1: \documentclass[pra,twocolumn]{revtex4}
2:
3: \usepackage{graphicx}
4: \usepackage{hyperref}
5: \usepackage{amssymb,amsmath}
6: \usepackage{array}
7: \usepackage{color}
8:
9: \begin{document}
10:
11: \title{Type-II Optical Parametric Oscillator :\\
12: a versatile source of quantum correlations and entanglement}
13:
14: \author{Julien Laurat, Thomas Coudreau and Claude Fabre{\footnote[1]{To whom
15: correspondence should be addressed (fabre@spectro.jussieu.fr)}}}
16:
17: \affiliation{Laboratoire Kastler Brossel, Universit{\'e} P. et M.
18: Curie, Case 74, 4 Place Jussieu, 75252 Paris cedex 05, France}
19:
20: \begin{abstract}
21: Type-II Optical Parametric Oscillators are efficient sources of
22: quadrature squeezed or polarization-squeezed light, intensity
23: correlated beams, and entangled light. We review here the
24: different levels of quantum correlations and entanglement that are
25: reached in this device, and present some applications.
26: \end{abstract}
27:
28: \date{\today}
29:
30:
31: \maketitle
32:
33: \section{Introduction}
34:
35: Quantum correlations play a key role in quantum mechanics, in
36: basic issues such as non-locality or decoherence and also in
37: potential applications such as quantum information processing and
38: computation. The existence of correlations between different
39: physical systems is obviously not a specific property of quantum
40: physics : it is simply the consequence of a former interaction,
41: whatever its character, between the systems submitted to the
42: measurement. Consequently, the observation or prediction of a
43: correlation, even perfect, between the measurements of two
44: variables is not at all a proof of the quantum character of the
45: phenomenon under study. One can find in the literature a great
46: deal of criteria setting a border between the classical and the
47: quantum effects, differing by the definitions of the quantum
48: character of a given physical situation. The purpose of this paper
49: is to review some criteria for quantum correlations (section
50: \ref{criteres}) and to describe how a single device, namely a
51: type-II Optical Parametric Oscillator (section \ref{montage})
52: produces various kinds of correlations fulfilling these criteria
53: (sections \ref{2quad} and \ref{4quad}). We will also show that the
54: same device can generate entangled states in a non-standard form
55: (section \ref{manipulation}). This last section will provide a
56: good insight into general properties of two-mode gaussian states,
57: illustrated in terms of covariance matrices.
58:
59: The results presented in this paper are detailed in Refs.
60: \cite{criteres,optlett_above,mescond,theo_mescond,pra_bf,pra_epr}.
61:
62: \section{Correlation criteria}\label{criteres}
63:
64: Let us consider two light beams denoted by indices 1 and 2. We
65: denote by $\delta X_{1,2}$ one quadrature component of these
66: beams, which can be measured either by direct photodetection
67: (amplitude quadrature) or by an homodyne detection, normalized in
68: such a way that vacuum fluctuations have a variance equal to 1. We
69: restrict ourselves in this paper to the "balanced" case when the
70: two beams have equal variances $F$ on these quadratures, and also
71: equal frequencies. More general criteria in the unbalanced case
72: can be found in \cite{criteres}. Let us stress also that we are in
73: the case where the quantum properties of the system are well
74: described by a linearized approach of quantum fluctuations.
75:
76: \subsection{"Gemellity"}
77:
78: A first criterion of quantum correlations can be defined as
79: follows: \textit{the correlation measured in the system cannot be
80: described by a semi-classical model involving classical
81: electromagnetic fields having classical fluctuations}.
82:
83: It is easy to show that the classical character of light fields is
84: preserved by linear ”passive” optical devices, which involve only
85: linear, energy-preserving, optical elements like beamsplitters and
86: free propagation. In order to ascertain the quantum character of
87: correlations existing between $\delta X_{1}$ and $\delta X_{2}$,
88: the simplest way is therefore to process the two beams by all
89: possible linear passive optical devices : if one is able to
90: produce in such a way a beam having fluctuations below the quantum
91: noise limit, that is well-known to be "non-classical", the initial
92: correlation will also be termed as ”non-classical”.
93:
94: For balanced beams, the best linear processing is simply to send
95: them on a 50/50 beam-splitter: one obtains at one of its output
96: ports a beam with quadrature fluctuations $\delta X_{out}$ given
97: by
98: \begin{equation}
99: \delta X_{out} = \frac{\delta X_1 - \delta X_2}{\sqrt 2}
100: \end{equation}
101: having a variance given by:
102: \begin{equation}
103: G = \frac{1}{2}\left\langle \left(\delta X_1 - \delta X_2\right)^2
104: \right\rangle
105: \end{equation}
106: The correlation will be said to be non-classical when this
107: quantity, that can be called the "gemellity", is smaller than 1.
108: $G$ can also be written in terms of the noise variance of each
109: beam $F$ and of the normalized correlation coefficient $C_{12}$:
110: \begin{equation}
111: G = F (1 - \left|C_{12}\right|).\label{eq:gemellite}
112: \end{equation}
113: Therefore a correlation is non-classical when the normalized
114: correlation function fulfills the following condition:
115: \begin{equation}
116: |C_{12}| > 1 - \frac{1}{F}
117: \end{equation}
118: Thus the larger the classical noise is on each beam, the more
119: stringent the condition becomes.
120:
121: Finally, let us stress that $G$ can be easily measured
122: experimentally: this is done in all homodyne detection schemes of
123: squeezing, which actually measures the quantum character of the
124: correlation existing between the two beams produced by mixing the
125: field to measure with the local oscillator, and in all twin beams
126: experiments\cite{optlett_above,traditional}.
127:
128: \subsection{Quantum Non Demolition correlation}
129:
130: When two observables $M_1$ and $M_2$ are correlated, the
131: measurement of $M_2$ gives some information about the value of
132: $M_1$ without any interaction with system $1$. Correlations
133: provide therefore opportunities for Non Demolition measurements.
134: One is led to a second criterion of quantum correlation:
135: \textit{the correlation is such that the information extracted
136: from the measurement on one field provides a Quantum Non
137: Demolition measurement of the other \cite{qnd}}.
138:
139: This criterion is related to the conditional variance given by~:
140: \begin{equation}
141: V_{1|2} = F_1(1-C_{12}^2).\label{eq:varcond}
142: \end{equation}
143: where $F_1$ is the noise of beam 1 normalized to shot noise. QND
144: correlations correspond to values of $V_{1|2}$ below 1, and
145: therefore to a correlation satisfying the inequality:
146: \begin{equation}
147: |C_{12}| > \sqrt{1-\frac{1}{F}}
148: \end{equation}
149:
150: Eq. \ref{eq:varcond} can also be expressed in terms of the
151: gemellity:
152: \begin{equation}
153: V_{1|2} = V_{2|1} =
154: G(1+|C_{12}|)=2G-\frac{G^2}{F}.\label{eq:varcond2}
155: \end{equation}
156: It is easy to show from these relations that all QND-correlated
157: beams have a gemellity smaller than 1, whereas a gemellity smaller
158: than 0.5 is required to have QND-correlated beams (in the limit of
159: large individual noise).
160:
161: \subsection{Inseparability}
162: \label{sec:insep}
163:
164: Let us now define a new criterion related to entanglement:
165: \textit{the correlation cannot be described by separable quantum
166: states}. Can the state be written as (a sum of) tensor products or
167: not?
168:
169: If one is sure that the system is in a pure state, separable or
170: factorizable state vectors give rise to no correlations at all,
171: whatever the observables: the existence of a non-zero correlation,
172: even "classical", on a single quadrature is sufficient to prove
173: the inseparability of the state.
174:
175: When the state is mixed, which is the general case, this is no
176: longer the case. Let us consider for example the mixed state
177: described by the density matrix
178: \begin{equation}
179: \rho = \sum_n p_n \left(|1: n\rangle \otimes |2: n \rangle\right)
180: \left(\langle 1:n| \otimes \langle 2:n |\right)
181: \end{equation}
182: where $|1,2: n \rangle$ is a Fock state with $n$ photons in mode
183: 1, 2. This highly non-classical state has perfect intensity
184: correlations ($C_{12}=1$), so that $G=V_{1|2} = 0$). However, it
185: is a separable state, being a statistical mixture of factorized
186: state vectors.
187:
188: In order to ascertain the separable character of the physical
189: state of a system, one needs to make two joint correlation
190: measurements on non-commuting observables on the system, and not
191: only one, as was the case in the two previous criteria. More
192: precisely, Duan \emph{et al.}\cite{duan} have shown that, in the
193: case of Gaussian states for which the covariance matrix is
194: expressed in the so-called standard form, there exists a necessary
195: and sufficient criterion of separability in terms of the quantity
196: $\mathcal I$, that we will call ”separability”, and is given by :
197: \begin{eqnarray}
198: \mathcal I &=& \frac{1}{4} \left(\left\langle \left(\delta X_1 -
199: \delta X_2\right)^2\right\rangle + \left\langle \left(\delta P_1 +
200: \delta
201: P_2\right)^2\right\rangle\right)\nonumber\\&=&\frac{1}{2}\left(G_X+G_P\right).
202: \label{eq:insep}
203: \end{eqnarray}
204: The separability appears as the half-sum of the gemellity
205: measuring the correlations between quadratures $\delta X$ and the
206: (anti)gemellity measuring the anticorrelations between $\delta P$.
207: A state for which $\mathcal I$ is smaller than one will be a
208: non-separable or entangled state. As a consequence, classically
209: correlated beams, for which these two gemellities are larger than
210: 1, are separable.
211:
212: Let us note that in the case of symmetric gaussian states the
213: entanglement can be quantified by a quantity called entropy of
214: formation -- or entanglement of formation $EOF$ --, that was
215: introduced in Ref.\cite{giedke}. It represents the amount of pure
216: state entanglement needed to prepare the entangled state. This
217: quantity is related to the value of the inseparability $\mathcal
218: I$ by:
219: \begin{equation}\label{EOF1}
220: EOF=c_{+}\log_{2}(c_{+})-c_{-}\log_{2}(c_{-})
221: \end{equation}
222: with
223: \begin{equation}\label{EOF2}
224: c_{\pm}=(\mathcal{I}^{-1/2} \pm \mathcal{I}^{1/2})^{2}/4
225: \end{equation}
226: EOF takes a positive value only for entangled beams. Its interest
227: is that it constitutes a real measure of the amount of
228: entanglement. In addition, it is also used in the discrete
229: variable regime.
230:
231: A more general quantity has been introduced to characterize the
232: entanglement: the logarithmic negativity \cite{vidal}. This
233: quantity can be calculated for any arbitrary bipartite system. We
234: will consider it in more detail in the last section of this paper
235: where the generated two-mode state is not in a standard form.
236:
237: \subsection{Einstein-Podolsky-Rosen correlations}
238:
239: Two correlations give the opportunity of Non Demolition
240: measurements on two non-commuting variables. As for a single
241: quadrature, one can be interested in the quality of the
242: information that one gets on one beam by measuring the other. This
243: question is related to the question asked by Einstein, Podolsky
244: and Rosen in their famous paper \cite{epr35}. In particular, we
245: will say that we have \textit{"EPR beams" when the information
246: extracted from the the measurement of the two quadratures of one
247: field provide values for the quadratures of the other which
248: "violate" the Heisenberg inequality.} This criterion was
249: considered and discussed extensively by Reid and
250: co-workers\cite{reid}. They showed that to characterize this
251: property, one can use the product of the conditional variances,
252: \begin{equation}
253: \mathcal V = V_{X_1|X_2} . V_{P_1|P_2}.
254: \end{equation}
255: When this quantity is smaller than one we will say that we have
256: "EPR-correlated beams". Let us note that when this condition is
257: fulfilled, one can perform double QND-measurements, that is two
258: QND-measurements on non-commuting quadratures. One can show that
259: all EPR-correlated beams are not separable, whereas the reverse is
260: not true. EPR-correlation is therefore the strongest of the
261: correlation criteria that we have listed here. One can envision
262: other criteria which are even stronger, but not relevant for the
263: problem of measuring correlated quadratures with Gaussian
264: statistics that we are considering here.
265:
266: \section{Experimental investigation of quantum correlations}
267: \label{experiences}
268:
269: In this section, we will show how these various criteria can be
270: tested using the states produced by a triply resonant type-II
271: Optical Parametric Oscillator. Such a system consists of a
272: triply-resonant optical cavity containing a type-II phase matched
273: $\chi^{(2)}$ crystal. Spontaneous parametric down-conversion which
274: occurs in such crystals is well known to produce twin photons,
275: that is photons created in pairs. When such a crystal is placed
276: inside a cavity and the system pumped above a critical value
277: (threshold), one generates intense beams which are correlated. The
278: system transfers the correlations existing in the discrete regime
279: to the continuous variable one. The OPO is thus an ideal system to
280: test the various criteria that we have mentioned above.
281:
282:
283: \begin{figure}[h]
284: \centerline{\includegraphics[width=0.9\columnwidth]{Schema_twin.eps}}
285: \caption{A cw doubled Nd:YAG laser pumps above threshold a type-II
286: OPO. Intensity correlations are directly measured by a balanced
287: detection scheme. PD Lock: FND-100 photodiode for locking of the
288: OPO.}\label{setup_twin}
289: \end{figure}
290:
291: \subsection{Experimental set-up}
292: \label{montage}
293:
294: The experimental setup is shown in Fig. \ref{setup_twin}. A
295: continuous-wave frequency-doubled Nd:YAG laser pumps a triply
296: resonant OPO above threshold, made of a semi-monolithic linear
297: cavity. The intensity reflection coefficients for the input
298: coupler are 95\% for the pump at 532 nm and almost 100\% for the
299: signal and idler beams at 1064 nm. The output mirror is highly
300: reflective for the pump and its transmission coefficient $T$ can
301: be chosen to be 5 or 10\%. With $T=5\%$, at exact triple
302: resonance, the oscillation threshold is less than 15 mW. In spite
303: of the triple resonance which generally makes OPOs much more
304: sensitive to disturbances, length and temperature controls enable
305: stable operation over more than one hour without mode-hopping.
306:
307: \subsection{"$2 \times 1$ quadrature" case}
308: \label{2quad}
309:
310:
311: \subsubsection{Twin beams}\label{sec:twin}
312:
313: Type II optical parametric oscillators are well-known to generate
314: above threshold highly quantum correlated bright twin beams.
315: Intensity correlations were experimentally observed several years
316: ago and applied to measurements of weak physical
317: effects\cite{traditional}. We describe here a recent improvement
318: of the observed correlation.
319:
320: Intensity correlations are directly measured by a balanced
321: detection scheme (Fig. \ref{setup_twin}). The signal and idler
322: orthogonally polarized beams are separated on a polarizing beam
323: splitter and detected on a pair of high quantum efficiency
324: photodiodes. With a transmission $T=10\%$ for the output mirror,
325: we have obtained a noise reduction of $9.7 \pm 0.5$ dB (89\%)
326: around 5 MHz (Fig. \ref{result_twin}), which corresponds to a
327: gemellity of $G=0.11$. To the best of our knowledge, this noise
328: reduction is the strongest reported to date in the experimental
329: quantum optics field.
330:
331: \begin{figure}[h]
332: \centerline{\includegraphics[width=0.85\columnwidth]{result_twin.eps}}
333: \caption{Normalized noise power of the intensity difference of the
334: signal and idler as a function of the frequency, after correction
335: of the electronic noise.}\label{result_twin}
336: \end{figure}
337:
338: \subsubsection{QND correlations and conditional preparation of a non-classical state}
339:
340: The observed correlation is strong enough to yield a conditional
341: variance well below 1. We will show now that such a
342: QND-correlation can be used to produce a squeezed state \emph{via}
343: conditional preparation performed on continuous variables.
344:
345: A well-known technique to generate a single photon state from twin
346: photons is to use the method of conditional measurement: if one
347: labels (1) and (2) the two modes in which the twin photons are
348: emitted, it consists in retaining in the information collected in
349: mode (1) only the counts occurring when a photon is detected in
350: mode (2) within a given time window $\Delta T$. State preparation
351: by conditional measurement can be readily extended to the
352: continuous variable regime, where the instantaneous values of the
353: signal and idler photocurrents play the role of the occurrence of
354: counts in the photon counting regime. The technique consists in
355: selecting the signal photocurrent $I_s$ only during the time
356: intervals when the idler intensity $I_i$ has a given value $I_0$
357: (within a band $\Delta I$ smaller than the photocurrent standard
358: deviation). The measurements outside these time intervals are
359: discarded. If the correlation is perfect and the interval $\Delta
360: I$ close to zero, the recorded signal intensity is perfectly
361: constant, and an intense number state is generated; in a real
362: experiment, the correlation between the signal and idler
363: photocurrents is not perfect, and the selection band $\Delta I$ is
364: finite, so that the method will not prepare a perfect number
365: state, but a sub-Poissonian state instead.
366:
367: A theoretical analysis of this protocol \cite{theo_mescond} shows
368: that in the limit where $\Delta I$ is very small the conditional
369: measurement produces a beam characterized by a Fano factor equal
370: to the conditional variance of the signal and idler beams. This
371: means that the present protocol produces a sub-Poissonian beam
372: when the signal and idler beams are QND-correlated. As shown in
373: Eq. \ref{eq:varcond2} in the limit of large correlations, the
374: residual intensity noise $F$ on the conditionally prepared state
375: will be equal to twice the gemellity ($F=V \simeq 2 G$).
376:
377:
378: \begin{figure}
379: \centerline{\includegraphics[width=.85\columnwidth]{MesCond_book.eps}}
380: \caption{\label{courbes_exp}Experimental results: (a) Idler
381: intensity fluctuations: 200 000 acquired points at 3.5 MHz
382: demodulation frequency (only 20 000 shown). (b) Corresponding
383: probability distribution. The unit is the width $\sigma_0$ of the
384: Poisson distribution of same mean intensity (shot noise). (c)
385: Values of the signal intensity conditionally selected by the value
386: of the idler intensity recorded at the same time (selection
387: bandwidth $\Delta I$ equal to 0.1 $\sigma_0$ around the mean),
388: superimposed to the corresponding experimentally measured shot
389: noise. (d) Corresponding probability distribution, compared to the
390: Poisson distribution (grey line), displaying the sub-Poissonian
391: character of the conditionally generated state. The black line is
392: a gaussian fit of the intensity distribution.}
393: \end{figure}
394:
395: Figure \ref{courbes_exp} sums up the experimental results. The
396: Fano factor $F$ of both the signal and idler beams exceeds 100
397: (20~dB above the shot noise level), and the measured gemellity $G$
398: is equal to $0.18$ ($0.14$ after correction of dark noise). The
399: ensemble of values of the signal intensity for which the idler
400: intensity falls within the selection band is given in figure
401: \ref{courbes_exp} (c): one indeed observes a significant narrowing
402: of the probability distribution below the shot noise level. With a
403: selection bandwidth $\Delta I$ equal to 0.1 times the standard
404: deviation $\sigma_0$ of a coherent state having the same power
405: (shot noise level), the conditionally prepared light state has a
406: measured Fano factor $F=0.36$, which turns out to be equal, as
407: expected, to the conditional variance of the twin beams. The
408: success rate of the conditional preparation is around 0.85\% (1700
409: points out of 200 000 are accepted). An advantage of the
410: conditional preparation using continuous variables is that one can
411: use at the same time different selection non-overlapping bands on
412: the idler beam. Each allows one to conditionally prepare a
413: different sub-Poissonian state, each having a Fano factor
414: $F=0.36$. With 200 different selection bands, the overall success
415: rate is close to $100\%$.
416:
417: \subsection{"$2\times 2$" quadratures case} \label{4quad}
418:
419:
420: \subsubsection{Entanglement below threshold}
421:
422: \begin{figure*}
423: \centerline{\includegraphics[width=1.5\columnwidth]{ManipEPR.eps}}
424: \caption{A doubled Nd:YAG laser pumps a type II OPO, below or
425: above threshold. The generated two-mode state is characterized by
426: two simultaneous homodyne detections. PD Split: split two-element
427: InGaAs photodiode for tilt-locking of the filtering
428: cavity.}\label{setup}
429: \end{figure*}
430:
431: Type-II OPO below threshold are well-known to generate entangled
432: beams. The first experimental demonstration of EPR correlations in
433: the continuous variable regime in 1992 was performed with such a
434: device\cite{ou92}. Our experimental setup is similar to the
435: previous one (Fig. \ref{setup_twin}) but the detection system is
436: now based on two simultaneous homodyne detections (Fig.
437: \ref{setup}). In order to measure the separability $\mathcal{I}$,
438: one must characterize the noise of the superposition modes
439: oriented $\pm45^{\circ}$ from the axes of the crystal:
440: \begin{eqnarray} A_{+}=\frac{A_{1}+A_{2}}{\sqrt{2}} \qquad
441: \textrm{and} \qquad A_{-}=\frac{A_{1}-A_{2}}{\sqrt{2}} \nonumber
442: \end{eqnarray}
443: Eq. \ref{eq:insep} shows that the signal and idler fields are
444: entangled as soon as these two modes have squeezed fluctuations on
445: orthogonal quadratures. The orthogonally polarized modes are
446: separated on a first polarizing beam splitter at the output of the
447: OPO. A half-wave plate inserted before this polarizing beam
448: splitter enables us to choose the fields to characterize: the
449: signal and idler modes which are entangled, or the $\pm45^{\circ}$
450: rotated modes which are squeezed. The detection setup is able to
451: characterize simultaneously the two chosen modes with the same
452: phase reference, and to measure the noise reductions either in
453: quadrature ("in phase homodyne detection") or in phase ("in
454: quadrature homodyne detection"), by inserting or not a $\lambda/4$
455: plate in the beam exiting the OPO. This configuration permits a
456: direct and instantaneous verification of the inseparability
457: criterion by simply adding the two squeezed variances.
458:
459:
460: Typical spectrum analyzer traces while scanning the local
461: oscillator phase are shown on Fig. \ref{verysmallscan}. Normalized
462: noise variances of the $\pm 45^{\circ}$ vacuum modes at a given
463: noise frequency of 3.5 MHz are superimposed for in-phase and
464: in-quadrature homodyne detections. One indeed observes, as
465: expected, correlations and anti-correlations of the emitted modes
466: on orthogonal quadratures. The homodyne detection can be locked on
467: the squeezed quadrature (Figure \ref{verysmallscan}). The observed
468: amount of simultaneous squeezing for the two rotated modes is
469: $-4.3 \pm 0.3$ dB and $-4.5 \pm 0.3$ dB below the standard quantum
470: limit ($-4.7 \pm 0.3$ dB and $-4.9 \pm 0.3$ dB after correction of
471: the electronic noise). This gives a value of the separability of
472: $\mathcal I=0.33 \pm 0.02$, well below the unit limit for
473: inseparability. With a measured value of the parameter $F$ of
474: $6.6$, one obtains a product of conditional variances of $0.42\pm
475: 0.05$, well below 1, which establishes the EPR character of the
476: measured correlations. The entanglement of formation $EOF$ of the
477: two beams is, according to formulae (\ref{EOF1}) and (\ref{EOF2}),
478: equal to $1.1 \pm 0.1\,ebits$. To the best of our knowledge, this
479: setup generates the best EPR/entangled beams to date produced in
480: the continuous variable regime.
481:
482:
483: \begin{figure}[htpb!]
484: \centerline{\includegraphics[width=.95\columnwidth,clip=]{Scan_lock.eps}}
485: \caption{(a) Normalized noise variances at 3.5 MHz of the $\pm
486: 45^{\circ}$ modes while scanning the local oscillator phase. The
487: first plot corresponds to in-phase homodyne detections and the
488: second one in-quadrature. Squeezing is well observed on orthogonal
489: quadratures. (RBW 100 kHz, VBW 1 kHz) (b) Normalized noise
490: variances at 3.5 MHz of the $\pm 45^{\circ}$ modes and
491: inseparability $\mathcal I$ for signal and idler modes. The
492: homodyne detections are in-quadrature and locked on the squeezed
493: quadratures. After correction of the electronic noise, the
494: inseparability criterion reaches $0.33\pm 0.02$. (RBW 100 kHz, VBW
495: 300 Hz).}\label{verysmallscan}
496: \end{figure}
497:
498: Non-classical properties are generally measured in the MHz range
499: of Fourier frequencies, because of the presence of large classical
500: noise at lower frequencies. In the present device significant
501: quantum correlations and EPR entanglement have been observed in
502: from 50 kHz to 10 MHz. Figure \ref{bf} gives the squeezed
503: variances for low noise frequencies, between 40 kHz and 150 kHz.
504: Let us mention that squeezing from a single type-I OPA was
505: recently reported at a record Fourier frequency of 200 Hz
506: \cite{200Hz}.
507:
508: \begin{figure}[htpb!]
509: \centerline{\includegraphics[width=0.85\columnwidth,clip=]{BF.EPS}}
510: \caption{\label{bf}Normalized noise variances from 40 kHz to 150
511: kHz of the $\pm 45^{\circ}$ modes after correction of the
512: electronic noise and inseparability criterion for signal and idler
513: modes. Squeezing and entanglement are observed down to 50 kHz.
514: (RBW 3 kHz, VBW 10 Hz)}
515: \end{figure}
516:
517: \subsubsection{Bright EPR beams above threshold and polarization
518: squeezing}
519: \label{phlock}
520:
521: A type-II OPO pumped above threshold has been theoretically
522: predicted to be a very efficient source of bright entangled and
523: EPR beams. This means that, in addition to the already
524: demonstrated intensity correlations, phase anticorrelations exist
525: in the system. However, they can be easily measured by usual
526: homodyne detection techniques only in the frequency-degenerate
527: regime. Frequency degeneracy occurs only accidentally above
528: threshold because it corresponds to a single point in the
529: experimental parameter space. Actually, up to now, no direct
530: evidence of such phase anti-correlations has been observed. In
531: 1998, Mason and Wong proposed an elegant way to achieve frequency
532: degenerate operation above threshold \cite{Mason98,Fabre99}: they
533: inserted inside the OPO cavity a birefringent plate making an
534: angle with the axis of the non-linear crystal. The induced linear
535: coupling between the signal and idler results in a locking
536: phenomenon\cite{Pikovsky}. It has been shown theoretically that in
537: such a "self-phase-locked" OPO the quantum correlations are
538: preserved for small angles of the plate and that the system
539: produces entangled states in a wide range of parameters
540: \cite{longcham1,longcham2}.
541:
542:
543: \begin{figure}
544: \centerline{\includegraphics[width=0.85\columnwidth]{Audessus_book.eps}}
545: \caption{(a) Noise power of the mode $A_-$ while scanning
546: simultaneously the phase of the local oscillator and the noise
547: frequency between 1.5 and 10 MHz. The lower trace gives the shot
548: noise level. (b) Noise power of the mode $A_+$ while scanning the
549: phase of the local oscillator, for a noise frequency between 1.5
550: and 10 MHz. The shot noise level is given by the lower trace plus
551: 3 dB.}\label{audessus}
552: \end{figure}
553:
554: In the experiment, the frequency locking phenomenon can be
555: maintained during more than hour. Degenerate operation is
556: confirmed by the fact that the generated mode has now a fixed
557: polarization: at the minimum threshold point, the generated state
558: is linearly polarized at +45$^{\circ}$. Due to the defined phase
559: relation existing now between the signal and idler fields, $A_+$
560: is a bright mode, and $A_-$ has a zero mean value. Figure
561: \ref{audessus} (a) gives the noise power of the mode $A_-$ while
562: scanning the local oscillator phase, for a transmission $T=5\%$
563: and a plate angle of 0.1$^{\circ}$. A noise reduction of 4.5dB is
564: observed. This strong noise reduction on the mode $A_-$ confirms
565: the quantum intensity correlation between the signal and idler
566: modes. Figure \ref{audessus} (b) shows the noise power of the mode
567: $A_+$ in the same condition. As the plate angle is very small, a
568: similar amount of noise reduction is expected. However, a slight
569: excess noise of 3 dB is measured for the minimal noise quadrature:
570: the phase anticorrelations appear to be slightly degraded,
571: probably by external noise sources.
572:
573: Despite this slight excess noise which prevents from reaching the
574: proof of entanglement in the OPO above threshold, the generated
575: state turns out to be squeezed in the polarization orthogonal to
576: the mean field: $A_+$ is the main mode and $A_-$ the squeezed
577: vacuum one. This condition is required to obtain a so-called
578: "polarization squeezed" state \cite{Korolkova,Bowen,Josse}. 4.5 dB
579: of polarization squeezing has been thus generated in the
580: self-phase-locked OPO. Such states have recently raised great
581: interest, in particular because of the possibility to map quantum
582: polarization state of light onto an atomic ensemble \cite{Hald}.
583:
584:
585: \begin{figure}[htpb!]
586: \centerline{\includegraphics[width=.8\columnwidth]{Fresnel.eps}}
587: \caption{\label{ellipse} Fresnel representation of the noise
588: ellipse of the $\pm 45^{\circ}$ rotated modes when the plate angle
589: is increased. Without coupling, squeezing is predicted on
590: orthogonal quadratures. The noise ellipse of the $-45^{\circ}$
591: mode rotates and the noise reduction is degraded when the coupling
592: increases while the $+45^{\circ}$ rotated mode is not affected.}
593: \end{figure}
594:
595: \begin{figure*}[t]
596: \small
597: \begin{eqnarray}
598: \Gamma_{A_{1}\,\!A_{2}} = \left( \begin{array}{cc|cc}
599: 181.192 & 0 & 179.808 & -0.255 \\
600: 0 & 0.386 & -0.255 & -0.383 \\
601: \hline 179.808 & -0.255 & 181.192 & 0\\
602: -0.255 & -0.383 & 0 & 0.386
603: \end{array} \right)
604: &\Longrightarrow& \Gamma'_{A_{1}\,\!A_{2}} = \left(
605: \begin{array}{cc|cc}
606: 180.839 & 0 & 180.161 & 0 \\
607: 0 & 0.739 & 0 & -0.736 \\
608: \hline 180.161 & 0 & 180.839 & 0\\
609: 0 & -0.736 & 0 & 0.739
610: \end{array} \right)
611: \nonumber
612: \end{eqnarray}
613: \begin{eqnarray}
614: \Gamma_{A_{+}\,\!A_{-}} = \left( \begin{array}{cc|cc}
615: 361 & 0 & 0 & 0 \\
616: 0 & 0.00277 & 0 & 0 \\
617: \hline 0& 0 & 1.383 & -0.256\\
618: 0 & 0 & -0.256 & 0.770
619: \end{array} \right)
620: &\Longrightarrow& \Gamma'_{A_{+}\,\!A_{-}} = \left(
621: \begin{array}{cc|cc}
622: 361 & 0 & 0 & 0 \\
623: 0 & 0.00277 & 0 & 0 \\
624: \hline 0& 0 & 0.677 & 0\\
625: 0 & 0 & 0 & 1.476
626: \end{array} \right)\nonumber
627: \end{eqnarray}
628: \caption{Numerical example of covariance matrix of the
629: $A_{1}$/$A_{2}$ modes and the $A_{+}$/$A_{-}$ modes before and
630: after the non-local operation for a plate angle of
631: $\rho=1.3^{\circ}$. ($\sigma=0.9$ and $\Omega=0$)} \label{matrix}
632: \end{figure*}
633:
634:
635: \section{Manipulating entanglement with polarization elements}
636: \label{manipulation}
637:
638: The self-phase-locked OPO can also be operated below threshold. It
639: produces a two-mode state with strong quantum features which
640: manifest themselves in terms of noise reduction properties in a
641: given polarization basis, and in terms of entanglement and EPR
642: correlations in another. This last section is devoted to the
643: general study of this two-mode Gaussian quantum state.
644:
645: \subsection{Manipulation of entanglement in the two-mode state produced by the type-II OPO with mode coupling}
646:
647: In a standard OPO the correlated quadratures are orthogonal to the
648: anti-correlated ones which results in squeezing of the rotated
649: modes on orthogonal quadratures. It is no more the case when a
650: linear coupling is introduced. When the plate angle increases, the
651: correlated quadratures rotate and the correlations are degraded.
652: The evolution is depicted in Fig. \ref{ellipse} through the noise
653: ellipses of the rotated (squeezed) modes. In order to maximize the
654: entanglement between the signal and idler modes, the optimal
655: quadratures have to be made orthogonal \cite{Wolf}. Such an
656: operation consists of a phase-shift of $A_{-}$ relative to
657: $A_{+}$. This transformation is passive and "non-local" in the
658: sense of the EPR argument: it acts simultaneously on the two
659: considered sub-systems. In the type II OPO, such "non-local"
660: transformations are easy to perform by inserting polarizing
661: birefringent elements in the total beam, because the two
662: polarization modes are produced by the OPO in the same transverse
663: spatial mode.
664:
665: As the generated two-mode state is not in the standard form, we
666: need to use a general measure of entanglement. Let us introduce
667: the covariance matrix formalism and the logarithmic negativity. In
668: a given mode basis, the quantum properties of the generated state,
669: of zero mean value, are completely contained in the covariance
670: matrix $\Gamma_{A\,\!B}$ defined as:
671: \begin{equation}
672: \Gamma_{A\,\!B}=\left( \begin{array}{cc}
673: \gamma_{A} & \sigma_{A\,\!B} \\
674: \sigma_{A\,\!B}^{T} & \gamma_{B}
675: \end{array} \right)\nonumber
676: \end{equation}
677: $\gamma_{A}$ and $\gamma_{B}$ are the covariance matrix of the
678: individual modes while $\sigma_{A\,\!B}$ describes the intermodal
679: correlations. The elements of the covariance matrix are written
680: $\Gamma_{ij}=\langle \delta R_{i}\delta R_{j}+\delta R_{j}\delta
681: R_{i}\rangle/2$ where
682: $R_{\{i,i=1,..,4\}}=\{X_{A},Y_{A},X_{B},Y_{B}\}$. $X$ and $Y$
683: corresponds to an arbitrary orthogonal basis of quadratures. In
684: order to measure the degree of entanglement of Gaussian states, a
685: simple computable formula of the logarithmic negativity
686: $E_{\mathcal{N}}$ has been obtained \cite{vidal} (see also
687: \cite{Adesso} for a general overview). $E_{\mathcal{N}}$ can be
688: easily evaluated from the largest positive symplectic eigenvalue
689: $\xi$ of the covariance matrix which can be obtained from
690: \begin{eqnarray}
691: \xi^{2}=\frac{1}{2}(D-\sqrt{D^{2}-4\det\Gamma_{A\,\!B}}\,)
692: \end{eqnarray}
693: with
694: \begin{eqnarray}
695: D=\det\gamma_{A}+\det\gamma_{B}-2\det\sigma_{A\,\!B}
696: \end{eqnarray}
697: The two-mode state is entangled if and only if $\xi<1$. The
698: logarithmic negativity can thus be expressed by
699: $E_{\mathcal{N}}=-\log_{2}(\xi)$. The maximal entanglement which
700: can be extracted from a given two-mode state by passive operations
701: is related to the two smallest eigenvalues of $\Gamma$,
702: $\lambda_1$ and $\lambda_2$, by
703: $E_{\mathcal{N}}^{max}=-\log_{2}(\lambda_1\lambda_2)/2$
704: \cite{Wolf}.
705:
706: We give here a numerical example for realistic experimental values
707: $\rho=1.3^{\circ}, \sigma=0.9$ and $\Omega=0$, where $\rho$ stands
708: for the plate angle, $\sigma$ the pump power normalized to the
709: threshold and $\Omega$ the noise frequency. The covariance
710: matrices for the $A_{1}$/$A_{2}$ modes and for the $A_{+}$/$A_{-}$
711: modes are given in Fig. \ref{matrix} with and without the
712: phase-shift. The matrix of the $A_{+}$/$A_{-}$ modes are
713: well-suited to understand the behavior of the device. At first,
714: the intermodal blocks are zero, showing that these two modes are
715: not at all correlated and consequently are the most squeezed modes
716: of the system: there is no way to extract more squeezing. But one
717: can also note that the diagonal blocks are not diagonalized
718: simultaneously. This corresponds to the tilt angle $\theta$ of the
719: squeezed quadrature of $A_{-}$. A phase-shift of the angle
720: $\theta$ permits to diagonalize simultaneously the two blocks and
721: to obtain squeezing on orthogonal quadratures. The logarithmic
722: negativity $E_{\mathcal{N}}$ has increased in the transformation
723: from $4.06$ to $4.53$. The maximal entanglement available has been
724: extracted in this way as
725: $E_{\mathcal{N}}^{max}=-\log_{2}(\lambda_1\lambda_2)/2=4.53$.
726:
727:
728: \subsection{Experimental optimization of entanglement}
729:
730: Let us now describe how to experimentally optimize the EPR
731: entanglement generated by the self-phase-locked OPO below
732: threshold.
733:
734: In order to extract the maximal entanglement, one must perform an
735: appropriate phase-shift on the rotated modes. This is achieved by
736: using an association of one $\lambda/2$ and one $\lambda/4$ plates
737: added at the output of the OPO. The double homodyne detection we
738: have developed is necessary in order to be able to characterize
739: simultaneously the two modes with the same phase reference.
740:
741:
742: \begin{figure}[htpb!]
743: \centerline{\includegraphics[width=.85\columnwidth]{aveci_ncetcorrige.EPS}}
744: \caption{\label{corrige}Normalized noise variances at 3.5 MHz of
745: the rotated modes while scanning the local oscillator phase for a
746: plate angle $\rho=0.3^{\circ}$, before and after the non-local
747: operation. The homodyne detections are in-quadrature. After this
748: operation, squeezing is observed on orthogonal quadratures.}
749: \end{figure}
750:
751: Figure \ref{corrige} displays the normalized noise variances of
752: the rotated modes for a plate angle of $\rho=0.3^{\circ}$, before
753: and after the phase-shift. The homodyne detections are operated in
754: quadrature so that squeezing on orthogonal quadratures is observed
755: simultaneously on the spectrum analyzers. After the operation is
756: performed, squeezing is obtained on orthogonal quadratures as in a
757: standard type-II OPO without mode coupling. Experimentally, the
758: logarithmic negativity goes from 1.13 to 1.32, showing that we are
759: able to extract more quantum resource from the state after the
760: operation.
761:
762: \section{Conclusion}
763:
764: We have seen that type II triply resonant OPO produce in a very
765: stable way the strongest intensity correlation and EPR
766: entanglement to date. Phase-locked, frequency degenerate operation
767: can be obtained using an intracavity birefringent plate. This
768: locking permits the experimental realization of homodyne detection
769: of the quadratures even when the system is operated above
770: threshold. This result opens a very promising way to the direct
771: generation of intense entangled beams and offers a new and simple
772: method to achieve strong polarization squeezing. Below threshold,
773: the self-phase-locked OPO exhibits a very rich and interesting
774: behavior which provides a good insight into entanglement
775: manipulation by passive operations. This opens the way to the
776: manipulation and optimization of quantum properties in highly
777: multimode Hilbert spaces.
778:
779: \begin{thebibliography}{99}
780: \bibitem{criteres} N. Treps, C. Fabre, Laser Physics \textbf{15}, 187 (2005)
781:
782: \bibitem{optlett_above} J. Laurat, L. Longchambon, T. Coudreau, C. Fabre, Opt. Lett \textbf{30}, 1177 (2005)
783:
784: \bibitem{mescond} J. Laurat, T. Coudreau, N. Treps, A. Ma\^{\i}tre, C.
785: Fabre, Phys. Rev. Lett. \textbf{91}, 213601 (2003)
786:
787: \bibitem{theo_mescond} J. Laurat, T. Coudreau, N. Treps, A. Ma\^itre, C.
788: Fabre, Phys. Rev. A, \textbf{69}, 33808 (2004)
789:
790: \bibitem{pra_bf} J. Laurat, T. Coudreau, G. Keller, N. Treps, C.
791: Fabre, Phys. Rev. A \textbf{70}, 042315 (2004)
792:
793: \bibitem{pra_epr} J. Laurat, T. Coudreau, G.
794: Keller, N. Treps, C. Fabre, Phys. Rev. A \textbf{71}, 022313
795: (2005)
796:
797: \bibitem{traditional} A. Heidmann, R.J. Horowicz, S. Reynaud,
798: E. Giacobino, C. Fabre, G. Camy, Phys. Rev. Lett. \textbf{59},
799: 2555 (1987); J. Mertz, T. Debuisschert, A. Heidmann, C. Fabre, E.
800: Giacobino, Opt. Lett. \textbf{16}, 1234 (1991); J. Gao, F. Cui, C.
801: Xue, C. Xie, K. Peng, Opt. Lett. \textbf{23}, 870 (1998)
802:
803: \bibitem{qnd} P. Grangier, J.-M Courty, S. Reynaud Opt. Commun.
804: \textbf{89}, 99 (1992)
805:
806: \bibitem{duan} L.-M. Duan, G. Giedke, J. I. Cirac, P. Zoller,
807: Phys. Rev. Lett. \textbf{84}, 2722 (2000)
808:
809: \bibitem{giedke}G. Giedke, M.M. Wolf, O. Kr\"{u}ger, R.F. Werner, J.I. Cirac, Phys. Rev. Lett \textbf{91}, 107901 (2003)
810:
811: \bibitem{vidal} G. Vidal, R.F. Werner, Phys. Rev. A \textbf{65},
812: 032314 (2002)
813:
814: \bibitem{epr35} A. Einstein, B. Podolsky, N. Rosen,
815: Phys. Rev. \textbf{47}, 777 (1935)
816:
817: \bibitem{reid} M. Reid, P. Drummond, Phys. Rev. Lett. \textbf{60},
818: 2731 (1989); M. Reid, Phys. Rev. A \textbf{40}, 913 (1989); M.
819: Reid, P. Drummond, Phys. Rev A \textbf{41}, 3930 (1991)
820:
821: \bibitem{ou92} Z.Y. Ou, S.F. Pereira, H.J. Kimble, K.C. Peng,
822: Phys. Rev. Lett. \textbf{68}, 3663 (1992)
823:
824: \bibitem{200Hz}K. McKenzie, N. Grosse, W.P. Bowen, S.E. Whitcomb, M.B. Gray, D.E. McClelland, P.K. Lam, Phys. Rev. Lett \textbf{93},
825: 161105 (2004)
826:
827: \bibitem{Mason98} E.J. Mason, N.C. Wong, Opt. Lett. \textbf{23}, 1733 (1998)
828:
829: \bibitem{Fabre99} C. Fabre, E.J. Mason, N.C Wong, Optics Communications \textbf{170},
830: 299 (1999)
831:
832: \bibitem{Pikovsky} A. Pikovsky, M. Rosenblum, J. Kurths, Synchronization (Cambridge University Press,
833: 2001)
834:
835: \bibitem{longcham1} L. Longchambon, J. Laurat, T. Coudreau, C. Fabre, Eur. Phys. J.
836: D \textbf{30}, 279 (2004)
837:
838: \bibitem{longcham2} L. Longchambon, J. Laurat, T. Coudreau, C. Fabre, Eur. Phys. J.
839: D \textbf{30}, 287 (2004)
840:
841: \bibitem{Korolkova} N. Korolkova, G. Leuchs, R. Loudon, T.C. Ralph, C. Silberhorn, Phys. Rev. A \textbf{65},
842: 052306 (2002)
843:
844: \bibitem{Bowen} W.P. Bowen, R. Schnabel, H.-A. Bachor, P.K. Lam, Phys. Rev. Lett. \textbf{88},
845: 093601 (2002)
846:
847: \bibitem{Josse} V. Josse, A. Dantan, L. Vernac, A. Bramati, M. Pinard, E. Giacobino, Phys. Rev. Lett. \textbf{91},
848: 103601 (2003)
849:
850: \bibitem{Hald} J. Hald, J.L. S\o rensen, C. Schori, E.S. Polzik, Phys. Rev. Lett. \textbf{83},
851: 1319 (1999)
852:
853: \bibitem{Wolf} M.M. Wolf, J. Eisert, M.B. Plenio, Phys. Rev. Lett \textbf{90},
854: 047904 (2003)
855:
856: \bibitem{Adesso} G. Adesso, A. Serafini, F. Illuminati, Phys. Rev. A \textbf{70},
857: 022318 (2004)
858:
859: \end{thebibliography}
860:
861: \end{document}
862: