \cite{gho} and later by Ou \cite{ou4} and by
360: Grice and Walmsley \cite{wam}. In general, the unitary evolution
361: operator for weakly pumped type-II process is given by
362: \begin{eqnarray}
363: \hat U = 1 + \eta \int d\omega_1d\omega_2 \Phi(\omega_1,\omega_2)
364: \hat a_H^{\dag}(\omega_1)\hat a_V^{\dag}(\omega_2), \label{12}
365: \end{eqnarray}
366: where $\eta$ is some parameter that is proportional to the pump
367: strength and nonlinear coupling. For simplicity without losing
368: generality, we assume the two processes are identical and are
369: governed by the evolution operator in Eq.(\ref{12}). Furthermore,
370: we assume the symmetry relation $\Phi(\omega_1,\omega_2) =
371: \Phi(\omega_2,\omega_1)$ which is in general not satisfied but can
372: be achieved with some symmetrizing tricks \cite{wam2,wong}. So for
373: the first process, because the input is vacuum, we obtain the
374: output state as
375: \begin{widetext}
376: \begin{eqnarray}
377: |\Psi_1\rangle = \hat U |vac\rangle= |vac\rangle + \eta \int
378: d\omega_1d\omega_2 \Phi(\omega_1,\omega_2) \hat
379: a_H^{\dag}(\omega_1)\hat a_V^{\dag}(\omega_2)|vac\rangle.
380: \label{13}
381: \end{eqnarray}
382: The second crystal has the state of $|\Psi_1\rangle$ as its input.
383: So after the second crystal, the output state becomes
384: \begin{eqnarray}
385: |\Psi_2\rangle = \hat U |\Psi_1\rangle =... + \eta^2 \int
386: d\omega_1d\omega_2d\omega_1^{\prime}d\omega_2^{\prime}
387: \Phi(\omega_1,\omega_2) \Phi(\omega_1^{\prime},\omega_2^{\prime})
388: \hat a_H^{\dag}(\omega_1^{\prime})\hat
389: a_{V}^{\dag}(\omega_2^{\prime})\hat a_H^{\dag}(\omega_1)\hat
390: a_{V_1}^{\dag}(\omega_2)|vac\rangle, \label{14}
391: \end{eqnarray}
392: \end{widetext}
393: where $V_1$ and $V$ denote the two non-overlapping vertical
394: polarization mode from the first and second crystals,
395: respectively. Here we only keep the four-photon term. Although
396: there are other four-photon terms in the $|\Psi_2\rangle$ state
397: corresponding to two-pair generation from one crystal alone, they
398: won't contribute to what we are going to calculate. So we omit
399: them in Eq.(\ref{14}).
400:
401: The field operators at the four detectors are given by
402: \begin{eqnarray}
403: \begin{cases}
404: \hat E_A(t) =\tau_1 \hat E_H(t) + \rho_1 \hat E_V(t),\cr \hat
405: E_B(t) = \big[\tau_2 \hat E_V(t) + \rho_2\hat
406: E_H(t)\big]/\sqrt{2}+... = \hat E_C(t) ,
407: \end{cases}
408: \label{15}
409: \end{eqnarray}
410: with $\tau_1=(1-2e^{i\varphi})/3, \tau_2 = (e^{i\varphi}-2)/ 3,
411: \rho_1= - \rho_2 = \sqrt{2} (1+e^{i\varphi})/3$. Here we used the
412: equivalent relations in Eqs.(\ref{7}, \ref{8}) to establish the
413: connection between the field operators $\hat E_A,\hat E_B,\hat
414: E_C$ and $\hat E_H,\hat E_V$ and for detector D, we have
415: \begin{eqnarray}
416: \hat E_D(t) = \hat E_{V_1}(t),\label{16}
417: \end{eqnarray}
418: with
419: \begin{eqnarray}
420: \hat E_k(t) = {1\over \sqrt{2\pi}}\int d\omega \hat a_k(\omega)
421: e^{-i\omega t}. ~~~~~(k=H, V, V_1) \label{17}
422: \end{eqnarray}
423:
424: The four-photon coincidence rate of ABCD is proportional to a time
425: integral of the correlation function
426: \begin{widetext}
427: \begin{eqnarray}
428: \Gamma^{(4)}(t_1,t_2,t_3,t_4) = \langle \Psi_2|\hat
429: E_D^{\dag}(t_4)\hat E_C^{\dag}(t_3)\hat E_B^{\dag}(t_2)\hat
430: E_A^{\dag}(t_1)\hat E_A(t_1)\hat E_B(t_2)\hat E_C(t_3)\hat
431: E_D(t_4)|\Psi_2\rangle.\label{18}
432: \end{eqnarray}
433: It is easy to first evaluate $\hat E_A(t_1)\hat E_B(t_2)\hat
434: E_C(t_3)\hat E_D(t_4)$:
435: \begin{eqnarray}
436: \hat E_A(t_1)\hat E_B(t_2)\hat E_C(t_3)\hat E_D(t_4) =
437: \big[(HHV+HVH)D\tau_1\tau_2\rho_2 + VHHD\rho_1\rho_2^2\big]/2
438: +...,\label{19}
439: \end{eqnarray}
440: where $H=\hat E_H$, $V= \hat E_V$, $D= \hat E_D$ for short and we
441: keep the time ordering of $t_1t_2t_3$. We also drop five terms
442: that give zero result when they operate on $|\Psi_2\rangle$. It is
443: now straightforward to calculate the quantity $\hat E_A(t_1)\hat
444: E_B(t_2)\hat E_C(t_3)$ $\hat E_D(t_4)|\Psi_2\rangle$, which has
445: the form of
446: \begin{eqnarray}
447: &&\hat E_A(t_1)\hat E_B(t_2)\hat E_C(t_3)\hat
448: E_D(t_4)|\Psi_2\rangle = {\eta^2\over 2}
449: \Big\{\big[G(t_1,t_2,t_3,t_4)+G(t_2,t_1,t_3,t_4)+
450: G(t_1,t_3,t_2,t_4)+ G(t_3,t_1,t_2,t_4)\big]\tau_1\tau_2\rho_2
451: \cr&&\hskip 3in
452: + \big[G(t_2,t_3,t_1,t_4)+G(t_3,t_2,t_1,t_4)\big]
453: \rho_1\rho_2^2\Big\}|vac\rangle,\label{20}
454: \end{eqnarray}
455: where
456: \begin{eqnarray}
457: G(t_1,t_2,t_3,t_4)= g(t_1,t_3)g(t_2,t_4)~~~~{\rm
458: with}~~~g(t,t^{\prime})\equiv {1\over 2\pi}\int d\omega_1d\omega_2
459: \Phi(\omega_1,\omega_2) e^{-i(\omega_1t +\omega_2t^{\prime})}.
460: \label{21}
461: \end{eqnarray}
462: Substituting Eq.(\ref{20}) into Eq.(\ref{18}) and carrying out the
463: time integral, we obtain
464: \begin{eqnarray}
465: P_4 &\propto& \int dt_1dt_2dt_3dt_4
466: \Gamma^{(4)}(t_1,t_2,t_3,t_4)\cr &=&{|\eta|^4\over 4}\int
467: d\omega_1d\omega_2d\omega_1^{\prime}d\omega_2^{\prime}
468: \Big|\big[\Phi(\omega_1,\omega_2)
469: \Phi(\omega_1^{\prime},\omega_2^{\prime})+\Phi(\omega_1,\omega_1^{\prime})
470: \Phi(\omega_2,\omega_2^{\prime})\big](\tau_1\tau_2\rho_2+\rho_1\rho_2^2)\cr
471: &&\hskip 3in+2\Phi(\omega_1^{\prime},\omega_2)
472: \Phi(\omega_1,\omega_2^{\prime})\tau_1\tau_2\rho_2\Big|^2.
473: \label{22}
474: \end{eqnarray}
475: \end{widetext}
476: With $\tau_1,\rho_1,\tau_2,\rho_2$, we can further reduce
477: Eq.(\ref{22}) to
478: \begin{eqnarray}
479: P_4\propto {2|\eta|^4(17{\cal A}+7{\cal E})\over 243}\Big[1+{\cal
480: V}_3 \cos 3\varphi +{\cal V}_1\cos \varphi\Big] \label{23}
481: \end{eqnarray}
482: where
483: \begin{eqnarray}
484: {\cal V}_3 = {8({\cal A}+2{\cal E})\over 17{\cal A} + 7{\cal E}},
485: ~~~ {\cal V}_1 = {9({\cal A}-{\cal E})\over 17{\cal A} + 7{\cal
486: E}}.\label{24}
487: \end{eqnarray}
488: and
489: \begin{eqnarray}
490: &&{\cal A} =\int d\omega_1d\omega_2d\omega_1^{\prime}
491: d\omega_2^{\prime} |\Phi(\omega_1,\omega_2)
492: \Phi(\omega_1^{\prime},\omega_2^{\prime})|^2 \label{25}\\ &&{\cal
493: E} =\int d\omega_1d\omega_2d\omega_1^{\prime} d\omega_2^{\prime}
494: \Phi^*(\omega_1,\omega_2)
495: \Phi^*(\omega_1^{\prime},\omega_2^{\prime})\cr&&\hskip 1.5in
496: \times \Phi(\omega_1^{\prime},\omega_2)
497: \Phi(\omega_1,\omega_2^{\prime}). ~~~~~\label{26}
498: \end{eqnarray}
499: In deducing Eqs.(\ref{22}--\ref{26}), we used the symmetry
500: relation $\Phi(\omega_1,\omega_2)=\Phi(\omega_2,\omega_1)$.
501:
502: Obviously, when ${\cal A}={\cal E}$, Eq.(\ref{23}) completely
503: recovers to Eq.(\ref{9}). In practice, we always have ${\cal
504: A}\ge{\cal E}$ by Schwartz inequality. When ${\cal E} < {\cal A}$,
505: Eq.(\ref{23}) has the same form as Eq.(\ref{11}), indicating that
506: the multi-mode analysis indeed correctly predicts the imperfect
507: cancellation of the $|2_H,1_V\rangle$ and $|2_V,1_H\rangle$ terms
508: in Eqs.(\ref{4}, \ref{9}). If we use the experimentally measured
509: ${\cal V}_3$ and ${\cal V}_1$ in Eq.(\ref{24}), we will obtain two
510: inconsistent values of ${\cal E}/{\cal A}$: $({\cal E}/{\cal
511: A})_3=0.65$ and $({\cal E}/{\cal A})_1=0.87$. The discrepancy is
512: the result of the break up of the symmetry relation of
513: $\Phi(\omega_1,\omega_2) = \Phi(\omega_2,\omega_1)$ for type-II
514: parametric down-conversion, which is reflected in the
515: less-than-unit visibility of the two-photon interference. This
516: imperfection can be modelled as spatial mode mismatch and
517: approximately modifies Eq.(\ref{24}) as
518: \begin{eqnarray}
519: {\cal V}_3 = v_1^3{8({\cal A}+2{\cal E})\over 17{\cal A} + 7{\cal
520: E}}, ~~~ {\cal V}_1 = v_1 {9({\cal A}-{\cal E})\over 17{\cal A} +
521: 7{\cal E}},\label{27}
522: \end{eqnarray}
523: where $v_1$ is the equivalent reduced visibility in single-photon
524: interference due to spatial mode mismatch. With the extra
525: parameter $v_1$ in Eq.(\ref{27}), we obtain a consistent $({\cal
526: E}/{\cal A})=0.86$ with $v_1 = 0.96$.
527:
528: \section{NOON state projection measurement}
529:
530: The projection measurement discussed in the previous section
531: relies on the cancellation of some specific terms and therefore
532: cannot be applied to an arbitrary photon number. In the following,
533: we will discuss another projection scheme that can cancel all the
534: unwanted terms at once and thus can be scaled up.
535:
536: \subsection{Principle of experiment}
537:
538:
539: The NOON-state projection measurement scheme was first proposed by
540: Sun {\it et al} \cite{sun1} and realized by Resch {\it et al}
541: \cite{res} for six-photon case and by Sun {\it et al} \cite{sun2}
542: for the four-photon case. Since it is based on a multi-photon
543: interference effect, it was recently used to demonstrate the
544: temporal distinguishability of an N-photon state \cite{xia,ou5}.
545: Here we will apply it to a three-photon superposition state for
546: the demonstration of three-photon de Broglie wave length without
547: the NOON state.
548:
549: The NOON-state projection measurement scheme for three-photon case
550: is sketched in inset (b) of Fig.1. In this scheme, the input field
551: is first divided into three equal parts. Then each part passes
552: through a phase retarder that introduces a relative phase
553: difference of $0, 2\pi/3,4\pi/3$ respectively between the H- and
554: V-polarization. The phase shifted fields are then projected to
555: $135^{\circ}$ direction by polarizers before being detected by A,
556: B, C detectors, respectively. It was shown that the three-photon
557: coincidence rate is proportional to
558: \begin{equation}
559: P_3 \propto {1\over 18}\Big|\langle
560: NOON_3|\Phi_3\rangle\Big|^2,\label{28}
561: \end{equation}
562: where $|NOON_3\rangle = (|3_H,0_V\rangle
563: -|0_H,3_V\rangle)/\sqrt{2}$ and $|\Phi_3\rangle =
564: c_0|3_H,0_V\rangle + c_1|2_H,1_V\rangle + c_2|1_H,2_V\rangle +
565: c_3|0_H,3_V\rangle$. Note that since $|2,1\rangle ,|1,2\rangle$
566: are orthogonal to the NOON-state, their contributions to $P_3$ are
567: zero. Assuming that $|c_0|=|c_3|=c$ and there is a relative phase
568: of $\varphi$ between H and V so that $c_0=c, c_3=ce^{i3\varphi}$,
569: we obtain from Eq.(\ref{28})
570: \begin{equation}
571: P_3 \propto {|c|^2\over 18}(1-\cos 3\varphi),\label{29}
572: \end{equation}
573: which is exactly in the form of Eq.(\ref{2}) with $N=3$, showing
574: the three-photon de Broglie wave length.
575:
576: \subsection{Experiment}
577:
578: From Sect.IIB, we learned that a state of $|2_H,1_V\rangle$ can be
579: produced with two parametric down-conversion processes. This state
580: will of course give no contribution to the NOON state projection
581: since it is orthogonal to the NOON state. On the other hand, we
582: can rotate the state by 45$^{\circ}$. Then the state becomes
583: \cite{ou3,cam}
584: \begin{eqnarray}
585: &&|\Phi_3\rangle = \sqrt{3\over 8}\Big(|3_H,0_V\rangle
586: -|0_H,3_V\rangle\Big) \cr &&\hskip 1in +
587: {1\over\sqrt{8}}\Big(|2_H,1_V\rangle -
588: |1_H,2_V\rangle\Big),~~~~~~\label{30}
589: \end{eqnarray}
590: which has the NOON state component with $c=\sqrt{3/8}$.
591:
592: Experimentally, the three-photon state of $|2_H,1_V\rangle$ is
593: prepared in the same way described in Sect.IIB and shown in Fig.2.
594: Different from Sect.IIB, the polarizations of the prepared state
595: are rotated 45$^{\circ}$ by HWP1 to achieve the state in
596: Eq.(\ref{30}). The phase shifter (PS) then introduces a relative
597: phase difference $\varphi$ between the H and V polarizations and
598: HWP2 is set at zero before the NOON state projection measurement
599: is performed [Inset (b) of Fig.2]. As before, a four-photon
600: coincidence measurement among ABCD detectors is equivalent to a
601: three-photon coincidence measurement by ABC detectors gated on the
602: detection at D, which is required for the production of
603: $|2_H,1_V\rangle$.
604:
605: \begin{figure}[tbp]
606: \includegraphics[width=8cm]{Fig4.eps}
607: \caption{Experimental result for the NOON state projection
608: measurement. The data is least-square-fitted to $P_{40}(1+{\cal
609: V}_{3}\cos 3\protect\varphi$ with ${\cal V}_{3}=84\%$ after
610: background subtraction.}
611: \end{figure}
612:
613: Four-photon coincidence count among ABCD detectors is registered
614: in 200 sec as a function of the phase $\varphi$ (PS). The data
615: after subtraction of background contributions is plotted in Fig.4.
616: It clearly shows a sinusoidal dependence on $\varphi$ with a
617: period of $2\pi/3$. The solid curve is a chi-square fit to the
618: function of $P_4=P_{40}[1+{\cal V}_3\cos 3(\varphi+\varphi_0)]$
619: with $P_{40}=103/200$sec and ${\cal V}_3=0.84$. The $\chi^2$ of
620: the fit is 24.3, which is comparable to the number of data of 25
621: indicating a good statistical fit.
622:
623: The less-than-unit visibility is a result of temporal
624: distinguishability among the three photons produced from two
625: crystals. It can only be accounted for with a multi-mode model of
626: the state given in Sect.IIC. Let's now apply it to the current
627: scheme.
628:
629: \subsection{Multi-mode analysis}
630:
631: The input state is same as Eq.(\ref{14}). But the field operators
632: are changed to
633: \begin{eqnarray}
634: \begin{cases}
635: \hat E_A(t) = \big[\hat E_+(t) - e^{i\varphi}\hat
636: E_-(t)\big]/\sqrt{6} + ...,\cr \hat E_B(t) = \big[\hat E_+(t)
637: -e^{i\varphi} \hat E_-(t)e^{i2\pi/3}\big]/\sqrt{6} + ...,\cr \hat
638: E_C(t) = \big[\hat E_+(t) - e^{i\varphi}\hat
639: E_-(t)e^{i4\pi/3}\big]/\sqrt{6} + ...,
640: \end{cases}
641: \label{31}
642: \end{eqnarray}
643: with
644: \begin{eqnarray}
645: \begin{cases}
646: \hat E_+(t) = \big[\hat E_H(t) + \hat E_V(t)\big]/\sqrt{2},\cr
647: \hat E_-(t) = \big[\hat E_H(t) - \hat E_V(t)\big]/\sqrt{2},
648: \end{cases}
649: \label{32}
650: \end{eqnarray}
651: where we omit the vacuum input fields and $\varphi$ is the phase
652: shift introduced by PS in Fig.2. The field operator for detector D
653: is same as Eq.(\ref{16}).
654:
655: As in Sect.IIC, the four-photon coincidence rate is related to the
656: correlation function in Eq.(\ref{18}) and we can first evaluate
657: $\hat E_A(t_1)\hat E_B(t_2)\hat E_C(t_3)\hat E_D(t_4)$. With the
658: field operators in Eq.(\ref{31}), we obtained
659: \begin{eqnarray}
660: &&\hat E_A(t_1)\hat E_B(t_2)\hat E_C(t_3)\hat E_D(t_4) \cr
661: &&\hskip 0.3in = \big(HHV a_1 + HVH a_2 +VHH
662: a_3\big)/12\sqrt{12},~~~~~\label{33}
663: \end{eqnarray}
664: with
665: \begin{eqnarray}
666: \begin{cases} a_1=1+ e^{i3\varphi}+ 2e^{i(2\varphi+2\pi/3)}+
667: 2e^{i(\varphi+4\pi/3)}\cr a_2 = 1+ e^{i3\varphi}+
668: 2e^{i(2\varphi+4\pi/3)}+ 2e^{i(\varphi+2\pi/3)}\cr a_3=1+
669: e^{i3\varphi}+ 2e^{i2\varphi}+ 2e^{i\varphi},\label{34}
670: \end{cases}
671: \end{eqnarray}
672: where the notations are same as in Eq.(\ref{19}) and we used the
673: identity $1+e^{i2\pi/3}+e^{i4\pi/3} = 0$. As before, we also drop
674: five terms that give zero result when they operate on
675: $|\Psi_2\rangle$. Now we can calculate the quantity $\hat
676: E_A(t_1)\hat E_B(t_2)\hat E_C(t_3)\hat E_D(t_4)|\Psi_2\rangle$,
677: which has the form of
678: \begin{widetext}
679: \begin{eqnarray}
680: \hat E_A(t_1)\hat E_B(t_2)\hat E_C(t_3)\hat E_D(t_4)|\Psi_2\rangle
681: = {\eta^2\over \sqrt{12^3}}
682: \Big\{\big[G(t_1,t_2,t_3,t_4)+G(t_2,t_1,t_3,t_4)\big] a_1+\big[
683: G(t_1,t_3,t_2,t_4)+ G(t_3,t_1,t_2,t_4)\big]a_2\nonumber
684: \end{eqnarray}
685: \begin{eqnarray}
686: \hskip 3in
687: + \big[G(t_2,t_3,t_1,t_4)+G(t_3,t_2,t_1,t_4)\big]
688: a_3\Big\}|vac\rangle,\label{35}
689: \end{eqnarray}
690: \end{widetext}
691: where $G(t_1,t_2,t_3,t_4)$ is given in Eq.(\ref{21}). After the
692: time integral, we obtain
693: \begin{eqnarray}
694: P_4(NOON) \propto {|\eta|^4(2{\cal A}+{\cal E})\over
695: 72}\big(1+{\cal V}_3 \cos 3\varphi \big) \label{36}
696: \end{eqnarray}
697: with
698: \begin{eqnarray}
699: {\cal V}_3(NOON) = {{\cal A} +2{\cal E}\over 2{\cal A}+{\cal E}},
700: \label{37}
701: \end{eqnarray}
702: where ${\cal A}$ and ${\cal E}$ are given in Eqs.(\ref{25},
703: \ref{26}). Note that the terms such as $\cos2\varphi, \cos\varphi$
704: are absent in Eq.(\ref{37}) even in the non-ideal case of ${\cal
705: E}<{\cal A}$. This is because of the symmetry among the three
706: detectors A, B, C involved in the three-photon NOON state
707: projection measurement. When spatial mode mismatch is considered,
708: the visibility is changed to
709: \begin{eqnarray}
710: {\cal V}_3(NOON) = v_1^3{{\cal A} +2{\cal E}\over 2{\cal A}+{\cal
711: E}}. \label{38}
712: \end{eqnarray}
713: With $v_1$ and the quantity ${\cal E}/{\cal A}$ obtained in
714: Sect.IIC, we have ${\cal V}_3(NOON) =0.85$, which is close to the
715: observed value of 0.84 in Sect.IIIB.
716:
717: \section{Summary and Discussion}
718:
719: In summary, we demonstrate a three-photon de Broglie wavelength by
720: using two different schemes of projection measurement without the
721: need for a hard-to-produce NOON state. Quantum interference is
722: responsible for the cancellation of the unwanted terms. The first
723: scheme by asymmetric beam splitters targets specific terms while
724: the second one by NOON state projection cancels all the unwanted
725: terms at once. We use a multi-mode model to describe the non-ideal
726: situation encountered in the experiment and find good agreements
727: with the experimental results.
728:
729: Since the scheme by asymmetric beam splitters is only for some
730: specific terms, it cannot be easily scaled up to arbitrary number
731: of photons although the extension to the four-photon case is
732: available. The extension of the scheme by NOON state projection to
733: arbitrary number of photons is straightforward. In fact,
734: demonstrations with four and six photons have been done with
735: simpler states \cite{res,sun2}.
736:
737: On the other hand, the scheme of NOON state projection need to
738: divide the input fields into $N$ equal parts while the scheme with
739: asymmetric beam splitters requires less partition. So the latter
740: will have higher coincidence rate than the former. In fact, Fig.3
741: and Fig.4 show a ratio of 4 after pump intensity correction. This
742: is consistent with the ratio of 4.8 from Eqs.(\ref{23}, \ref{36})
743: when ${\cal E= A}$. The difference may come from the different
744: collection geometry in the layout.
745:
746: The dependence of the visibility in Eqs.(\ref{27}, \ref{37}) on
747: the quantity ${\cal E/A}$ reflects the fact that the interference
748: effect depends on the temporal indistinguishability of the three
749: photons. From previous studies \cite{ou6,tsu,ou7,sun1,liu}, we
750: learned that the quantity ${\cal E/A}$ is a measure of
751: indistinguishability between two pairs of photons in parametric
752: down-conversion. In our generation of the $|2_H, 1_V\rangle$
753: state, one of the H-photon is from another pair of down-converted
754: photons. So to form an indistinguishable three-photon state, we
755: need pair indistinguishability, i.e., ${\cal E/A}\rightarrow 1$.
756:
757:
758: \begin{acknowledgments}
759: This work was funded by National Fundamental Research Program of
760: China (2001CB309300), the Innovation funds from Chinese Academy of
761: Sciences, and National Natural Science Foundation of China (Grant
762: No. 60121503 and No. 10404027)). ZYO is also supported by the US
763: National Science Foundation under Grant No. 0245421 and No.
764: 0427647.
765: \end{acknowledgments}
766:
767: \begin{thebibliography}{99}
768: \bibitem{Jac} J. Jacobson, G. Bj\"{o}rk, I. Chuang and Y. Yamamoto,
769: Phys. Rev. Lett. {\bf 74}, 4835 (1995).
770:
771: \bibitem {hei} W. Heisenberg, {\it Zeitschr.\ f.\ Physik} {\bf 43}, 172
772: (1927).
773:
774: \bibitem{hol} M. J. Holland and K. Burnett, Phys. Rev. Lett. 71, 1355
775: (1993).
776:
777: \bibitem{bol} J. J. Bollinger, W. M. Itano, D. J. Wineland, and D. J.
778: Heinzen, Phys. Rev. A54, R4649 (1996).
779:
780: \bibitem{ou} Z. Y. Ou, Phys. Re. Lett. 77, 2352 (1996); Phys. Rev. A {\bf 55}, 2598 (1997).
781:
782: \bibitem{kok} A. N. Boto, P. Kok, D. S. Abrams, S. L. Braunstein, C. P.
783: Williams, and J. P. Dowling, Phys. Rev. Lett. 85, 2733 (2000); P.
784: Kok, A. N. Boto, D. S. Abrams, C. P. Williams, S. L. Braunstein,
785: and J. P. Dowling, Phys. Rev. A63, 063407 (2001).
786:
787: \bibitem{ou2} Z. Y. Ou, X. Y. Zou, L. J. Wang, and L. Mandel, Phys. Rev. A {\bf 42},
788: 2957 (1990).
789:
790: \bibitem {rar} J. G. Rarity, P. R. Tapster, E. Jakeman, T. Larchuk, R. A. Campos, M. C. Teich, and B. E. A.
791: Saleh, Phys. Rev. Lett. {\bf 65}, 1348 (1990).
792:
793: \bibitem{hof} H. F. Hofmann, Phys. Rev. A {\bf 70}, 023812 (2004).
794:
795: \bibitem{sha} F. Shafiei, P. Srinivasan, and Z. Y. Ou, Phys. Rev. A {\bf 70},
796: 043803 (2004).
797:
798: \bibitem{liub} B. Liu and Z. Y. Ou, Phys. Rev. A {\bf 74},
799: 035802 (2006).
800:
801: \bibitem{wal} P. Walther, J. W. Pan, M. Aspelmeyer, R. Ursin, S.
802: Gasparoni, and A. Zeilinger, Nature (London) {\bf 429}, 158
803: (2004).
804:
805: \bibitem{mit} M. W. Mitchell, J. S. Lundeen, and A. M. Steinberg, Nature
806: (London) {\bf 429}, 161 (2004).
807:
808: \bibitem{wan} H. B. Wang, and T. Kobayashi, Phys. Rev. A {\bf 71},
809: 021802(R) (2005).
810:
811: \bibitem{sun1} F. W. Sun, Z. Y. Ou, and G. C. Guo, Phys. Rev. A {\bf 73}, 023808
812: (2006).
813:
814: \bibitem{res} K. J. Resch, K. L. Pregnell, R. Prevedel, A. Gilchrist, G. J.
815: Pryde, J. L. O'Brien, and A. G. White, quant-ph/0511214.
816:
817: \bibitem{sun2} F. W. Sun, B. H. Liu, Y. F. Huang, Z. Y. Ou, and G. C.
818: Guo, Phys. Rev. A {\bf 74}, 033812 (2006).
819:
820:
821: \bibitem{san} K. Sanaka, K. J. Resch, and A. Zeilinger,
822: Phys. Rev. Lett. {\bf 96}, 083601 (2006).
823:
824: \bibitem{hon} C.K. Hong, Z.Y. Ou, and L. Mandel, Phys. Rev. Lett. {\bf 59}, 2044 (1987).
825:
826: \bibitem {ou3} Z. Y. Ou, C. K. Hong, and L. Mandel, Opt. Commun. {\bf 63}, 118
827: (1987).
828:
829: \bibitem {cam} R. A. Campos, B. E. A. Saleh, and M. C. Teich, Phys. Rev. A {\bf 40}, 1371 (1989).
830:
831:
832: \bibitem{m-w8} L. Mandel and E. Wolf, {\it Optical Coherence and Quantum Optics}, (Cambridge
833: University Press, New York, 1995).
834:
835: \bibitem{liu} B. H. Liu, F. W. Sun, Y. X. Gong, Y. F. Huang, Z. Y. Ou, and G. C. Guo,
836: quant-ph/0606118.
837:
838: \bibitem{gho} R. Ghosh, C. K. Hong, Z. Y. Ou, and L. Mandel, Phys. Rev. A{\bf 34},
839: 3962 (1986).
840:
841: \bibitem {ou4} Z. Y. Ou, Quan. Semiclass. Opt. {\bf 9} 599 (1997).
842:
843: \bibitem {wam} W. P. Grice and I. A. Walmsley, Phys. Rev. A {\bf 56}, 1627
844: (1997).
845:
846: \bibitem {wam2} D. Branning, W. P. Grice, R. Erdmann, and I. A.
847: Walmsley, Phys. Rev. Lett. {\bf 83}, 955 (1999)
848:
849: \bibitem{wong} O. Kuzucu, M. Fiorentino, M. A. Albota, F. N. C.
850: Wong, and F. X. K\"artner, Phys. Rev. Lett. {\bf 94}, 083601
851: (2005).
852:
853: \bibitem {ou5} Z. Y. Ou, quant-ph/0601118.
854:
855: \bibitem{xia} G. Y. Xiang, Y. F. Huang, F. W. Sun, P. Zhang, Z. Y. Ou, and G. C. Guo,
856: Phys. Rev. Lett. {\bf 97}, 023604 (2006).
857:
858: \bibitem{ou6} Z. Y. Ou, J. K. Rhee, and L. J. Wang, Phys. Rev. A {\bf 60}, 593
859: (1999).
860:
861: \bibitem{tsu} K. Tsujino, H. F. Hofmann, S. Takeuchi, and K. Sasaki,
862: Phys. Rev. Lett. {\bf 92}, 153602 (2004).
863:
864: \bibitem {ou7} Z. Y. Ou, Phys. Rev. A {\bf 72}, 053814 (2005).
865: \end{thebibliography}
866:
867: \end{document}
868: