1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2: %

3: \documentclass[12pt]{iopart}

4: % Uncomment next line if AMS fonts required

5: \usepackage{iopams}

6: \usepackage{setstack}

7: \usepackage[dvips]{graphicx}

8: %\usepackage{amsmath}

9: 

10: \begin{document}

11: 

12: \title[Polarization squeezing]{Polarization squeezing in a 4-level system}

13: 

14: \author{V. Josse\footnote[3]{To whom correspondence should be addressed

15: (josse@spectro.jussieu.fr)}, A. Dantan, A. Bramati, M. Pinard, E.

16: Giacobino}

17: 

18: 

19: \address{Laboratoire Kastler Brossel, Universit\'{e} Pierre et Marie Curie,

20: 4 place Jussieu, F75252 Paris Cedex 05, France}

21: 

22: 

23: \begin{abstract}

24: We present a theoretical study of an ensemble of X-like 4-level

25: atoms placed in an optical cavity driven by a linearly polarized

26: field. We show that the self-rotation (SR) process leads to

27: polarization switching (PS). Below the PS threshold, both the mean

28: field mode and the orthogonal vacuum mode are squeezed. We provide

29: a simple analysis of the phenomena responsible for the squeezing

30: and trace the origin of vacuum squeezing not to SR, but to crossed

31: Kerr effect induced by the mean field. Last, we show that this

32: vacuum squeezing can be interpreted as \textit{polarization

33: squeezing}.

34: \end{abstract}

35: 

36: %Uncomment for PACS numbers title message

37: \pacs{42.50.Lc, 42.65.Pc, 42.50.Dv}

38: 

39: % Uncomment for Submitted to journal title message

40: %\submitto{}

41: 

42: % Comment out if separate title page not required

43: %\maketitle

44: 

45: \section{Introduction}

46: 

47: The principal limit in high precision measurements and optics

48: communication is given by the quantum fluctuations of light. For

49: several years, in order to beat the standard quantum limit, a

50: number of methods consisting in generating squeezed states of

51: light have been developed \cite{davidovich}. In connection with

52: quantum information technology the quantum features of the

53: polarization of light has raised a lot of attention. The

54: generation of polarization squeezing has been achieved

55: experimentally by mixing an OPO-produced squeezed vacuum with a

56: coherent field \cite{grangier,polzik}, or more recently by mixing

57: two independent OPA-originated squeezed beams on a polarizing

58: beamsplitter \cite{bachor}. Several schemes using Kerr-like media

59: have also been proposed \cite{chirkin2,boivin, korolkova3}, and

60: very recently, Matsko et al. proposed to propagate a linearly

61: polarized field through a self-rotative atomic medium to produce

62: vacuum squeezing on the orthogonal polarization \cite{matsko}. The

63: Kerr-like interaction between cold cesium atoms placed in a high

64: finesse optical cavity and a circularly polarized field has been

65: studied in our group and a field noise reduction of 40\% has been

66: obtained \cite{lambrecht,coudreau}. We recently observed

67: experimental evidence of polarization squeezing when the incoming

68: polarization is linear \cite{josse}. In this paper, we present a

69: theoretical investigation of polarization squeezing generated by

70: an ensemble of X-like 4-level atoms illuminated by a linearly

71: polarized field. To be as realistic as possible, the experimental

72: parameters values of Ref \cite{lambrecht,coudreau,josse} are taken

73: as references. In the first part of the paper, we give a detailed

74: study of the steady state and show that self-rotation is

75: responsible for polarization switching and saturation leads to

76: tristability. We derive simple analytical criteria for the

77: existence of elliptically polarized solutions and the stability of

78: the linearly polarized solution. This steady state study is

79: essential to figure out the interesting working points for

80: squeezing. In the second part, we focus on the case in which the

81: polarization remains linear (below the PS threshold) and show that

82: both the linearly polarized field mode and the orthogonal vacuum

83: mode are squeezed. Analytical spectra are derived in the low

84: saturation limit and enable a clear discussion of the physical

85: effects responsible for polarization squeezing; in particular, we

86: demonstrate that self-rotation is associated to strong atomic

87: noise terms preventing vacuum squeezing at low frequency. On the

88: other hand, saturation accounts for the squeezing on the mean

89: field and crossed-Kerr effect enables to retrieve vacuum squeezing

90: at high frequency. The analytical results are compared with a full

91: quantum calculation. Finally, we derive the Stokes parameters

92: \cite{chirkin} and relate their fluctuations to those of the

93: vacuum field. The vacuum squeezing obtained is then equivalent to

94: the squeezing of one Stokes parameter, the so-called

95: \textit{polarization squeezing} \cite{korolkova}.

96: 

97: 

98: \section{The model}

99: 

100: \begin{figure}[h]

101:   \centering

102:   \includegraphics[width=7cm]{fig1}

103:   \caption{X-like 4-level configuration.}\label{fig1}

104: \end{figure}

105: 

106: 

107: The system considered in this paper is a set of N 4-level cold

108: atoms interacting in an optical cavity driven by a linearly

109: polarized field as represented in Fig \ref{fig1}. We denote

110: $A_\pm$ the slowly-varying envelope operators associated with the

111: $\sigma_{\pm}$ components of the light \cite{fabre}. They are

112: defined from the standard linear polarization components

113: 

114: \begin{equation}

115: A_+=-\frac{A_x-iA_y}{\sqrt{2}}\;\;,\;\;\;A_-=\frac{A_x+iA_y}{\sqrt{2}}

116: \label{passage}

117: \end{equation}

118: 

119: The atomic frequencies are both equal to

120: $\omega_{at}=\omega_{13}=\omega_{24}$. The field frequency is

121: $\omega$ and the detunings from atomic resonance are equal on both

122: transitions to $\Delta=\omega_{at}-\omega$. The 4-level system is

123: described using collective operators for the N atoms of the

124: ensemble, the optical dipoles being defined in the rotating frame

125: associated to the laser frequency (e.g.

126: $\sigma_{14}=\sum\limits_{i=1}^N e^{i\omega t}|1\rangle_i\langle

127: 4|_i$). The coupling constant between the atoms and the field is

128: defined by $g=\mathcal{E}_0 d/\hbar$, where $d$ is the atomic

129: dipole and

130: $\mathcal{E}_0=\sqrt{\hbar\omega/2\epsilon_0\mathcal{S}c}$. With

131: this definition, the mean square value of the field is expressed

132: in number of photons per second. As in Fig \ref{fig1}, the

133: population of level 3 decays with rate $2\gamma_{\perp}$ on level

134: 1 and with rate $2\gamma_{//}$ on level 2, the dipole decay rate

135: being $\gamma=\gamma_{//}+\gamma_{\perp}$. We consider the case of

136: saturated optical pumping and neglect the relaxation rate of the

137: ground states populations. This approximation is well verified for

138: alkali cold atoms \cite{chu}. With these conventions, the

139: atom-field hamiltonian is

140: 

141: \begin{equation}

142: H=\hbar

143: g[A_+\sigma_{41}+A_+^{\dagger}\sigma_{14}+A_-\sigma_{32}+A_-^{\dagger}\sigma_{23}]

144: \end{equation}

145: 

146: The atomic evolution is then governed by a set of quantum

147: Heisenberg-Langevin equations

148: 

149: \begin{eqnarray}

150: \frac{d\sigma_{14}}{dt} & = & -(\gamma+i\Delta)\sigma_{14}-ig

151: A_+(\sigma_{11}-\sigma_{44})+F_{14}\\

152: \frac{d\sigma_{23}}{dt} & = & -(\gamma+i\Delta)\sigma_{23}-ig

153: A_-(\sigma_{22}-\sigma_{33})+F_{23}\\

154: \frac{d\sigma_{11}}{dt} & = &

155: 2\gamma_{\perp}\sigma_{33}+2\gamma_{//}\sigma_{44}-ig

156: (A_+^{\dagger}\sigma_{14}-A_+\sigma_{41})+F_{11}\\

157: \frac{d\sigma_{22}}{dt} & = &

158: 2\gamma_{//}\sigma_{33}+2\gamma_{\perp}\sigma_{44}-ig

159: (A_-^{\dagger}\sigma_{23}-A_-\sigma_{32})+F_{22}\\

160: \frac{d\sigma_{33}}{dt} & = & -2\gamma\sigma_{33}+ig

161: (A_-^{\dagger}\sigma_{23}-A_-\sigma_{32})+F_{33}\\

162: \frac{d\sigma_{44}}{dt} & = & -2\gamma\sigma_{44}+ig

163: (A_+^{\dagger}\sigma_{14}-A_+\sigma_{41})+F_{44}

164: \end{eqnarray}

165: 

166: Note that we have not reproduced all the atomic equations, but

167: only those of interest for the following. The Langevin operators

168: $F_{\mu\nu}$ are $\delta$-correlated and their correlation

169: functions are calculated via the quantum regression theorem

170: \cite{cohen}. We consider a ring cavity with $T$ the transmission

171: of the cavity coupling mirror, $\omega_c$ the cavity resonance

172: frequency closest to $\omega$ and $\tau$ the cavity round-trip

173: time. The cavity dephasing is $\Phi_c=(\omega-\omega_c)\tau$. The

174: incoming quantum fields are $A_{\pm}^{in}$ and the field equations

175: read

176: 

177: \begin{eqnarray}

178: \tau\frac{dA_+}{dt} & = &

179: -(T/2+i\Phi_c)A_+-ig\sigma_{14}+\sqrt{T}A_+^{in}\label{Aplus}\\

180: \tau\frac{dA_-}{dt} & = &

181: -(T/2+i\Phi_c)A_--ig\sigma_{23}+\sqrt{T}A_-^{in}\label{Amoins}

182: \end{eqnarray}

183: 

184: \section{Steady-state}

185: 

186: \subsection{Atomic steady state}

187: The atomic steady state is readily obtained by setting the time

188: derivatives to zero and using the fact that a Langevin operator

189: mean value is zero. Defining saturation parameters $s_{\pm}$ for

190: both polarizations,

191: 

192: \begin{eqnarray}

193: s_{\pm}=\frac{2g^2|\langle

194: A_{\pm}\rangle|^2}{\Delta^2+\gamma^2}=\frac{\Omega_{\pm}^2/2}{\Delta^2+\gamma^2},\label{S}

195: \end{eqnarray}

196: 

197: the atomic steady state is given by

198: 

199: \begin{eqnarray}

200: \langle \sigma_{14}\rangle=\frac{-igN\langle

201: A_+\rangle}{\gamma+i\Delta}\frac{s_-}{s_++s_-}\frac{1}{1+S} \;,

202: \langle \sigma_{23}\rangle=\frac{-igN\langle A_-\rangle}{\gamma+i\Delta}\frac{s_+}{s_++s_-}\frac{1}{1+S}\\

203: \langle \sigma_{11}\rangle=

204: N\frac{s_-}{s_++s_-}\frac{1+s_+/2}{1+S} \;\;\;\;\;\;\;\;\;\;,

205: \langle \sigma_{22}\rangle=

206: N\frac{s_+}{s_++s_-}\frac{1+s_-/2}{1+S}\\

207: \langle \sigma_{33}\rangle=\langle \sigma_{44}\rangle =

208: \frac{N}{4}\frac{S}{1+S}\;\;\;with:\;\;\;

209: S=\frac{2s_+s_-}{s_++s_-}

210: \end{eqnarray}

211: 

212: $\Omega_{\pm}$ are the Rabi frequencies and $S$ is the coupling

213: saturation parameter which plays a symmetrical role with respect

214: to both polarization components. For an x-polarized field,

215: $S=s_+=s_-=s_x/2$ is directly related to the intracavity field

216: intensity.

217: 

218: \subsection{Polarization switching}\label{ps}

219: It is well known that such a coupled system may exhibit

220: polarization switching when driven by a linearly polarized field

221: \cite{walls,elisabeth}. In fact, the intracavity field intensities

222: depend on the atomic dephasings $\Phi_{\pm}$ and absorptions

223: $a_{\pm}$

224: 

225: \begin{eqnarray}

226: \Phi_{\pm}=2\Phi_0\frac{s_{\mp}}{s_++s_-}\frac{1}{1+S} & \;\;,\;\; & a_{\pm}=2a_0\frac{s_{\mp}}{s_++s_-}\frac{1}{1+S}\\

227: \Phi_0=\frac{Ng^2\Delta}{2(\Delta^2+\gamma^2)} & \;\;,\;\; &

228: a_0=\frac{Ng^2\gamma}{2(\Delta^2+\gamma^2)}

229: \end{eqnarray}

230: 

231: with $\Phi_0$ and $a_0$ the linear dephasing and absorption in the

232: absence of saturation. These quantities depend in turn on the

233: intensities to yield a complex coupled system. In order to derive

234: analytical criteria for polarization switching, we follow the

235: method given in \cite{elisabeth} and decompose dephasings and

236: losses into their linear and non-linear parts,

237: 

238: \begin{eqnarray}

239: \Phi_{\pm} =  \Phi_l\pm\Phi_{SR}\;\;\; & with\;\;\; &

240: \Phi_l=\frac{\Phi_0}{1+S}\;,\;\; \Phi_{SR}=\Phi_lx_{SR}\label{phil}\\

241: a_{\pm} = a_l\pm a_{SR}\;\;\; & with\;\;\; &

242: a_l=\frac{a_0}{1+S}\;,\;\; a_{SR}=a_lx_{SR}\label{al}

243: \end{eqnarray}

244: 

245: where $\Phi_{SR}$ and $a_{SR}$ are the non-linear circular

246: birefringence and dichroism, related to the ellipticity $\epsilon$

247: \cite{huard}

248: 

249: \begin{equation}

250: x_{SR}=\frac{s_--s_+}{s_++s_-}=-\sin 2\epsilon \label{epsilon}

251: \end{equation}

252: 

253: Thus, as pointed out in the literature \cite{savage,yashchuk}, the

254: optical pumping induces non-linear self-rotation (SR) of

255: elliptically polarized light. It will be shown in the next section

256: that this effect is responsible for PS in a cavity configuration.

257: Let us first focus on the solution for the $\sigma_{\pm}$

258: components. Normalizing all the dephasings and absorptions by

259: $T/2$ ($\delta_{j}=2\Phi_{j}/T$ and $\alpha_{j}=2a_{j}/T$), Eqs

260: (\ref{Aplus}),(\ref{Amoins}) read in steady state

261: 

262: \begin{equation}

263: s_{\pm}=\frac{s_{max}}{(1+\alpha_l\pm

264: \alpha_{SR})^2+(\delta_l\pm\delta_{SR}-\delta_c)^2} \label{spm}

265: \end{equation}

266: 

267: with $s_{max}=2/T s_x^{in}$ the maximal intracavity intensity in

268: the absence of absorption. Replacing (\ref{spm}) in

269: (\ref{epsilon}), we derive the equation for $x_{SR}$: non zero

270: solutions correspond to elliptically polarized states. After

271: straightforward calculations, we obtain

272: 

273: \begin{eqnarray}

274: x_{SR}=0 \;\;\;\;&or\;\;\;\;& (\alpha_l^2+\delta_l^2)x_{SR}^2=

275: \delta_l^2+\alpha_l^2-\delta_c^2-1\label{xsr}

276: \end{eqnarray}

277: 

278: The first trivial solution corresponds to the linearly polarized

279: field. It follows from the second equation and

280: (\ref{phil}),(\ref{al}) that elliptically polarized states may

281: exist as soon as the existence criterion $C_{ex}$ is satisfied

282: 

283: \begin{equation}

284: C_{ex}=\frac{\delta_0^2+\alpha_0^2}{(1+S)^2}-\delta_c^2-1\geq

285: 0\;\;\;\;\;\;(s_+\neq s_-) \label{critereexistence}

286: \end{equation}

287: 

288: Note that the absorption brings a positive contribution to the

289: existence of asymmetrical solutions: this is due to the fact that

290: non-linear circular dichroism produces "self-elliptization" of the

291: field. However, this criterion gives no information on the

292: stability of the solutions. In order to get some physical insight

293: into this complicated problem it is useful to look at the

294: evolution of the linearly polarized solution.

295: 

296: 

297: \subsection{Interpretation of polarization switching}\label{opo}

298: In this section, we give a simple interpretation of PS as the

299: threshold for laser oscillations. Let us consider the linearly

300: polarized solution along the x axis. The adiabatic elimination of

301: the atomic variables leads to

302: 

303: \begin{eqnarray}

304: \frac{1}{\kappa}\frac{dA_y}{dt}=-(1+i\delta_c)A_y+(i\delta_l-\alpha_l)A_y-

305: (\delta_{SR}+i\alpha_{SR})A_x+\frac{2}{\sqrt{T}}A_y^{in}\label{equationAy2}

306: \end{eqnarray}

307: 

308: where $\kappa=T/2\tau$ is the intracavity field decay rate. In

309: (\ref{equationAy2}) all terms have zero mean value and are of

310: order 1 in fluctuations ($\langle A_x\rangle\neq 0$). Using

311: $x_{SR}A_x=i(A_y-A_x^2/|A_x|^2A_y^{\dagger})$, one obtains

312: 

313: \begin{eqnarray}

314: \frac{1}{\kappa}\frac{dA_y}{dt}=-(1+i\delta_c)A_y+(i\delta_l-\alpha_l)\frac{A_x^2}{|A_x|^2}A_y^{\dagger}

315: +\frac{2}{\sqrt{T}}A_y^{in} \label{equationAy3}

316: \end{eqnarray}

317: 

318: Owing to SR the fluctuations of the orthogonal mode undergo a

319: phase dependent gain. A similar equation has already been derived

320: in previous theoretical works in a single pass scheme

321: \cite{matsko}. In our configuration the presence of the cavity

322: will lead to oscillations of this mode as soon as the phase

323: sensitive gain is larger than the losses. This condition may be

324: expressed as follows

325: 

326: \begin{equation}

327: C_{PS}=\frac{\delta_0^2+\alpha_0^2}{(1+s_x/2)^2}-\delta_c^2-1\geq

328: 0 \label{criterestabilite}

329: \end{equation}

330: 

331: Obviously, the linearly polarized solution is not stable when

332: $C_{PS}\geq 0$. However, the adiabatical elimination of the atomic

333: variables does not \textit{a priori} take all causes for

334: instability into account. Yet, we checked that this threshold

335: analysis was consistent with a numerical calculation of the

336: atom-field stability matrix. In the following we use $C_{PS}$ as a

337: stability criterion for the linearly polarized solution.

338: Nevertheless, it does not yield information on the stability of

339: the elliptically polarized solutions, which has been evaluated numerically.\\

340: Besides, the ability of a system to produce squeezing being

341: closely related to its static properties, the fluctuations of the

342: vacuum field are expected to be strongly modified in the vicinity

343: of the PS threshold. Since Eq (\ref{equationAy3}) is similar to

344: that of a degenerate optical parametric oscillator (OPO) below the

345: threshold \cite{gardiner}, perfect squeezing could be obtained via

346: SR. However, the atomic noise is not included in

347: (\ref{equationAy3}) and is to be carefully evaluated.

348: 

349: \subsection{Optical pumping regime}\label{opticalpumping}

350: 

351: \begin{figure}

352:   \centering

353:   \includegraphics[width=10cm]{fig2}

354:   \caption{Resonance curves in the optical pumping regime.

355:   The parameters are $\delta_0=5$, $s_{max}=0.01$. The cavity dephasing

356:   corresponding to PS is $\delta_ {PS}=4.85$, close to $\delta_0$

357:   as given by the PS criterion. When the cavity detuning is scanned from the

358:   right, the linear solution is stable until $\delta_c=\delta_{PS}$

359:   and unstable afterwards. Then

360:   the elliptically polarized solutions, $s_+$ and $s_-$,

361:   become stable for $|\delta_c|\leq\delta_{PS}$.

362:   We plot also the resonance peaks (dashed line) for the cavity with $0$ or $N$ atoms, in the

363:   absence of SR phenomenon.}\label{fig2}

364: \end{figure}

365: 

366: PS is caused by a competition between the two $\sigma_{\pm}$

367: optical pumping processes. We can understand the main features of

368: this effect by restraining ourselves to the case where absorption

369: and saturation are negligible: $\Delta\gg\gamma$ and $s_x\ll 1$.

370: Neglecting the excited state populations, the optical pumping

371: equations for the ground state populations are

372: 

373: \begin{eqnarray}

374: \left(\frac{d\sigma_{11}}{dt}\right)_{pumping}

375: =-\gamma_{\perp}s_+\sigma_{11}+\gamma_{\perp}s_-\sigma_{22}\label{op1}\\

376: \left(\frac{d\sigma_{22}}{dt}\right)_{pumping}

377: =-\gamma_{\perp}s_-\sigma_{22}+\gamma_{\perp}s_+\sigma_{11}\label{op2}

378: \end{eqnarray}

379: 

380: so that the $\sigma_+$ component tends to pump the atoms into

381: level 2, the $\sigma_-$ into 1, and, in steady state,

382: $\sigma_{11}\propto s_-$ and $\sigma_{22}\propto s_+$. The

383: circular birefringence $\delta_{SR}$ is proportional to the ground

384: state population difference, and consequently, to the intensity

385: difference $s_+-s_-$ [see (\ref{epsilon})]. This simple analysis

386: allows for relating self-rotation to competitive optical

387: pumping and will help us interpret the resonance curves.\\

388: Under the previous conditions both criteria

389: (\ref{critereexistence}) and (\ref{criterestabilite}) are

390: equivalent and it follows that the linearly polarized solution

391: bifurcates into an elliptically polarized state for

392: $|\delta_c|\leq\delta_{PS}=\sqrt{\delta_0^2-1}$. Consequently, PS

393: is observed as soon as the linear dephasing is greater than half

394: the cavity bandwidth ($\delta_0\geq 1$). This represents an easily

395: accessible condition from an experimental point of view : in our

396: cesium experiment using a magneto-optical trap

397: \cite{lambrecht,coudreau}, the number of atoms interacting with

398: the light is $N\simeq 7\;10^6$. To find realistic experimental

399: parameters, we assimilate each one of our X-model transitions to

400: the transition $6S_{1/2}\;(F=4)-6P_{3/2}\;(F=5)$ of the $D_2$ line

401: of $^{133}Cs$, for which $\gamma/2\pi=2.6$ MHz. The square of the

402: coupling constant $g$ is proportional to the ratio of the

403: diffusion section at resonance to the transversal surface $S=0.1$

404: mm$^2$ of the beam, $g^2=3\gamma\lambda^2/4\pi S=4.24$ Hz. The

405: cavity transmission is 10\%. To obtain a sufficiently high

406: non-linearity, keeping the absorption low, a good detuning is

407: $\Delta\simeq 20\gamma$, so that an approximate value for the

408: linear detuning is $\delta_0\simeq 5$. Note that the saturation

409: parameters of (\ref{S}) are simply

410: 

411: \begin{equation}

412: s_{\pm}=\frac{\gamma^2}{\gamma^2+\Delta^2}\frac{I_{\pm}}{I_{sat}}

413: \end{equation}

414: 

415: The saturation intensity being

416: $I_{sat}=\epsilon_0c\gamma^2\hbar^2/d^2=1.05$ mW/cm$^2$ \cite{lambrecht},

417: typical values for $s_{\pm}$ are 0.1-1.\\

418: 

419: In Fig \ref{fig2} are represented the admissible intensities for

420: the $\sigma_+$ and $\sigma_-$ components versus the cavity

421: detuning for typical experimental values of the parameters. The

422: peak centered on $\delta_c=\delta_0$ corresponds to the

423: symmetrical solution. When the cavity is scanned from right to

424: left, the linearly polarized field ($s_+=s_-$) intensity increases

425: until the PS threshold is reached

426: ($\delta_c=\delta_{PS}\simeq\delta_0$). Then one elliptically

427: polarized state becomes stable. The predominant circular

428: component, say $\sigma_+$, creates, via the optical pumping

429: process (\ref{op1}-\ref{op2}), a positive orientation of the

430: medium $\sigma_{11}\simeq 0$, $\sigma_{22}\simeq N$. Since the

431: atomic dephasing decreases to zero for the $\sigma_+$ component

432: ($\delta_+\simeq 0$), as if it were propagating in an empty

433: cavity. Hence, the solution draws close to the zero-dephasing

434: peak, that is, close to resonance in the range

435: $|\delta_c|\leq\delta_{PS}$. On the other hand the $\sigma_-$

436: component "sees" all the atoms ($\delta_-\simeq 2\delta_0$) and

437: breaks down to fit the peak centered on $\delta_c=2\delta_0$,

438: which is far from resonance. In order to illustrate this

439: interpretation of the resonance curves, the two Airy peaks

440: centered on $\delta_c=0$ and $\delta_c=2\delta_0$ are represented

441: in Fig \ref{fig2}. As the cavity detuning is decreased both

442: asymmetrical solutions reunite when the criterion

443: (\ref{criterestabilite}) is no longer satisfied and the linear

444: solution becomes stable again. These simple interpretations will

445: help us understand the much more complex general case, when

446: absorption and saturation come into play.\\

447: As discussed in Ref \cite{elisabeth}, taking into account the

448: ground state relaxation rate $\gamma_0$ yields tristability in the

449: unsaturated optical pumping regime

450: ($\gamma_{\perp}s_{\pm}\ll\gamma_0$). We will now show that the

451: optical saturation also leads to tristability.

452: 

453: \subsection{Tristability}\label{kerrstat}

454: 

455: \begin{figure}

456:   \flushright

457:   \includegraphics[width=13cm]{fig3}

458:   \caption{Upper plot: resonance curves for $\delta_0=7$,

459:   $\Delta=20\gamma$ and $s_{max}=2$. The linear absorption is $\alpha_0=0.35$. Dashed

460:   parts indicate unstable solutions. The switching occurs for

461:   $\delta_{PS}=2.6$ and $S_{PS}=1.5$. $C_{ex}=0$ for $\delta_{ex}=5.6$, so that

462:   the tristability range is $2.6\leq\delta_c\leq5.6$.

463:   The arrows on the hysteresis cycle correspond to increasing and decreasing cavity

464:   detuning scan.

465:   Below are plotted the two criteria: $C_{PS}$ giving the stability

466:   of the linear solution (plain) and

467:   $C_{ex}$ giving the existence of asymmetrical solutions (dashed).}\label{fig3}

468: \end{figure}

469: 

470: It is well-known that saturation may induce multistability for the

471: linearly polarized field \cite{gibbs} in our configuration and

472: substantially modify the steady state. When the non-linearity is

473: sufficient, there may be three possible values for the x-polarized

474: field intensity. Therefore, saturation is an additional cause of

475: instability for the symmetrical solution. In Fig \ref{fig3}, we

476: plotted the same curves as in Fig \ref{fig2}, but for higher

477: values of $s_{max}$ and $\delta_0$. As expected, the linearly

478: polarized state solution is distorted as a consequence of the

479: non-linear effect. The effect of absorption is also clear: whereas

480: the symmetrical peak height is reduced, the $\sigma_+$ dominant

481: peak height is not. Indeed,

482: the $\sigma_+$ component "sees" no atoms after the switching.\\

483: Besides, the system now exhibits tristability for a certain range

484: of the cavity detuning. As mentioned previously the existence of

485: asymmetrical solutions is related to the positivity of $C_{ex}$,

486: whereas the stability of the symmetrical solution is given by

487: $C_{PS}$. For instance, on Fig \ref{fig3}, $C_{ex}=0$ for

488: $\delta_c=\delta_{ex}=5.6$ and the threshold $C_{PS}=0$ is reached

489: for a dephasing $\delta_c=\delta_{PS}=2.6$. Thus, in the range

490: $\delta_{PS}\leq\delta_c\leq \delta_{ex}$, two different sets of

491: asymmetrical solutions exist, in addition to the linear

492: polarization state. This phenomenon is due to the saturation

493: experienced by the $\sigma_+$ and $\sigma_-$ components. As

494: expected, we checked that only the lower branch of each

495: asymmetrical curve is stable, leading to tristability for the

496: polarization state : linear, $\sigma_+$-dominant or

497: $\sigma_-$-dominant. The system switches for a different value of

498: the cavity detuning if the cavity is scanned from left to right,

499: or from right to left (see Fig \ref{fig3}). Hence, unlike the

500: unsaturated case, saturation induces a multistable behavior and a

501: hysteresis cycle now appears in the resonance curve.\\

502: 

503: This brief study of the resonance curve for typical parameters

504: leads to an essential observation: the lower branch of the

505: bistability curve for the linear polarized field is not stable. We

506: may wonder if there is a domain of the parameter space for which

507: it is not the case. Since the quantum fluctuations are expected to

508: be most reduced in the vicinity of the lower turning point

509: \cite{reynaud,hilico}, the answer is of crucial importance for

510: squeezing and will be treated in the next section.

511: 

512: \subsection{Competition between SR and

513: saturation}\label{competition}

514: 

515: \begin{figure}

516:   \centering

517:   \includegraphics[width=10cm]{fig4}

518:   \caption{Bistability curve: linearly polarized field intensity $S=s_x/2$ as a function of

519:   $s_{max}$. PS is the switching threshold, $HT$ and $LT$ the higher and lower turning

520:   points. For certain incident intensities $s_{max}$ three

521:   solutions (1,2,3) exist for the intracavity intensity $s_x/2$, of which

522:   only one (3) is stable, instead of the usual two (1,3) in the

523:   absence of PS phenomenon.

524: %Parameters: $\delta_0=7$, $\Delta=20\gamma$, $\delta_c=0.3$.

525:   }\label{fig4}

526: \end{figure}

527: 

528: To complete the analysis of the steady state, we would like to

529: emphasize that, when the cavity is scanned, PS always happens

530: before reaching the higher turning point of the bistability curve.

531: In order to get some insight into this complicated problem, it is

532: worth looking at Fig \ref{fig4}. We plotted the typical S-shaped

533: variation of the linearly polarized field intensity $s_x$ versus

534: the incoming intensity $s_{max}$, for a fixed value of the cavity

535: detuning $\delta_c$. We choose the parameters so that there is

536: bistability for this state of polarization and report the position

537: of the lower ($LT$) and the higher ($HT$) turning points. In the

538: absence of the PS phenomenon, the solutions between $HT$ and $LT$

539: are unstable (like 2 on Fig \ref{fig4}), whereas solutions on the

540: lower (1) and higher (3) branches are stable. However the

541: stability of the linear polarization is modified by the PS

542: effects. To a fixed value of the dephasing corresponds the PS

543: intensity $S_{PS}$ cancelling $C_{PS}$ in

544: (\ref{criterestabilite}); if $S=s_x/2\leq S_{PS}(\delta_c)$, then

545: the linear polarization is unstable. Hence, if

546: $S_{PS}(\delta_c)\geq S_{HT}$ is satisfied in the whole parameter

547: space, then PS occurs before reaching $HT$, and, consequently, the

548: lower branch is never stable.\\

549: This general feature is shown on Fig \ref{fig5}, in which we

550: represented different bistability curves as in Fig \ref{fig4}. The

551: upper branch of AB is the $HT$ curve (the ensemble of the higher

552: turning points when $\delta_c$ is varied), the lower branch is the

553: $LT$ curve. The dashed curve shows the ensemble of the intensities

554: $S_{PS}$ for which the polarization switches. This curve is always

555: above the $HT$ curve, confirming that the linear polarization

556: always becomes unstable on account of PS first. What is more, we

557: see that PS is closer to $HT$ for low values of $s_x$. We thus

558: expect this situation to be the most favorable to achieve

559: squeezing via optical bistability. We checked that varying the

560: parameters $\delta_0$ and $\Delta$ does not change the conclusion.

561: 

562: \begin{figure}

563:   \centering

564:   \includegraphics[width=11cm]{fig5}

565:   \caption{Bistability curves $S=s_x/2$ as a function of $s_{max}$ for field $A_x$.

566:   The three S-shaped curves correspond to different values of the cavity dephasing

567:   ($\delta_c=-0.25,0.3,1.1$ from left to right),

568:   $\delta_0$ and $\Delta$ having the same value as in Fig \ref{fig3}.

569:   The AB segments represent the $HT$ (higher) and $LT$ (lower) curves.

570:   The dashed curve is the ensemble of the intensities $S_{PS}$ for which

571:   polarization switching occurs. The system exhibits bistability for $s_A\leq s_{max}\leq s_B$.}\label{fig5}

572: \end{figure}

573: 

574: To conclude this section, we would like to point out that

575: bistability, as well as PS, may disappear when the saturation is

576: too high, as can be seen from Fig \ref{fig4}. However, we will

577: focus on the low saturation case in the large detuning limit which

578: is the most favorable case for squeezing, and provides analytical

579: results, as well as a clear physical understanding.

580: 

581: 

582: 

583: \section{Mean field fluctuations}

584: 

585: Since we are interested in the quantum fluctuations, we linearize

586: the quantum operators around their steady state values following

587: the standard linear input-output method \cite{reynaud}. The

588: elliptically polarized solutions are not of great interest for

589: squeezing since the predominant circular component sees no atom

590: and the other has negligible intensity. Therefore, in all the

591: following, we focus on the linearly polarized state and study how

592: both the mean field $A_x$ and the orthogonal vacuum field $A_y$

593: may be squeezed. We have calculated the outgoing fields noise

594: spectra via a full quantum treatment (see e.g. \cite{hilico})

595: involving the four-level system. The outgoing fields are

596: standardly defined from the input-output relation \cite{reynaud}:

597: 

598: \begin{equation}

599: A_{x,\;y}^{out}=\sqrt{T}A_{x,\;y}-A_{x,\;y}^{in}

600: \end{equation}

601: 

602: Yet, to provide clear interpretations as well as analytical

603: results, we derive simplified equations, first for

604: the mean field mode $A_x$, then for the vacuum mode $A_y$.\\

605: A similar equation to (\ref{equationAy2}) can be derived for the

606: field $A_x$ with a term arising from SR in $\delta_{SR}A_y$. In

607: the linearization, this product of zero mean value operators

608: vanishes, so that we only have to take saturation into account to

609: derive the spectra of $A_x$. Field squeezing owing to optical

610: bistability has been widely studied \cite{hilico,lugiato,reid} and

611: is known to occur on a frequency range given by the cavity

612: bandwidth $\kappa$. The most favorable configuration is the bad

613: cavity limit: $\kappa$ is greater than $\gamma$ (in our

614: experiment, $\kappa\simeq 2\gamma$). In the large detuning limit,

615: $\Delta\gg g |A_x|\gg\gamma$, the equation for $A_x$ reads at

616: order 3 in $\gamma/\Delta$,

617: 

618: \begin{equation}

619: \frac{1}{\kappa}\frac{d}{dt}\delta

620: A_x=-(1+i\delta_c-i\delta_0)\delta

621: A_x+i\delta_0\frac{s_x}{2}\left[2\delta

622: A_x+\frac{A_x^2}{|A_x|^2}\delta

623: A_x^{\dagger}\right]+\frac{2}{\sqrt{T}}A_x^{in} \label{Ax1}

624: \end{equation}

625: 

626: where $A_x$ is short for $\langle A_x\rangle$. This simplified

627: equation yields the classical Kerr terms in $A_x^2\delta

628: A_x^{\dagger}$ producing squeezing. Note that absorption,

629: dispersion and the associated atomic noise are not included in

630: (\ref{Ax1}). The spectra taking absorption and dispersion into

631: account can be easily derived and are shown on Fig \ref{fig7}. The

632: associated susceptibility and correlation matrices,

633: $[\chi]_{Kerr}$ and $[\sigma]_{Kerr}$, of the linear input-output

634: theory are reproduced in Appendix, and the comparison with a Kerr

635: medium is discussed in \cite{hilico}. The situation is more

636: complex for the orthogonal mode on account of SR.

637: 

638: 

639: \section{Vacuum fluctuations}

640: 

641: As mentioned in Sec \ref{opo}, SR seems to be a very promising

642: candidate for generating vacuum squeezing. However, a careful

643: analysis of the atomic noise, which cannot be neglected, is

644: necessary in the squeezing calculations. In the optical pumping

645: regime the circular birefringence, $\delta_{SR}=-2\delta_0J_z/N$,

646: is proportional to the ground state population difference

647: $J_z=(\sigma_{22}-\sigma_{11})/2$ (see Sec \ref{opticalpumping}).

648: The SR effect is thus closely related to the fluctuations of

649: $J_z$, and consequently to the fluctuations of $A_y$ via the

650: coupling term in $A_x\delta_{SR}$ [Eq (\ref{equationAy2})].

651: Therefore, we derive general equations for $\delta A_y$ and

652: $\delta J_z$ in the Fourier domain, and examine their low and high

653: frequency limits. For the sake of simplicity, absorption and

654: linear dispersion, again, are not shown; however, the additional

655: terms are included in the Appendix. As previously we place

656: ourselves in the large detuning limit with $s_x\ll 1$ and obtain,

657: discarding terms of order greater than $(\gamma/\Delta)^3$,

658: 

659: \begin{eqnarray}

660: \nonumber -i(\omega/\kappa)\delta A_y & = &

661: -(1+i\delta_c-i\delta_0)\delta

662: A_y-i\delta_0\frac{s_x}{4}\left[3\delta

663: A_y-\frac{A_x^2}{|A_x|^2}\delta

664: A_y^{\dagger}\right]\\

665: & & +\beta(\omega)\frac{2\delta_0}{N}A_x\delta

666: J_z+\frac{2}{\sqrt{T}}\delta A_y^{in}+F_{A_y}\label{Ay1}\\

667: -i\omega\delta J_z & = & -\gamma_p\alpha(\omega)\left[\delta

668: J_z-\lambda(\omega)\left(1-\frac{s_x}{2}\right)\frac{N}{2}\frac{\delta

669: S_z}{|A_x|^2}\right]+F_z\label{Jz1}

670: \end{eqnarray}

671: 

672: where $S_z=i(A_xA_y^{\dagger}-A_x^*A_y)$ is the usual Stokes

673: parameter (see Sec \ref{polarizationsqueezing}) and

674: 

675: \begin{equation}

676: \alpha(\omega)=\left[1-\frac{i\omega}{4\gamma_{\perp}}\right]\frac{\beta(\omega)}{\lambda(\omega)}

677: \;\;,\;\;\;\beta(\omega)=1-\frac{s_x}{4\lambda(\omega)}\;\;,\;\;\;

678: \lambda(\omega)=\frac{2\gamma-i\omega}{2(\gamma-i\omega)}

679: \end{equation}

680: 

681: with $F_{A_y}$ and $F_z$ the Langevin operators associated to

682: $A_y$ and $J_z$ after the adiabatical eliminations,

683: $\gamma_p=\gamma_{\perp}s_x$ the optical pumping rate. In

684: (\ref{Ay1}), the last term of the first line is a crossed Kerr

685: term and clearly contributes to squeezing. We also see that the

686: coupling with $J_z$ is strongly frequency-dependant and requires a

687: careful investigation. In the next sections, we discuss the low

688: and high frequency limits to further simplify the previous

689: equations and give simple interpretations for the squeezing.

690: 

691: \subsection{Low frequency: SR effect}\label{faradaysection}

692: 

693: \begin{figure}

694:   \flushright

695:   \includegraphics[width=14cm]{fig6bis}

696:   \caption{Minimal (left) and maximal (right) outgoing field spectra obtained by considering only SR effects.

697:   Parameters: $\delta_0=5$, $\Delta=40\gamma$, $\gamma/\gamma_{\perp}=3$,

698:   $\kappa=2\gamma$, $s_x=0.1$, $\delta_c=3$.}\label{fig6}

699: \end{figure}

700: 

701: Again, to stress the effects on the fluctuations only due to SR,

702: we neglect the terms in $s_x$ responsible for the Kerr effect,

703: which will be studied in the next section, and place ourselves in

704: the optical pumping regime, keeping only terms of order 1 in

705: $\gamma/\Delta$. Note that this approximation consists in

706: adiabatically eliminating the optical dipoles and neglecting the

707: excited state populations and thus limits the analysis to the

708: range of frequencies $\omega\ll\gamma,\gamma_{\perp}$. Under these

709: conditions, one has $\alpha(\omega)\simeq 1$, $\beta(\omega)\simeq

710: 1$, $\lambda(\omega)\simeq 1$ and Eq (\ref{Jz1}) reduces to the

711: linearized optical pumping equation

712: 

713: \begin{equation}

714: -i\omega\delta J_z = -\gamma_p\left[\delta

715: J_z-\frac{N}{2}\frac{\delta S_z}{|A_x|^2}\right]+F_z \label{Jz2}

716: \end{equation}

717: 

718: It is clear that the fluctuations of $J_z$ are governed by the

719: time constant $\gamma_p$, consistently with the optical pumping

720: approximation $\gamma_p\ll\gamma$. SR is effective only at low

721: frequency. Plugging (\ref{Jz2}) back into (\ref{Ay1}), one gets

722: 

723: \begin{eqnarray}

724: \nonumber (1+i\delta_c-i\omega/\kappa)\delta A_y & = &

725: i\delta_0\left[1-\frac{\gamma_p}{\gamma_p-i\omega}\right]\delta

726: A_y\\& &

727: +i\delta_0\frac{\gamma_p}{\gamma_p-i\omega}\frac{A_x^2}{|A_x|^2}\delta

728: A_y^{\dagger}+\frac{2}{\sqrt{T}}\delta A_y^{in}+\tilde{F}_{A_y}

729: \label{Ay2}

730: \end{eqnarray}

731: 

732: The SR term comes with an amplitude $\delta_0$

733: ($\propto\gamma/\Delta$) around zero frequency, which is much

734: greater than the usual third order saturation non-linearity. Very

735: good squeezing could be expected if it were not for the noise

736: coming from the atoms $\tilde{F}_{A_y}$, which we now study. The

737: fluctuation operator arising from atomic and field fluctuations

738: reads

739: 

740: \begin{equation}

741: \tilde{F}_{A_y}=\frac{2\delta_0A_x}{N}\frac{F_z}{\gamma_p-i\omega}+F_{A_y}

742: \end{equation}

743: 

744: The second term $F_{A_y}$ is responsible for the noise due to

745: absorption. The first term includes the optical pumping noise. One

746: calculates the correlation function of $F_z$ via the quantum

747: regression theorem \cite{cohen}

748: 

749: \begin{equation}

750: \langle

751: F_z(\omega)F_z(\omega')\rangle=2\pi\delta(\omega+\omega')N\gamma_p/2

752: \end{equation}

753: 

754: so that

755: 

756: \begin{eqnarray}

757: \langle

758: \tilde{F}_{A_y}(\omega)\tilde{F}_{A_y}^{\dagger}(\omega')\rangle&\simeq&

759: \frac{4\delta_0^2A_x^2}{N^2}\frac{1}{\gamma_p^2+\omega^2}\langle

760: F_z(\omega)F_z(\omega')\rangle\\

761: &=&2\pi\delta(\omega+\omega')\frac{C}{4}\frac{\gamma

762: }{\gamma_{\perp}}\frac{\gamma_p^2}{\gamma_p^2+\omega^2}

763: \end{eqnarray}

764: 

765: in which we introduced $C=g^2N/T\gamma$ the cooperativity

766: parameter quantifying the strength of the atom-field coupling via

767: the cavity ($C\sim100$ in our Cs experiment). For

768: $\omega\ll\gamma_p$ the noise is thus much more important than the

769: losses due to absorption and therefore has a dramatic influence on

770: the squeezing that could have been produced by the SR term.

771: Following the method given in \cite{hilico}, we derive the

772: susceptibility and correlation matrices which are given in

773: Appendix. We can then calculate the outgoing vacuum field spectrum

774: for all the quadratures. Minimal and maximal spectra are plotted

775: on Fig \ref{fig6} in the "close-to-bad" cavity limit

776: ($\kappa=2\gamma$) corresponding to our experimental

777: configuration. Whereas the first is close to the shot-noise level,

778: the second is extremely noisy. In the good cavity limit, the noise

779: is even more important. The conclusion is that the optical pumping

780: process adds too much noise at zero frequency for SR to generate

781: vacuum squeezing. However, this low frequency noise does not

782: prevent squeezing at higher frequencies.

783: 

784: \subsection{High frequency limit: crossed Kerr effect}

785: If one repeats the previous calculation keeping the first order

786: saturation terms in $s_x$ and considers frequencies

787: $\omega\gg\gamma$, one finds

788: 

789: \begin{equation}

790: \delta J_z(\omega)=\frac{N}{2}\frac{s_x}{4}\frac{\delta

791: S_z(\omega)}{|A_x|^2} \label{Jzkerr}

792: \end{equation}

793: 

794: This is not surprising, since the evolution times considered are

795: small with respect to the atomic relaxation time. The system

796: behaves as if $\sigma_+$ and $\sigma_-$ were independent. In fact,

797: let us consider two independent two-level systems, 1-4 and 2-3,

798: each with $N/2$ atoms. In the large detuning limit, one has

799: $\sigma_{44}=\sigma_{11}s_+/2$ and $\sigma_{33}=\sigma_{22}s_-/2$,

800: and the atomic fluctuations follow the field fluctuations

801: \cite{hilico} (still at order 3 in $(\gamma/\Delta)$)

802: 

803: \begin{equation}

804: \delta \sigma_{11}(\omega)=-\delta

805: \sigma_{44}(\omega)=-\frac{N}{2}\frac{\delta

806: s_+}{2}\;\;,\;\;\;\delta \sigma_{22}(\omega)=-\delta

807: \sigma_{33}(\omega)=-\frac{N}{2}\frac{\delta s_-}{2}

808: \end{equation}

809: 

810: so that, using $|A_x|^2\delta(s_+-s_-)=s_x\delta S_z$, we retrieve

811: (\ref{Jzkerr}). This equation shows that the fluctuations of $J_z$

812: are only caused by saturation and their contribution adds to the

813: crossed Kerr terms already mentioned in (\ref{Ay1}) to retrieve a

814: similar "Kerr" equation for $A_y$ to that of $A_x$ at high

815: frequency

816: 

817: \begin{eqnarray}

818: \nonumber -i(\omega/\kappa)\delta A_y & = &-(1+i\delta_c-i\delta_0)\delta A_y\\

819: & &-i\delta_0\frac{s_x}{2}\left[2\delta

820: A_y-\frac{A_x^2}{|A_x|^2}\delta

821: A^{\dagger}_y\right]+\frac{2}{\sqrt{T}}\delta A_y^{in}+F_{A_y}

822: \label{Aykerr}

823: \end{eqnarray}

824: 

825: This high frequency behavior is thus characterized by the same

826: Kerr-induced optimal squeezing on both polarization modes,

827: consistently with the previous analysis for two independent

828: two-level systems. More precisely, the optimal squeezing spectra

829: are the same for each mode, but involve orthogonal quadratures

830: [because of the sign difference in the Kerr terms between

831: (\ref{Ax1}) and (\ref{Aykerr})]. We now plot the outgoing fields

832: $A_{x,\;y}^{out}$ squeezing spectra and discuss the squeezing

833: optimization.

834: 

835: 

836: \subsection{Squeezing spectra}

837: To derive spectra for the whole frequency range, we combine both

838: effects by adding the matrices obtained in the two asymptotical

839: regimes studied previously. We write the complete susceptibility

840: and correlation matrix under the form

841: $[\chi(\omega)]_y=[\chi(\omega)]_{Kerr}+[\chi(\omega)]_{SR}$ and

842: $[\sigma(\omega)]_y=[\sigma(\omega)]_{Kerr}+[\sigma(\omega)]_{SR}$,

843: where the Kerr matrices are those obtained in the high frequency

844: limit, and the SR matrices those obtained at low frequency (see

845: Appendix for analytical expressions). This approximation is good

846: since Kerr effect is negligible at low frequency compared to SR,

847: while SR breaks down at high frequency.\\

848: In Fig \ref{fig7}, typical spectra for a working point close to

849: the PS threshold are represented. The parameters are chosen to be

850: as close to the experimental situation as possible \cite{josse}.

851: We compared these approximate spectra (a) with those obtained with

852: a full 4-level calculations based on the linear input-output

853: theory (b). The analytical spectra combining Kerr and SR effects

854: show indeed an excellent agreement with the exact calculations, as

855: long as the saturation is low. As shown previously, the SR

856: spectrum is close to the shot-noise level. The Kerr spectrum is

857: accurate for $A_y$ only for $\omega\gg\gamma$ and extends on a

858: range of several $\kappa$ as expected \cite{hilico}. The combined

859: spectrum (a) confirms that SR destroys completely the squeezing at

860: low frequency and reproduces well the exact behavior (b). The best

861: squeezing for $A_y$, obtained at intermediate frequencies, is

862: about 25\%.\\

863: Note that the Kerr spectrum in (a) is also valid for field $A_x$

864: for all frequencies and 45\% of squeezing is obtained at zero

865: frequency. The situation for the mean field $A_x$ is identical to

866: that of a circularly polarized field with intensity $s_x/2$

867: interacting with $N/2$ two-level atoms as in \cite{hilico}, for

868: which the Kerr spectrum shows good agreement with the exact

869: spectrum.

870: 

871: \begin{figure}

872:   \centering

873:   \includegraphics[width=10cm]{fig7}

874:   \caption{(a) Analytical minimal spectra for $A_y^{out}$ given by the SR effect (dashed),

875:    by the Kerr effect (light) and by both effects

876:   (dark).

877:   (b) Exact spectra for the mean field mode $A_x^{out}$ (light) and the

878:   orthogonal vacuum mode $A_y^{out}$ (dark). Parameters values:

879:   $\delta_0=5$, $\Delta=20\gamma$, $\gamma/\gamma_{\perp}=3$,

880:   $\kappa=2\gamma$, $s_{max}=0.1$, $\delta_c=4.6$.}\label{fig7}

881: \end{figure}

882: 

883: As mentioned in Sec \ref{competition}, we expect squeezing to

884: improve in the vicinity of the PS threshold. We verified this

885: behavior by plotting on Fig \ref{fig8} the evolution of the

886: spectra when the cavity is scanned while keeping the incident

887: intensity ($\propto s_{max}$) constant. It appears clearly that

888: the best squeezing is obtained at the peak of the resonance curve,

889: right before the switching. This is due to the fact that, in the

890: low saturation regime, the PS threshold is close to the point

891: where saturation process is the most efficient.

892: 

893: \begin{figure}

894:   \centering

895:   \includegraphics[width=10cm]{fig8}

896:   \caption{Minimal spectra for the vacuum field (plain) and the mean field (dashed) for different working points.

897: $\delta_0=5$, $\Delta=20\gamma$, $\kappa=2\gamma=6\gamma_{\perp}

898: $, $s_{max}=0.1$. The working points coordinates are: A

899: ($\delta_c=\delta_{SR}=4.6$, $s_x/2=0.065$), B ($\delta_c=5.78$,

900: $s_x/2=0.04$), C ($\delta_c=6.79$, $s_x/2=0.02$). The inset shows

901: the working points positions on the resonance curve.}\label{fig8}

902: \end{figure}

903: 

904: We then study the effect of saturation and plot on Fig \ref{fig9}

905: various spectra corresponding to working points close to PS with

906: increasing saturation. The conclusion is that, for given values of

907: the detuning $\Delta$ and linear dephasing $\delta_0$, there is an

908: optimal value of $s_x$ for squeezing. This is due to the fact that

909: the range for which SR adds noise increases with the saturation

910: and eventually destroys Kerr-induced squeezing. The optimal

911: saturation value thus corresponds to a compromise between added

912: noise and Kerr squeezing in the intermediate frequency range.

913: Therefore, a bad cavity is preferable ($\kappa\gg\gamma$), since

914: Kerr-induced squeezing occurs on a frequency range given by

915: $\kappa$ and SR destroys squeezing for frequencies smaller than

916: $\gamma$. Spectra for different cavities are represented in Fig

917: \ref{fig10}. The case $\kappa=2\gamma$ corresponds to the

918: experimental situation, "close-to-bad cavity", the other curves to

919: increasingly bad cavities. Since SR is effective on a range

920: smaller and smaller compared to the cavity bandwidth, its effect

921: becomes negligible, and 75\% of squeezing can be obtained.\\

922: The conclusion is that very interesting squeezing values can be

923: reached in the bad cavity limit for both the mean field mode and

924: the orthogonal field mode. In the next section, we establish the

925: link between polarization squeezing and the vacuum squeezing

926: obtained in our system.

927: 

928: \begin{figure}

929:   \centering

930:   \includegraphics[width=10cm]{fig9}

931:   \caption{Minimal spectra for the vacuum field for different saturation values $s_{max}$.

932:   For each value of $s_{max}$, the working point is chosen close to

933:   PS. Parameters: $\delta_0=5$, $\Delta=20\gamma$,

934:   $\kappa=2\gamma=6\gamma_{\perp}$.}\label{fig9}

935: \end{figure}

936: 

937: 

938: \begin{figure}

939:   \centering

940:   \includegraphics[width=10cm]{fig10}

941:   \caption{Minimal spectra for the vacuum for different $\rho=\kappa/\gamma$ (short dash: $\rho=2$,

942:   long dash: $\rho=10$, plain: $\rho=50$). For each value of $\rho$, the saturation is

943:   optimized. Parameters: $\delta_0=5$, $\Delta=20\gamma$, $\gamma/\gamma_{\perp}=3$.

944: }\label{fig10}

945: \end{figure}

946: 

947: 

948: \section{Polarization squeezing}\label{polarizationsqueezing}

949: 

950: The noise of the mode with orthogonal polarization with respect to

951: the mean field is commonly referred to as polarization noise.

952: However, the study of the polarization state fluctuations requires

953: the introduction of the quantum Stokes operators

954: \cite{chirkin,korolkova,korolkova2}

955: 

956: \begin{eqnarray}

957: S_{0}(t)=A^{\dag}_{x}A_{x}+A^{\dag}_{y}A_{y}&\hspace{0.1cm},\hspace{1cm}&

958: S_{x}(t)=A^{\dag}_{x}A_{x}-A^{\dag}_{y}A_{y}\\

959: S_{y}(t)=A^{\dag}_{x}A_{y}+A^{\dag}_{y}A_{x}&\hspace{0.1cm},\hspace{1cm}&

960: S_{z}(t)=i(A^{\dag}_{y}A_{x}-A^{\dag}_{x}A_{y})

961: \end{eqnarray}

962: 

963: To be consistent with the definition of our slowly-varying

964: envelope operators $A_x$, $A_y$, these Stokes operators are

965: time-dependent and expressed in number of photons per second

966: \cite{fabre}. They obey the following commutation relationships

967: 

968: \begin{eqnarray}

969: \left[S_{0}(t),S_{i}(t')\right]=0\hspace{0.7cm}and\hspace{0.7cm}

970: \left[S_{i}(t),S_{j}(t')\right]= 2i\epsilon_{ijk}S_{k}\delta(t-t')

971: \end{eqnarray}

972: 

973: with $i,j,k=x,y,z$ and then the spectral noise densities of these

974: operators, defined by $\langle

975: S_i(\omega)S_i(\omega')\rangle=2\pi\delta(\omega+\omega')V_{S_i}(\omega)$,

976: satisfy uncertainty relations

977: 

978: \begin{equation}

979: V_{S_{i}}(\omega)V_{S_{j}}(\omega)\geq\epsilon_{ijk}|\langle{S_{k}\rangle}|^{2}\label{Heisenberg}

980: \end{equation}

981: 

982: The coherent polarization state correspond to the case where both

983: modes $A_x$ and $A_y$ are coherent states. Then the noise

984: densities of the Stokes parameters are constant and all equal to $

985: V_{S_{i}}(\omega)= \langle S_0 \rangle = |\langle A_x

986: \rangle|^2+|\langle A_y \rangle|^2$ for $i=0,x,y,z$. The so called

987: polarization squeezing is achieved if one or more of these

988: quantities (except $V_{S_0}$) is reduced below the coherent state

989: value

990: 

991: \begin{equation}

992: \frac { V_{S_i}(\omega)}{S_0}\leq 1 \hspace{1.5cm} (i=x,y,z)

993: \end{equation}

994: 

995: If the mean field is x-polarized, then $\langle S_0

996: \rangle=\langle S_x \rangle= |\langle A_x \rangle|^2 $ and

997: $\langle S_y \rangle=\langle S_z \rangle =0$. At first order in

998: noise fluctuations, $\delta S_y$ and $\delta S_z$ read

999: 

1000: \begin{eqnarray}

1001: \delta S_y & = & |\langle A_x \rangle |\left( \delta A^{\dag}_y

1002: e^{i\theta _x}+\delta A_y e^{-i\theta

1003: _x}\right)\;=|\langle A_x \rangle |\;\delta X_{A_y}(\theta_x) \\

1004: \delta S_z & = & i|\langle A_x \rangle |\left(\delta A^{\dag}_y

1005: e^{i\theta _x}-\delta A_y e^{-i\theta _x}\right)=|\langle A_x

1006: \rangle |\; \delta X_{A_y}(\theta_x+\pi/2)

1007: \end{eqnarray}

1008: 

1009: where $\theta_x$ is the phase of the mean field and

1010: $X_{A_y}(\theta)= A^{\dag}_y e^{i\theta}+A_y e^{-i\theta}$ is the

1011: quadrature with angle $\theta$ of the orthogonal mode. Therefore

1012: the fluctuations of these two Stokes parameters are proportional

1013: to the quadrature noise of $A_y$ and the polarization squeezing of

1014: $S_y$ and $S_z$ is simply related to the vacuum squeezing that we

1015: have studied in the previous sections. The physical meaning of

1016: this result is clear: let us choose $\theta_x = 0$, then geometric

1017: jitter on the polarization is due to the intensity fluctuations of

1018: $A_y$ ($\propto \delta S_y$), whereas the fluctuations of the

1019: ellipticity are caused by the phase fluctuations ($\propto \delta

1020: S_z$). In the general case, the squeezed and antisqueezed Stokes

1021: parameters are found to be

1022: 

1023: \begin{eqnarray}

1024: S_{sq} & = & \cos(\theta_x - \theta_{sq} )S_y + \sin (\theta_x -

1025: \theta_{sq} )S_z\\

1026: S_{antisq} & = & \sin(\theta_x - \theta_{sq} )S_y - \cos (\theta_x

1027: - \theta_{sq} )S_z

1028: \end{eqnarray}

1029: 

1030: Note that, unlike \cite{grangier,polzik,bachor,korolkova3}, there

1031: is no need to lock the phase-shift difference $\theta_x -

1032: \theta_{sq}$, since it is automatically done in this system; this

1033: property appears clearly in Eq (\ref{Aykerr}) where $\langle

1034: A_x^2\rangle/|\langle A_x\rangle|^2=e^{2i\theta_x}$. Since the new

1035: set of the Stokes parameters $S_0$, $S_x$, $S_{sq}$ and

1036: $S_{antisq}$ still satisfy the relationships (\ref{Heisenberg}),

1037: we obtain polarization squeezing in our system as soon as any

1038: quadrature of the vacuum field $A_y$ is squeezed, and the results

1039: of the previous sections can be applied to the squeezed Stokes

1040: component.

1041: 

1042: 

1043: \section{Conclusion}

1044: 

1045: We have presented a study of polarization switching in an X-like

1046: 4-level atoms ensemble illuminated by a linearly polarized light

1047: in an optical cavity. PS has been traced to self-rotation and

1048: simple criteria allow for a clear understanding of the switching

1049: effects and the multistable behavior of the system. The steady

1050: state analysis enables one to figure out the interesting working

1051: points for squeezing.\\

1052: In terms of squeezing the respective contributions of SR and

1053: saturation have been investigated and compared to a full quantum

1054: calculation. Since the propensity for squeezing of SR is cancelled

1055: by atomic noise at low frequency, the squeezing originates from

1056: Kerr effect. The mean field mode is squeezed via the usual

1057: saturation effects, whereas the vacuum mode squeezing is induced

1058: by the mean field via crossed Kerr effect. Both SR and crossed

1059: Kerr effects can be dissociated in a bad cavity configuration,

1060: thus allowing for high squeezing values. Last, this vacuum

1061: squeezing is shown to be equivalent to squeezing one Stokes

1062: operator.

1063: 

1064: \appendix

1065: \section{}

1066: 

1067: Using the input-output theory notations

1068: \cite{reynaud},\cite{hilico}, we give here the expressions of the

1069: susceptibility matrix $[\chi(\omega)]$ and the correlation matrix

1070: $[\sigma(\omega)]$ for field $A_y$. In the high frequency limit,

1071: they resume to those derived in \cite{hilico} in the large

1072: detuning limit

1073: 

1074: \begin{eqnarray}

1075: \nonumber[\chi(\omega)]_{Kerr}=\frac{1}{2\Delta}

1076: \left(\begin{array}{cc}

1077:   1 & 0 \\

1078:   0 & 1

1079: \end{array}\right)

1080:   &+\frac{1}{2\Delta^2}\left(

1081:   \begin{array}{cc}

1082:   i\gamma+\omega & 0 \\

1083:   0 & -i\gamma-\omega\end{array}\right)\\

1084:   & -\frac{g^2}{2\Delta^3}\left(

1085:   \begin{array}{cc}

1086:   2|A_x|^2 & \varepsilon A_x^2 \\

1087:   \varepsilon A_x^{*2} & 2|A_x|^2\end{array}\right)

1088: \end{eqnarray}

1089: 

1090: $\varepsilon=-1$ yields the susceptibility matrix for the vacuum

1091: mode. To retrieve the matrix for $A_x$, $\varepsilon$ should be

1092: taken equal to $+1$. This matrix corresponds to approximating the

1093: atoms ensemble with a Kerr medium: the term of order 1 in

1094: $1/\Delta$ is the linear dephasing, the second order matrix

1095: represents dispersion and absorption and the third order term is

1096: the non-linear dephasing corresponding to the Kerr effect. The

1097: associated correlation matrix is

1098: 

1099: \begin{equation}

1100: [\sigma(\omega)]_{Kerr}=\frac{\gamma}{\Delta^2}\left(\begin{array}{cc}

1101:   1 & 0 \\

1102:   0 & 0\end{array}\right)

1103: \end{equation}

1104: 

1105: In the Kerr limit, the atomic noise comes only from the frequency

1106: independent linear losses of the Kerr medium, which acts as a

1107: beamsplitter for the field. Similar matrices can be derived for

1108: field $A_x$ in agreement with (\ref{Ax1}). At low frequency,

1109: however, the previous matrices have to be completed by

1110: 

1111: 

1112: \begin{eqnarray}

1113: [\chi(\omega)]_{SR}=-\frac{1}{2\Delta}\frac{\gamma_p}{\gamma_p-i\omega}\left(

1114: \begin{array}{cc}

1115:   1 & -A_x^2/|A_x|^2 \\

1116:   -A_x^{*2}/|A_x|^2 & 1\end{array}\right)\label{chiy}

1117: \end{eqnarray}

1118: \begin{eqnarray}

1119: \nonumber[\sigma(\omega)]_{SR} & = &

1120: \frac{\gamma_p^2}{4\gamma_{\perp}(\gamma_p^2+\omega^2)}\left(\begin{array}{cc}

1121:   1 & -A_x^2/|A_x|^2 \\

1122:   -A_x^{*2}/|A_x|^2 &

1123:   1\end{array}\right)\\

1124:   & &

1125:   +\frac{\gamma_p}{2\Delta(\gamma_p^2+\omega^2)}\left(\begin{array}{cc}

1126:   -2\omega & \omega+i\gamma_p\\

1127:   \omega-i\gamma_p & 0\label{sigmay}\end{array}\right)

1128: \end{eqnarray}

1129: 

1130: For $\omega\ll\gamma_p$, the vacuum correlation matrix is

1131: equivalent to

1132: 

1133: \begin{equation}

1134: [\sigma(\omega)]_{SR}\sim\frac{\gamma_p^2}{4\gamma_{\perp}(\gamma_p^2+\omega^2)}\left(\begin{array}{cc}

1135:   1 & -A_x^2/|A_x|^2 \\

1136:   -A_x^{*2}/|A_x|^2 & 1\end{array}\right)

1137: \end{equation}

1138: 

1139: so that a lot of noise is reported on all the quadratures of $A_y$

1140: for frequencies of the order of $\gamma_p$, as pointed out in Sec

1141: \ref{faradaysection}. For frequencies $\omega\gg\gamma$, the SR

1142: noise terms vanish, allowing for crossed Kerr effect to produce

1143: squeezing.

1144: 

1145: \bigskip

1146: \begin{thebibliography}{10}

1147: 

1148: \bibitem{davidovich} L. Davidovich, Rev. Mod. Phys. \textbf{68},

1149: 127 (1996), and references therein.

1150: 

1151: \bibitem{grangier} P. Grangier, R.E. Slusher, B. Yurke, A.

1152: LaPorta, Phys. Rev. Lett. \textbf{59}, 2153 (1987).

1153: 

1154: \bibitem{polzik} J. Hald, J.L. Sorensen, C. Shori, E.S. Polzik,

1155: \textbf{83}, 1319 (1999).

1156: 

1157: \bibitem{bachor} W.P. Bowen, R. Schnabel, H.A. Bachor, P.K. Lam,

1158: Phys. Rev. Lett. \textbf{88}, 093601 (2002); W.P. Bowen, N. Treps,

1159: R. Schnabel, P.K. Lam, Phys. Rev. Lett. \textbf{89}, 253601

1160: (2002).

1161: 

1162: \bibitem{chirkin2} A.P. Alodjants, A.M. Arakelian, A.S. Chirkin, JEPT

1163: \textbf{108}, 63 (1995); N.V. Korolkova, A.S. Chirkin, J. Mod.

1164: Opt. \textbf{43}, 869 (1996).

1165: 

1166: \bibitem{boivin} L. Boivin, H.A. Haus, Optics. Lett. \textbf{21},

1167: 146 (1996).

1168: 

1169: \bibitem{korolkova3} J. Heersink, T. Gaber, S. Lorenz, O.

1170: Gl$\ddot{o}$ckl, N. Korolkova, G. Leuchs, e-print:

1171: quant-ph/0302100.

1172: 

1173: \bibitem{matsko} A.B. Matsko, I. Novikova, G.R. Welsch, D. Ducker,

1174: D.F. Kimball, S.M. Rochester, Phys. Rev. A \textbf{66}, 043815/1

1175: (2002).

1176: 

1177: \bibitem{lambrecht} A. Lambrecht, E. Giacobino, J.M. Courty,

1178: Optics Comm. \textbf{115},199 (1995).

1179: 

1180: \bibitem{coudreau} A. Lambrecht, T. Coudreau, A.M. Steimberg, E.

1181: Giacobino, Europhys. Lett. \textbf{36}, 93 (1996).

1182: 

1183: \bibitem{josse} V. Josse, A. Dantan, L. Vernac, A. Bramati, M.

1184: Pinard, E. Giacobino, to be published.

1185: 

1186: \bibitem{chirkin} A.S. Chirkin, A.A. Orlov, D.Yu Paraschuk, Kvant. Elektron.

1187: \textbf{20}, 999 (1993) [Quantum Electron. \textbf{23}, 870

1188: (1993).

1189: 

1190: \bibitem{korolkova} N. Korolkova, G. Leuchs, R. Loudon, T.C.

1191: Ralph, C. Silberhorn, Phys. Rev. A \textbf{65}, 052306 (2002).

1192: 

1193: \bibitem{fabre} C. Fabre, \textit{Quantum fluctuations in light

1194: beams}, Les Houches session LXIII (1996).

1195: 

1196: \bibitem{chu} N. Davidson, H.J. Lee, C.S. Adams, M. Kasevich, S.

1197: Chu, Phys. Rev. Lett. \textbf{74}, 1311 (1995).

1198: 

1199: \bibitem{cohen} C. Cohen-Tannoudji, J. Dupont-Roc, G. Grynberg,

1200: \textit{Processus d'interaction entre photons et atomes} (1996).

1201: 

1202: \bibitem{walls} D.F. Walls, P. Zoller, Optics Comm. \textbf{34}, 260 (1980);

1203: M. Kitano, T. Yabuzaki, T. Ogawa, Phys. Rev. Lett. \textbf{46},

1204: 926 (1981); C.M. Savage, H.J. Carmichael, D.F. Walls, Optics Comm.

1205: \textbf{42}, 211 (1982).

1206: 

1207: \bibitem{elisabeth} E. Giacobino, Opt. Comm. \textbf{56}, 249

1208: (1985).

1209: 

1210: \bibitem{huard} S. Huard, \textit{Polarization of light}, John

1211: Wiley and Sons, New York (1997), p. 26.

1212: 

1213: \bibitem{savage} P.D. Maker, R.W. Terhune, C.M. Savage, Phys. Rev. Lett.

1214: \textbf{12}, 507 (1964).

1215: 

1216: \bibitem{yashchuk} S.M. Rochester, D.S. Hsiung, D. Bucker, R.Y.

1217: Chiao, D.F Kimball, V.V. Yashchuk, Phys. Rev. A \textbf{63},

1218: 043814 (2001).

1219: 

1220: \bibitem{gardiner} G. Milburn, D.F. Walls, Optics. Comm.

1221: \textbf{39}, 401 (1981); M.J. Collet, C.W. Gardiner, Phys. Rev. A

1222: \textbf{30}, 1386 (1984); M.J. Collet, D.F. Walls, Phys. Rev. A

1223: \textbf{32}, 2887 (1985).

1224: 

1225: 

1226: \bibitem{gibbs} H.M. Gibbs, \textit{Optical bistability:

1227: controlling light with light}, Academic Press, Orlando (1985).

1228: 

1229: \bibitem{reynaud} S. Reynaud, C. Fabre, E. Giacobino, A. Heidmann,

1230: Phys. Rev. A \textbf{40}, 1440 (1989).

1231: 

1232: \bibitem{hilico} L. Hilico, C. Fabre, S. Reynaud, E. Giacobino,

1233: Phys. Rev. A \textbf{46}, 4397 (1992).

1234: 

1235: \bibitem{lugiato} L.A. Lugiato, G. Strini, Optics Comm.

1236: \textbf{41}, 67 (1982).

1237: 

1238: \bibitem{reid} M.D. Reid, Phys. Rev. A \textbf{37}, 4792 (1988).

1239: 

1240: \bibitem{korolkova2} N. Korolkova, Ch. Silberhorn, O.

1241: Gl\"{o}ckl, S. Lorentz, Ch. Marquardt, G. Leuchs, Eur. Phys. J. D.

1242: \textbf{18}, 229 (2002).

1243: 

1244: \end{thebibliography}

1245: 

1246: 

1247: \end{document}

1248: