1: \documentclass[prb,aps,twocolumn,groupedaddress,showpacs]{revtex4}

2: \usepackage{graphicx}

3: \bibliographystyle{PRBSTY}

4: 

5: \begin{document}

6: 

7: 

8: \title{ Radiative corrections to the excitonic molecule state in 

9: GaAs microcavities }

10: 

11: \author{A. L. Ivanov}

12: 

13: \address{Department of Physics and Astronomy, Cardiff University, 

14: Queen's Buildings, Cardiff CF24 3YB, United Kingdom}

15: 

16: \author{P. Borri, W. Langbein, and U. Woggon}

17: 

18: \address{Lehrstuhl f\"{u}r Experimentelle Physik

19: EIIb, Universit\"{a}t Dortmund, Otto-Hahn Str.4, 44227 Dortmund, Germany}

20: 

21: 

22: 

23: \date{\today}

24: 

25: 

26: \begin{abstract}

27: 

28: The optical properties of excitonic molecules (XXs) in GaAs-based 

29: quantum well microcavities (MCs) are studied, both theoretically and 

30: experimentally. We show that the radiative corrections to the XX state, 

31: the Lamb shift $\Delta^{\rm MC}_{\rm XX}$ and radiative width 

32: $\Gamma^{\rm MC}_{\rm XX}$, are large, about $10-30\,\%$ of the molecule 

33: binding energy $\epsilon_{\rm XX}$, and definitely cannot be neglected. 

34: The optics of excitonic molecules is dominated by the in-plane resonant 

35: dissociation of the molecules into outgoing 1$\lambda$-mode and 

36: 0$\lambda$-mode cavity polaritons. The later decay channel, ``excitonic 

37: molecule $\rightarrow$ 0$\lambda$-mode polariton +  0$\lambda$-mode 

38: polariton'',  deals with the short-wavelength MC polaritons invisible in 

39: standard optical experiments, i.e., refers to ``hidden'' optics of 

40: microcavities. By using transient four-wave mixing and pump-probe 

41: spectroscopies, we infer that the radiative width, associated with 

42: excitonic molecules of the binding energy $\epsilon_{\rm XX} \simeq 

43: 0.9-1.1$\,meV, is $\Gamma^{\rm MC}_{\rm XX} \simeq 0.2-0.3$\,meV in the 

44: microcavities and $\Gamma^{\rm QW}_{\rm XX} \simeq 0.1$\,meV in a 

45: reference GaAs single quantum well (QW). We show that for our 

46: high-quality quasi-two-dimensional nanostructures the $T_2 = 2 T_1$ 

47: limit, relevant to the XX states, holds at temperatures below 10\,K, 

48: and that the bipolariton model of excitonic molecules explains 

49: quantitatively and self-consistently the measured XX radiative widths. 

50: A nearly factor two difference between $\Gamma^{\rm MC}_{\rm XX}$ and 

51: $\Gamma^{\rm QW}_{\rm XX}$ is attributed to a larger number of the XX 

52: optical decay channels in microcavities in comparison with those in 

53: single QWs. We also find and characterize two critical points in the 

54: dependence of the radiative corrections against the microcavity detuning, 

55: and propose to use the critical points for high-precision measurements of 

56: the molecule binding energy and microcavity Rabi splitting. 

57: 

58: \pacs{78.66.-w,78.47.+p,78.66.Fd,71.36.+c}

59: \end{abstract}

60: \maketitle

61: 

62: 

63: \section{Introduction}

64: 

65: 

66: The optical properties of an excitonic molecule originate from the 

67: resonant interaction of its constituent excitons (Xs) with the light 

68: field. For semiconductor (GaAs) nanostructures we analyze in this paper, 

69: the above interaction refers to quasi-two-dimensional (quasi-2D) QW 

70: excitons and is different in single, MC-free quantum wells and in 

71: microcavities. In the first case, the breaking of translational 

72: invariance along the growth direction ($z$-direction) leads to the 

73: coupling of QW excitons to a continuum of bulk photon modes. This 

74: results in an irreversible radiative decay of low-energy QW excitons 

75: into the bulk photon modes and to interface, or QW, polaritons for 

76: the QW exciton states lying outside the photon cone 

77: \cite{Agranovich66,Nakayama85,Andreani90}. An interface polariton is 

78: the in-plane propagating eigenwave guided by a single QW, and the light 

79: field associated with interface polaritons is evanescent, i.e., it 

80: decays exponentially in the $z$-direction. In contrast, the MC polariton 

81: optics deals with the quasi-stationary mixed states of quasi-2D MC 

82: photons and QW excitons \cite{Weisbuch92,Savona94}, i.e., one realizes a 

83: nearly pure 2D exciton-photon system with resonant coupling between two 

84: eigenmodes (for a review of the MC polariton optics see, e.g., 

85: Refs.\,[\onlinecite{SkolnickSST98,KhitrovaRMP99}]). In this case the 

86: radiative lifetime of MC polaritons originates from a finite transmission 

87: through the cavity mirrors. The main aim of the present work is to 

88: develop coherent optics of quasi-2D excitonic molecules in semiconductor 

89: microcavities. 

90: 

91: The XX-mediated optical response from GaAs microcavities has been 

92: addressed only recently 

93: \cite{Gonokami97,Fan98,BorriPRBRb00,Jacob00,Saba00,Baars01,Tartakovskii02}. 

94: The Coulombic attractive interaction of cross-circular polarized 

95: ($\sigma^+$ and $\sigma^-$) excitons, which gives rise to the XX bound 

96: state, has been invoked and estimated in order to analyze the 

97: frequency-degenerate four-wave mixing (FWM) experiment \cite{Gonokami97}. 

98: Pump-probe specroscopy was used in Ref.\,[\onlinecite{Fan98}] to observe 

99: the XX-mediated pump-induced changes in the MC reflectivity spectrum. 

100: However, in the above first experiments the microcavity polariton 

101: resonance has large broadening so that the spectrally--resolved XX 

102: transition was not detected. Only recently the spectrally--resolved 

103: ``polariton $\leftrightarrow$ XX'' photon-assisted transition in 

104: GaAs-based MCs has been observed by using differential reflection 

105: spectroscopy \cite{BorriPRBRb00,Jacob00}. In particular, the transition 

106: is revealed in a pump--probe experiment as an induced absorption from 

107: the lower polariton dispersion branch to the XX state, at positive 

108: pump-probe time delays \cite{BorriPRBRb00}. In the latter work the MC 

109: Rabi splitting $\Omega_{\rm 1\lambda}^{\rm MC}$, associated with a 

110: heavy--hole QW exciton, exceeds the XX binding energy 

111: $\epsilon_{\rm XX}$ for more than a factor of three. The last experiments 

112: on excitonic molecules in GaAs microcavities use a high-intensity laser 

113: field to investigate the XX-mediated changes in the polariton spectrum 

114: \cite{Saba00,Baars01} and parametric scattering of MC polaritons 

115: \cite{Tartakovskii02}. In this work we are dealing with a 

116: {\it low-intensity} limit of the XX optics, aiming to study the radiative 

117: corrections to the molecule state in a high-quality GaAs single QW 

118: embedded in a co-planar $\lambda$-cavity. Recently, the optical properties 

119: of large binding energy excitonic molecules in a MC-embedded ZnSe QW has 

120: been studied \cite{Neukirch00}. The theoretical model we work out can 

121: straightforwardly be adapted to the quasi-2D molecules in II-VI 

122: nanostructures. 

123: 

124: In the previous theoretical studies \cite{LaRoccaJOSAB98,SiehEPJ99} the 

125: XX radiative corrections are not included, so that the models deal 

126: with the optically unperturbed molecule wavefunction $\Psi_{\rm XX}$ and 

127: binding energy $\epsilon_{\rm XX}$. According to 

128: Ref.\,[\onlinecite{LaRoccaJOSAB98}], the XX radiative corrections are 

129: rather small, even if $\Omega_{\rm 1\lambda}^{\rm MC} \gg 

130: \epsilon_{\rm XX}$. The authors argue qualitatively that a volume of 

131: phase space, where the resonant coupling of the constituent excitons with 

132: the light field occurs, is rather small to affect the XX state. As we 

133: show below, an exactly-solvable bipolariton model \cite{Ivanov95,Ivanov98}, 

134: adapted to excitonic molecules in GaAs-based quasi-2D nanostructures, 

135: yields $\Gamma^{\rm MC}_{\rm XX}$ and $\Delta^{\rm MC}_{\rm XX}$ of about 

136: $(0.15-0.30)\,\epsilon_{\rm XX}$ for microcavities, and 

137: $\Gamma^{\rm QW}_{\rm XX}$ and $\Delta^{\rm QW}_{\rm XX}$ of about 

138: $(0.10-0.15)\,\epsilon_{\rm XX}$ for single QWs. The calculated values 

139: refer to the weak confinement of QW excitons and QW excitonic molecules 

140: we deal with in our study. In the weak confinement limit, the 

141: QW thickness $d_z$ is comparable with the in-plane radius of the above 

142: electron-hole bound complexes, which are still constructed in terms of 

143: well-defined transversly-quantized quasi-2D electronic states. In contrast, 

144: in the strong confinement limit, $d_z$ is much less than the in-plane 

145: radius of QW excitons (excitonic molecules). 

146: 

147: The radiative corrections to the XX state cannot be neglected, because 

148: the exciton-photon coupling (polariton effect) changes the dispersion of 

149: excitons not only in a very close vicinity of the resonant crossover 

150: between the relevant exciton and photon energies, but in a rather broad 

151: band $p \sim p_0$. Here wavevector $p_0$ is given by the resonant 

152: condition $\hbar \omega^{\gamma}(p_0) = cp / \sqrt{\varepsilon_b} = 

153: E_{\rm X}(p_0)$ between the bulk photon and exciton dispersions 

154: ($\varepsilon_b$ is the background dielectric constant). For GaAs 

155: structures $p_0 \simeq 2.7 \times 10^5\,{\rm cm}^{-1}$. The dimensionless 

156: parameter $\delta^{\rm(D)}_{\rm R}$, which scales the XX radiative 

157: corrections, is $\delta^{\rm(D)}_{\rm R} = (a^{\rm (D)}_{\rm XX} 

158: p_0)^{\rm D}$, where $a^{\rm (D)}_{\rm XX}$ is the molecule radius 

159: and $D$ is the dimensionality of a semiconductor structure. Remarkably, 

160: as we demonstrate below, $\delta^{\rm(2D)}_{\rm R}$ does not depend upon 

161: the MC detuning between the $\lambda$-cavity mode and $E_{\rm X}$, i.e., 

162: is the same for microcavities and single QWs. For our high-quality GaAs 

163: QWs with weak confinement of excitons one estimates $a^{\rm (2D)}_{\rm XX} 

164: \simeq 200\,\AA$, so that $\delta^{\rm(2D)}_{\rm R} \simeq 0.3$. The 

165: latter value clearly shows that the exciton-photon coupling does change 

166: considerably the quasi-2D XX states. Even for the X wavevectors far away 

167: from the resonant crossover point $p_0$, the polariton effect can still 

168: have a considerable impact on the dispersion of optically-dressed 

169: excitons in bulk semiconductors and QWs. To illustrate this, note that 

170: for bulk GaAs, e.g., the effective mass associated with the upper 

171: polariton dispersion branch at $p=0$ is given by $M_{\rm eff} \simeq 

172: M_x/4$, i.e., by factor four is less than the translational mass $M_x$ 

173: of optically undressed excitons. In a similar way, the dispersion of QW 

174: excitons dressed by MC photons, which gives rise to 

175: $\Delta_{\rm XX}^{\rm MC}$ and $\Gamma^{\rm MC}_{\rm XX}$, refers to the 

176: in-plane wavevector domain $p_{\|} \lesssim p_0$ rather than to a close 

177: vicinity of the crossover point $p_{\|} \simeq 0$. 

178: 

179: An excitonic molecule can be described in terms of two quasi-bound 

180: polaritons (bipolariton), if the coupling of the molecule with the 

181: light field is much stronger than the incoherent scattering processes. 

182: In this case the sequence ``two incoming polaritons (or bulk photons) 

183: $\rightarrow$ quasi-bound XX state $\rightarrow$ two outgoing polaritons'' 

184: is a completely coherent process of the resonant polariton-polariton 

185: scattering and can be described in terms of the bipolariton wavefunction 

186: ${\tilde \Psi}_{\rm XX}$. The latter includes an inherent contribution 

187: from the outgoing (incoming) polaritons and should be found from the 

188: bipolariton wave equation. The solution also yields the radiative 

189: corrections to the XX energy, i.e., $-\epsilon_{\rm XX} = - 

190: \epsilon_{\rm XX}^{(0)} + \Delta_{\rm XX} - i \Gamma_{\rm XX}/2$, where 

191: $\epsilon_{\rm XX}^{(0)}$ is the ``input'' XX binding energy of an 

192: optically inactive molecule. For some particular model potentials of 

193: $\sigma^+$-exciton -- $\sigma^-$-exciton interaction, e.g., for the 

194: deuteron and Gaussian potentials, the bipolariton wave equation can be 

195: solved exactly \cite{Ivanov95,Ivanov98}. The bipolariton concept was 

196: verified in high-precision experiments with low-temperature bulk CuCl 

197: \cite{Chemla79,Akiyama90,Tokunaga99} and CdS \cite{Mann01}, and was also 

198: applied successfully to explain the XX-mediated optical response from 

199: GaAs/AlGaAs multiple QWs \cite{Ivanov97}. The latter experiment dealt 

200: with quasi-2D XXs in the limit of strong QW confinement. In this case 

201: the bipolariton model shows that the main channel of the optical decay 

202: of QW excitonic molecules in MC-free structures is the resonant 

203: photon-assisted dissociation of the molecule into two outgoing interface 

204: (QW) polaritons. Note that the Coulombic interaction between two 

205: constituent excitons of the molecule couples the radiative modes and the 

206: interface  polariton states, so that an ``umklapp'' process between the 

207: modes can intrinsically be realized. The above picture refers to the 

208: following scenario of the coherent optical generation and dissociation of 

209: QW molecules: ``$\sigma^+$ bulk photon + $\sigma^-$ bulk photon 

210: $\rightarrow$ $\sigma^+$ virtual QW exciton + $\sigma^-$ virtual QW 

211: exciton $\rightarrow$ QW molecule $\rightarrow$ $\sigma^+$ interface 

212: polariton + $\sigma^-$ interface polariton''. 

213: 

214: The experiments we report on deal with weakly confined QW excitonic 

215: molecules, i.e., the QW thickness $d_z = 250\,\AA$ is comparable with 

216: the radius of excitons in bulk GaAs. The quasi-2D weak confinement 

217: allows us to neglect inhomogeneous broadening in the detected X- and 

218: XX-mediated signals. The MC-free single QW is used as a reference 

219: structure: All the $\lambda$-microcavities, which we study, are embedded 

220: with a single QW nearly identical to the reference one. By analyzing the 

221: coherent dynamics of the XX-mediated signal in spectrally-resolved 

222: transient FWM, we infer the XX radiative width in the microcavities 

223: and in the reference single QW, $\Gamma^{\rm MC}_{\rm XX}$ and 

224: $\Gamma^{\rm QW}_{\rm XX}$, respectively. The measurements yield 

225: $\Gamma^{\rm MC}_{\rm XX}$ larger than $\Gamma^{\rm QW}_{\rm XX}$ by 

226: nearly factor two. Furthermore, by using pump-probe spectroscopy we 

227: also estimate the XX binding energies $\epsilon_{\rm XX}^{\rm MC}$ and 

228: $\epsilon_{\rm XX}^{\rm QW}$. Our measurements deal with the MC detuning 

229: band $-2\,\mbox{meV} \lesssim \delta \lesssim +2\,\mbox{meV}$. 

230: 

231: Similarly to quasi-2D XXs in high-quality single QWs, the main mechanism 

232: of the optical decay of MC molecules is their in-plane resonant 

233: dissociation into MC polaritons. Thus the coherent optical path of the 

234: XX-mediated signal in our experiments is given by ``$\sigma^+$ (pump) 

235: bulk photon + $\sigma^-$ (pump) bulk photon $\rightarrow$ MC molecule 

236: $\rightarrow$ $\sigma^+$ MC polariton + $\sigma^-$ MC polariton 

237: $\rightarrow$ $\sigma^+$ (signal) bulk photon + $\sigma^-$ (signal) bulk 

238: photon''. The latter escape of the MC polaritons into the bulk photon 

239: modes is due to a finite radiative lifetime of MC photons. In order to 

240: explain the experimental data, the bipolariton model is adapted to weakly 

241: confined quasi-2D molecules in (GaAs) microcavities and MC-free (GaAs) 

242: single QWs. One of the most important features of the optics of excitonic 

243: molecules in microcavities is a large contribution to the bipolariton 

244: state ${\tilde \Psi}_{\rm XX}$ from $0\lambda$-mode MC polaritons. The 

245: relevant $0\lambda$-mode polariton states refer to the in-plane 

246: wavevectors $p_{\|} \sim p_0$, i.e., are short-wavelength in comparison 

247: with the $1\lambda$-mode polariton states activated in standard optical 

248: experiments. An ``invisible'' decay channel of the MC molecule into two 

249: outgoing $0\lambda$-mode polaritons in combination with the directly 

250: observable dissociation path ``XX $\rightarrow$ $1\lambda$-mode MC 

251: polariton + $1\lambda$-mode MC polariton'' explain qualitatively the 

252: factor two difference between $\Gamma^{\rm MC}_{\rm XX}$ and 

253: $\Gamma^{\rm QW}_{\rm XX}$. The use of the microcavities embedded with a 

254: single QW allows us to apply the bipolariton model without complications 

255: due to the dark X states in multiple QWs \cite{LaRoccaJOSAB98}. The 

256: bipolariton model quantitatively reproduces our experimental data and 

257: predicts new spectral features, like $M_{1,2}$ critical points in the 

258: detuning dependent $\Gamma^{\rm MC}_{\rm XX} = 

259: \Gamma^{\rm MC}_{\rm XX}(\delta)$ and $\Delta^{\rm MC}_{\rm XX} = 

260: \Delta^{\rm MC}_{\rm XX}(\delta)$. 

261: 

262: Thus the main results of our study on weakly confined quasi-2D 

263: molecules in GaAs microcavities are (i) rigorous justification of the 

264: bipolariton model, (ii) importance of the XX radiative corrections, 

265: and (iii) existence of the efficient ``hidden'' XX decay channel, 

266: associated with 0$\lambda$-mode MC polaritons. 

267: 

268: In Sec.\,II, we apply the bipolariton model in order to analyze the XX 

269: radiative corrections, the XX Lamb shift $\Delta_{\rm XX}$ and XX 

270: radiative width $\Gamma_{\rm XX}$, relevant to our microcavities and 

271: reference QW. After a brief discussion of interface and MC polaritons, 

272: we demonstrated that in GaAs-based quasi-2D structures the XX radiative 

273: corrections can be as large as $10-30\,\%$ of the (input) XX binding 

274: energy $\epsilon^{(0)}_{\rm XX}$. It is shown that independently of the 

275: MC detuning $\delta$ the XX radiative corrections in microcavities and 

276: (reference) QWs are scaled by the same dimensionless parameter 

277: $\delta_{\rm R}^{\rm (2D)} = ( a^{\rm (2D)}_{\rm XX} p_0 )^2$, and that 

278: the main XX optical decay channels in microcavities are ``XX 

279: $\rightarrow$ 1$\lambda$-mode MC polariton + 1$\lambda$-mode MC 

280: polariton'' and ``XX $\rightarrow$ 0$\lambda$-mode MC polariton + 

281: 0$\lambda$-mode MC polariton'' against the main decay path in single 

282: QWs, ``XX $\rightarrow$ interface polariton + interface polariton''. 

283: We also find and classify two {\it critical points}, $M_1$ and $M_2$, in 

284: the spectrum of the XX radiative corrections in microcavities, 

285: $\Gamma^{\rm MC}_{\rm XX} = \Gamma^{\rm MC}_{\rm XX}(\delta)$ and/or 

286: $\Delta^{\rm MC}_{\rm XX} = \Delta^{\rm MC}_{\rm XX}(\delta)$, and 

287: propose to use the critical points for high-precision measurements of 

288: the MC Rabi splitting and the XX binding energy. 

289: 

290: In Sec.\,III, the investigated GaAs-based MC sample and the reference 

291: GaAs single QW are characterized. We describe the FWM measurements at 

292: $T=9$\,K, which allow us to estimate the XX dephasing width for the MC 

293: detuning band $-2\,\mbox{meV} \lesssim \delta \lesssim 2\,\mbox{meV}$, 

294: ${\tilde \Gamma}^{\rm MC}_{\rm XX}(T\!=\!9\,\mbox{K}) \simeq 

295: 0.3-0.4$\,meV, and the pump-probe experiments at $T=5$\,K, which yield 

296: the bipolariton (XX) binding energy in our microcavities, 

297: $\epsilon^{\rm MC}_{\rm XX} \simeq 0.9-1.1$\,meV. 

298: 

299: In Sec.\,IV, by analyzing a temperature-dependent contribution to the 

300: dephasing widths ${\tilde \Gamma}^{\rm MC}_{\rm XX}$ and 

301: ${\tilde \Gamma}^{\rm QW}_{\rm XX}$, which is associated with XX -- 

302: LA-phonon scattering, we estimate the corresponding XX radiative widths 

303: in the microcavities and reference QW ($\Gamma^{\rm MC}_{\rm XX} \simeq 

304: 0.2-0.3$\,meV and $\Gamma^{\rm QW}_{\rm XX} \simeq 0.1$\,meV), and show 

305: that the bipolariton model does reproduce {\it quantitatively and 

306: self-consistently} both $\Gamma^{\rm MC}_{\rm XX}$ and 

307: $\Gamma^{\rm QW}_{\rm XX}$. It is shown that the $T_2 = 2 T_1$ limit, 

308: which is crucial for the validity of the bipolariton model, starts to 

309: hold for excitonic molecules at cryostat temperatures below 10\,K. We 

310: also discuss the underlaying physical picture responsible for the large 

311: XX radiative corrections in high-quality quasi-2D (GaAs) nanostructures. 

312: 

313: A short summary of the results is given in Sec.\,V. 

314: 

315: 

316: \section{ The bipolariton states in microcavities and single quantum 

317: wells }

318: 

319: 

320: In this Section we briefly discuss interface (quantum well) and 

321: microcavity polaritons, and apply the bipolariton model 

322: \cite{Ivanov95,Ivanov98} in order to calculate the XX radiative 

323: corrections and to describe the optical decay channels of excitonic 

324: molecules in high-quality GaAs-based microcavities and single QWs. 

325: 

326: \subsection{Interface and microcavity polaritons}

327: 

328: 

329: For a single QW, the resonant coupling of excitons with the light field 

330: can be interpreted in terms of the radiative in-plane modes 

331: $|{\bf p}_{\|}| \leq p_0$, which ensure communication of low-energy QW 

332: excitons with incoming and outgoing bulk photons (the only photons used 

333: in standard pump-probe optical experiments with QWs), and interface 

334: polaritons, which refer to the states outside the photon cone, 

335: $|{\bf p}_{\|}| \geq \omega_t \sqrt{\varepsilon_b}/c$. The latter 

336: in-plane propagating polariton eigenmodes are trapped and waveguided by 

337: the X resonance; they are accompanied by the evanescent, interface light 

338: field, i.e., are invisible at macroscopic distances from the QW. 

339: 

340: For an ideal QW microcavity the MC photons with in-plane wavevector 

341: ${\bf p}_{\|}$ can be classified in terms of $n \lambda$- transverse 

342: eigenmodes ($n=0,1,2,...$). The MC polariton eigenstates arise when some 

343: of the MC photon eigenmodes resonate with the QW exciton state. As we 

344: show below, only $0 \lambda$- and $1 \lambda$- polariton eigenmodes are 

345: relevant to the optics of QW excitonic molecules in our MC structures. 

346: With increasing MC thickness towards infinity the microcavity polariton 

347: eigenstates evolve into the radiative and interface polariton eigenmodes 

348: associated with a MC-free single QW \cite{Savona94}. 

349: 

350: (i) {\it The light field resonantly interacting with quasi-2D 

351: excitons in a single (GaAs) QW.}

352: %

353: The interaction of a QW exciton with in-plane momentum $\hbar {\bf p}_{\|}$ 

354: with the transverse light field of frequency $\omega$ is characterized by 

355: the dispersion equation \cite{Agranovich66,Nakayama85,Andreani90}: 

356: \begin{equation}

357: { c^2 p_{\|}^2 \over \varepsilon_b } = 

358: \omega^2 + { \omega^2 R^{\rm QW}_{\rm X} \sqrt{p_{\|}^2 - \varepsilon_b 

359: (\omega/c)^2} \over \omega^2_t + \hbar \omega_t p_{\|}^2/M_x 

360: - i \omega \gamma_{\rm X} - \omega^2 \ } \ ,  

361: \label{pol}

362: \end{equation}

363: where $M_x$ is the in-plane translational X mass, $\hbar \omega_t = 

364: E_{\rm X}({\bf p}_{\|}$=0) is the X energy, $\gamma_{\rm X}$ is the 

365: rate of incoherent scattering of QW excitons, and $R^{\rm QW}_{\rm X}$ 

366: is the dimensional oscillator strength of exciton-photon interaction 

367: per QW unit area. Equation (\ref{pol}) refers to a single (GaAs) 

368: QW confined by two identical (AlGaAs) bulk barriers. 

369: 

370: For $|{\bf p}_{\|}| \geq \omega \sqrt{\varepsilon_b}/c$, i.e., for the 

371: momentum-frequency domain outside the photon cone, Eq.~(\ref{pol}) 

372: describes the in-plane polarized transverse interface polaritons ($Y$-mode 

373: polaritons). The evanescent light field associated with the interface 

374: polaritons is given by ${\bf E}(\omega,{\bf p}_{\|},z) = 

375: {\bf E}(\omega,{\bf p}_{\|}) \exp(-\kappa |z|)$, where 

376: $\kappa = \sqrt{p_{\|}^2 - \varepsilon_b (\omega/c)^2}$. The exciton 

377: and photon components of a QW polariton with in-plane wavevector 

378: ${\bf p}_{\|}$ are 

379: \begin{eqnarray}

380: && u^2_{\rm IP}(p_{\|}) = { \kappa R^{\rm QW}_{\rm X} \over \kappa 

381: R^{\rm QW}_{\rm X} + 2 [ \omega_t  + \hbar  p_{\|}^2/2M_x -  

382: \omega_{\rm IP}(p_{\|}) ]^2 } \ , 

383: \nonumber \\

384: && v^2_{\rm IP}(p_{\|}) = 1 - u^2_{\rm IP}(p_{\|}) \ , 

385: \label{comp}

386: \end{eqnarray}

387: respectively. Here $\omega = \omega_{\rm IP}(p_{\|})$ is the polariton 

388: dispersion determined by Eq.\,(\ref{pol}). Note that the $z$-polarized 

389: transverse interface polaritons ($Z$-mode QW polaritons) associated with 

390: the ground-state heavy-hole excitons are not allowed in GaAs QWs 

391: \cite{Andreani90}. 

392: 

393: The low-energy QW excitons from the radiative zone $|{\bf p}_{\|}| \leq 

394: p_0 = \omega_t \sqrt{\varepsilon_b}/c$ couple with bulk photons, i.e., 

395: can radiatively decay into the bulk photon modes. In this case 

396: Eq.\,(\ref{pol}) yields the X radiative decay rate into bulk in-plane 

397: ($Y$-) polarized transverse photons:  

398: \begin{equation} 

399: {1 \over \hbar}\Gamma^{\rm QW}_{\rm X}(p_{\|}) = {\varepsilon_b 

400: \over c^2} R^{\rm QW}_{\rm X} { \omega_t \over \sqrt{p^2_0 - 

401: p_{\|}^2} } \ .

402: \label{rad} 

403: \end{equation} 

404: One can also re-write Eq.\,(\ref{rad}) as 

405: $\Gamma^{\rm QW}_{\rm X}(p_{\|}) = \Gamma^{\rm QW}_{\rm X}(p_{\|}$=0) 

406: $p_0 / (p^2_0 - p_{\|}^2)^{1/2}$, where 

407: $\Gamma^{\rm QW}_{\rm X}(p_{\|}$=0) = $\hbar (\sqrt{\varepsilon_b} / c) 

408: R^{\rm QW}_{\rm X}$ is the radiative width of a QW exciton with 

409: in-plane momentum $\hbar p_{\|}=0$. In high-quality GaAs QWs 

410: at low temperatures, the condition $\Gamma^{\rm QW}_{\rm X} \gg \hbar 

411: \gamma_{\rm X}$ can be achieved, so that the X dispersion within the 

412: photon cone is approximated by $\hbar 

413: \omega^{\rm QW}_{\rm X}(p_{\|}$$\leq$$p_0) = \hbar \omega_t + \hbar^2 

414: p_{\|}^2/2M_x - i \Gamma^{\rm QW}_{\rm X}(p_{\|})/2$. 

415: 

416: The oscillator strength $R_{\rm X}^{\rm QW}$ associated with QW excitons 

417: is given by 

418: \begin{equation}

419: R_{\rm X}^{\rm QW} = {4 \pi \over \hbar} \, {\omega_t \over \varepsilon_b} 

420: \, |\phi_{\rm X}^{\rm (2D)}({\bf r} = 0)|^2 \, |d_{\rm cv}|^2 \, ,

421: \label{QWstr}

422: \end{equation}

423: where $\phi_{\rm X}^{\rm (2D)}(r)$ is the X wavefunction of relative 

424: electron-hole motion, and $d_{\rm cv}$ is the dipole matrix element of 

425: the interband optical transition. In the limits of strong and weak QW 

426: confinement Eq.\,(\ref{QWstr}) yields 

427: \begin{equation}

428: R^{\rm QW}_{\rm X} = \left\{

429: \begin{array}{ll}

430: 16 a_{\rm X}^{\rm (3D)} \omega_{\ell t} \omega_t \, , 

431: \ \ \ \ \ \  \ \ \ \ \ \ \ \ a^{\rm (3D)}_{\rm X} \gg d_z \, , \\

432: 2d_z \omega_{\ell t} \omega_t \, , \ \ \ \ \ \lambda = 2 \pi/p_0 \gg  

433: d_z \gtrsim a^{\rm (3D)}_{\rm X} \, , \\ 

434: \end{array}

435: \right.

436: \label{strength}

437: \end{equation}

438: respectively, where $a^{\rm (3D)}_{\rm X}$ is the Bohr radius of bulk 

439: excitons and $\omega_{\ell t}$ is the longitudinal-transverse splitting 

440: associated with bulk excitons (in bulk GaAs one has $a^{\rm (3D)}_{\rm X} 

441: \simeq 136\,\AA$ and $\hbar \omega_{\ell t} \simeq 80-86\,\mu{\rm eV}$, 

442: respectively \cite{Ulbrich82}). Thus we estimate the upper limit of the 

443: oscillator strength in narrow GaAs QWs as $\hbar^2 

444: R^{\rm QW}_{\rm X}(d_z$$\rightarrow$$0) \simeq 0.26 - 0.28\,{\rm eV}^2\AA$. 

445: For our GaAs QWs with weak confinement of excitons one evaluates from 

446: Eq.\,(\ref{strength}) that $\hbar^2 R^{\rm QW}_{\rm X}(d_z$$=$$250\,\AA) 

447: \simeq 0.061\,{\rm eV}^2\AA$. 

448: 

449: (ii) {\it The MC polariton dispersion relevant to excitonic molecules 

450: in (GaAs-based) microcavities.} 

451: %

452: The dispersion equation for MC polaritons, which contribute to the 

453: XX-mediated optics of a $\lambda$-cavity we study in our experiments, is 

454: given by 

455: \begin{eqnarray}

456: \omega^2_t &+& 

457: \hbar \omega_t p_{\|}^2/M_x - i \omega \gamma_{\rm X} - \omega^2 = 

458: \nonumber \\ 

459: &=&  \omega^2 \Bigg[ { (\Omega^{\rm MC}_{\rm 1 \lambda})^2 \over 

460: (\omega^{\gamma}_{1 \lambda})^2 - i \omega \gamma_{\rm R} - \omega^2 } 

461: \ + \ { (\Omega^{\rm MC}_{\rm 0 \lambda})^2 \over 

462: (\omega^{\gamma}_{0 \lambda})^2 - i \omega \gamma_{\rm R} - 

463: \omega^2 } \Bigg] \, , 

464: \nonumber \\

465: \label{MC}

466: \end{eqnarray}

467: where the photon frequencies, associated with the 1$\lambda$- and 

468: 0$\lambda$- microcavity eigenmodes, are $\omega^{\gamma}_{1 \lambda} =  

469: \omega^{\gamma}_{1 \lambda}(p_{\|}) = ( c^2 p_{\|}^2/\varepsilon_b + 

470: \omega_0^2 )^{1/2}$ and $\omega^{\gamma}_{0 \lambda} =  

471: \omega^{\gamma}_{0 \lambda}(p_{\|}) = c p_{\|}/\sqrt{\varepsilon_b}$, 

472: respectively. Here $\omega_0 = (2 \pi c)/(L_z \sqrt{\varepsilon_b})$ is 

473: the cavity eigenfrequency, $L_z$ is the MC thickness, and 

474: $\gamma_{\rm R}$ is the inverse radiative lifetime of MC photons, due 

475: to their escape from the microcavity into external bulk photon modes. 

476: The MC Rabi frequency $\Omega^{\rm MC}_{\rm 1 \lambda}$ refers to 

477: 1$\lambda$-eigenmode of the light field, ${\hat {\rm e}}_{1 \lambda}(z) 

478: = \sqrt{2/L_z} \cos[(2 \pi z)/L_z]$ (we assume that the QW is located 

479: at $z=0$ so that $|z| \leq L_z/2$), and is determined by 

480: \begin{equation}

481: (\Omega^{\rm MC}_{\rm 1 \lambda})^2 = {16 \pi \over \hbar} \, 

482: {\omega_t \over \varepsilon_b} \, |\phi_{\rm X}^{\rm (2D)}({\bf r} 

483: = 0)|^2 \, |d_{\rm cv}|^2 \, { |I_1|^2 \over L_z} \, ,

484: \label{MCstr1}

485: \end{equation}

486: where $I_1 = I_1(d_z/L_z) = [L_z/(\pi d_z)] \sin[(\pi d_z)/L_z] \simeq 1 - 

487: (\pi/6)(d_z/L_z)^2$. In turn, the Rabi frequency $\Omega^{\rm MC}_{\rm 0 

488: \lambda}$ is associated with 0$\lambda$-eigenmode of the MC light field, 

489: ${\hat {\rm e}}_{0 \lambda}(z) = 1/\sqrt{L_z} = {\rm const.}$, and 

490: \begin{equation}

491: (\Omega^{\rm MC}_{\rm 0 \lambda})^2 = {8 \pi \over \hbar} \, 

492: {\omega_t \over \varepsilon_b} \, |\phi_{\rm X}^{\rm (2D)}({\bf r} = 0)|^2 

493: \, |d_{\rm cv}|^2 \, { 1 \over L_z} \, .

494: \label{MCstr0}

495: \end{equation}

496: From Eqs.\,(\ref{QWstr}), (\ref{MCstr1}), and (\ref{MCstr0}) one gets

497: \begin{equation}

498: (\Omega^{\rm MC}_{\rm 1 \lambda})^2 = 2 |I_1|^2  

499: (\Omega^{\rm MC}_{\rm 0 \lambda})^2 = 4 \, { R_{\rm X}^{\rm MC} \over 

500: L_z } \, |I_1|^2  \, .

501: \label{strR}

502: \end{equation}

503: Because the factor $|I_1|^2 \simeq 1$ (for our microcavities $d_z = 

504: 250\,\AA$ and $L_z \simeq 2326\,\AA$, so that $|I_1|^2 \simeq 0.96$), 

505: we conclude that $\Omega^{\rm MC}_{\rm 1 \lambda} \simeq \sqrt{2} 

506: \Omega^{\rm MC}_{\rm 0 \lambda} \simeq 2 (R_{\rm X}^{\rm MC}/L_z)^{1/2}$. 

507: The factor two difference between $(\Omega^{\rm MC}_{\rm 1 \lambda})^2$ 

508: and $(\Omega^{\rm MC}_{\rm 0 \lambda})^2$ originates from the difference 

509: of the intensities of the light fields associated with microcavity 

510: 1$\lambda$- and 0$\lambda$-eigenmodes at the QW position, $z=0$, i.e., 

511: is due to $|{\hat {\rm e}}_{1 \lambda}(z$=$0)|^2/|{\hat {\rm e}}_{0 

512: \lambda}(z$=$0)|^2 = 2$. 

513: 

514: \begin{figure}

515: \includegraphics*[width=8cm]{figure1.eps}

516: \caption[]{ Three-branch microcavity polariton dispersion calculated 

517: with Eq.\,(\ref{MC}) for zero-detuning. The parameters are adapted to 

518: the GaAs microcavities used in our experiments: $\hbar 

519: \Omega^{\rm MC}_{1\lambda} = 3.70$\,meV, $\hbar \Omega^{\rm MC}_{0\lambda} 

520: = 2.67$\,meV, $\varepsilon_b = 12.3$, $M_x = 0.4\,m_0$, and 

521: $E_{\rm X}(p_{\|}$=$0) = \hbar \omega_t = 1.5219$\,eV. The dashed lines 

522: show the 1$\lambda$-mode MC photon and exciton dispersions (the 

523: 0$\lambda$-mode photon dispersion is not plotted). } 

524: \end{figure}

525: 

526: Thus, the dispersion Eq.\,(\ref{MC}) deals with a {\it three-branch} 

527: MC polariton model. In Fig.\,1 we plot the polariton dispersion 

528: branches, designated by 1$\lambda$-UB (upper branch), 1$\lambda$-LB 

529: (middle branch), and 0$\lambda$-LB (lower branch), respectively, and 

530: calculated by Eq.\,(\ref{MC}) for a zero-detuning GaAs-based microcavity 

531: with $\hbar \Omega^{\rm MC}_{\rm 1 \lambda} = 3.70$\,meV and $\hbar 

532: \Omega^{\rm MC}_{\rm 0 \lambda}= 2.67$\,meV. The ratio between the Rabi 

533: frequencies satisfies Eq.\,(\ref{strR}), and the used value of 

534: $\Omega^{\rm MC}_{\rm 1 \lambda}$ corresponds to that observed in our 

535: experiments. For small in-plane wavevectors $|{\bf p}_{\|}| \lesssim 

536: p_{\|}^{(1 \lambda)} \simeq 0.5 \times 10^5\,{\rm cm}^{-1}$ (see 

537: Fig.\,1) the 1$\lambda$-UB and 1$\lambda$-LB dispersion curves are 

538: identical to the upper and lower MC polariton branches calculated within 

539: the standard 1$\lambda$-eigenmode resonant approximation 

540: \cite{SkolnickSST98,KhitrovaRMP99}. In this case the 1$\lambda$-UB 

541: and 1$\lambda$-LB polaritons are purely 1$\lambda$-eigenwaves; the 

542: 0$\lambda$-LB dispersion is well-separated from the X resonance so that 

543: in Eq.\,(\ref{MC}) one can put $\Omega^{\rm MC}_{\rm 0 \lambda} = 0$ in 

544: order to describe the 1$\lambda$-UB and 1$\lambda$-LB dispersions in the 

545: wavevector domain $p_{\|} \lesssim p_{\|}^{(1 \lambda)}$. The 

546: anti-crossing between the X dispersion $\omega_t + \hbar p_{\|}^2/2M_x$ 

547: and the MC 0$\lambda$-mode photon frequency $c p_{\|}/

548: \sqrt{\varepsilon_b}$, which occurs at $p_{\|} = p_0 \simeq 2.7 \times 

549: 10^5\,{\rm cm}^{-1}$, gives rise to the MC 0$\lambda$-eigenmode 

550: dispersion associated with the 1$\lambda$-LB and 0$\lambda$-LB 

551: short-wavelength polaritons with $p_{\|} \gg p_{\|}^{(1\lambda)}$  

552: (see Fig.\,1). This picture is akin to the two-branch polariton 

553: dispersion in bulk semiconductors; for $p_{\|} \simeq p_0$ the 

554: 1$\lambda$-LB and 0$\lambda$-LB polariton dispersion can accurately 

555: be approximated by Eq.\,(\ref{MC}) with $\Omega^{\rm MC}_{\rm 1 \lambda} 

556: = 0$. In this case  Eq.\,(\ref{MC}) becomes identical to the dispersion 

557: equation for bulk polaritons, if in the latter the bulk Rabi splitting 

558: $\Omega^{\rm bulk}$ ($\hbar \Omega^{\rm bulk} \simeq 15.6$\,meV in GaAs) 

559: is replaced by $\Omega^{\rm MC}_{\rm 0 \lambda}$ and the bulk photon 

560: wavevector $p$ is replaced by $p_{\|}$. Note that for the MC 

561: 0$\lambda$-eigenmode the light field is homogeneous in the $z$-direction 

562: within the microcavity, i.e., for $|z| \leq L_z/2$. With increasing 

563: detuning from the X resonance the 1$\lambda$-LB and 0$\lambda$-LB 

564: polariton dispersions approach the photon frequencies 

565: $\omega^{\gamma}_{0 \lambda} = c p_{\|} / \sqrt{\varepsilon_b}$ and 

566: ${\tilde \omega}^{\gamma}_{0 \lambda} = c p_{\|} / 

567: \sqrt{\varepsilon^{(0)}_b}$, respectively, where the low-frequency 

568: dielectric constant is given by $\varepsilon^{(0)}_b = \varepsilon_b [1 + 

569: (\Omega_{0\lambda}^{\rm MC}/\omega_t)^2]$. The interconnection between 

570: two MC polariton domains occurs via the 1$\lambda$-LB polariton 

571: dispersion: With increasing $p_{\|}$ from $p_{\|} \lesssim p_{\|}^{(1 

572: \lambda)}$ towards $p_{\|} \gtrsim p_0$ the structure of the photon 

573: component of 1$\lambda$-LB polaritons smoothly changes, as a 

574: superposition of two modes, from purely 1$\lambda$-mode to purely 

575: 0$\lambda$-mode. 

576: 

577: % Zhopa

578: 

579: Because $1/a^{\rm (2D)}_{\rm XX} > p_0$, the non-zero exciton component 

580: of all three MC polariton dispersion branches contributes to the molecule 

581: state and, therefore, to the XX-mediated optics of microcavities. The 

582: X component, associated with the 0$\lambda$-LB, 1$\lambda$-LB, and 

583: 1$\lambda$-UB dispersions, is given by 

584: \begin{eqnarray}

585: (u^{\rm MC}_i)^2 = \Bigg[ 1 &+& { \omega^4_i (\Omega_{1 

586: \lambda}^{\rm MC})^2 \over \omega_t \omega^{\gamma}_{1 \lambda}

587: [\omega^2_i - (\omega^{\gamma}_{1 \lambda})^2]^2 } 

588: \nonumber \\

589: &+& { \omega^4_i (\Omega_{0 \lambda}^{\rm MC})^2 \over \omega_t 

590: \omega^{\gamma}_{0 \lambda} [\omega^2_i - 

591: (\omega^{\gamma}_{0 \lambda})^2]^2 } \Bigg]^{-1} \, , 

592: \label{MCcomp}

593: \end{eqnarray}

594: where $\omega_{i = {\rm 0\lambda LB,1\lambda LB, 1\lambda UB}} = 

595: \omega_{i = {\rm 0\lambda LB,1\lambda LB, 1\lambda UB}}^{\rm MC}(p_{\|})$ 

596: are the polariton dispersion branches calculated with Eq.\,(\ref{MC}). 

597: For a given $p_{\|}$ the X components satisfy the sum rule, 

598: $(u^{\rm MC}_{\rm 0\lambda LB})^2 + (u^{\rm MC}_{\rm 1\lambda LB})^2 + 

599: (u^{\rm MC}_{\rm 1\lambda UB})^2 = 1$. The exciton components, which 

600: correspond to the 0$\lambda$-LB, 1$\lambda$-LB, and 1$\lambda$-UB 

601: dispersions shown in Fig.\,1, are plotted in Fig.\,2. The above 

602: polariton branches have non-zero X component when the frequencies 

603: $\omega_i^{\rm MC}(p_{\|})$ resonate with the X state, i.e., at $p_{\|} 

604: \lesssim p_{\|}^{(1 \lambda)}$ for the 1$\lambda$-UB, at $p_{\|} \lesssim 

605: p_0$ for the 1$\lambda$-LB, and at $p_{\|} \gtrsim p_0$ for the 

606: 0$\lambda$-LB, respectively (see Fig.\,2). In our microcavities the X 

607: component of the $2\lambda$-, $3\lambda$- etc. eigenmode MC polaritons 

608: is negligible. 

609: 

610: \begin{figure}

611: \includegraphics*[width=8cm]{figure2.eps}

612: \caption[]{ The exciton component of 0$\lambda$-LB (dotted line), 

613: 1$\lambda$-LB (solid line), and 1$\lambda$-UB (dashed line) polaritons 

614: in a zero-detuning GaAs microcavity. } 

615: \end{figure}

616: 

617: A non-ideal optical confinement of the MC photon modes by distributed 

618: Bragg reflectors (DBRs) leads to the leakage of MC photons and gives rise 

619: to the radiative rate $\gamma_{\rm R}$ in Eq.\,(\ref{MC}). Thus the 

620: radiative width of MC polaritons, due to their optical escape through 

621: the DBRs, is $\Gamma^{\rm MC}_{i={\rm 0\lambda LB,1\lambda LB,1\lambda 

622: UB}} = \hbar (v^{\rm MC}_{i={\rm 0\lambda LB,1\lambda LB,1\lambda UB}})^2 

623: \gamma_{\rm R}$, where the photon component of the polaritons is given 

624: by $(v_i^{\rm MC})^2 = 1 - (u_i^{\rm MC})^2$. Note that for our 

625: GaAs-based macrocavities at low temperatures, $\gamma_{\rm R}$ is much 

626: larger than $\gamma_{\rm X}$ \cite{LangbeinPRB2000,BorriPRB2000}. 

627: 

628: 

629: \subsection{Bipolaritons in GaAs quantum wells}

630: 

631: 

632: The quasi-2D excitonic molecules in single QWs without co-planar optical 

633: confinement of the light field can either resonantly dissociate into 

634: interface polaritons or decay radiatively into the bulk photon modes. In our 

635: optical experiments, which deal with pump and signal bulk photons only, 

636: the first route of the XX optical decay cannot be visualized directly. Thus 

637: this channel refers to the ``hidden'' optics associated with the evanescent 

638: light field resonantly guided by QW excitons.

639: 

640: (i) {\it Resonant dissociation of QW excitonic molecules into outgoing 

641: interface polaritons.} The bipolariton model allows us to calculate the 

642: XX radiative width $\Gamma_{\rm XX}^{\rm QW(1)} = 

643: \Gamma_{\rm XX}^{\rm QW(1)}({\bf K}_{\|})$, associated with the resonant 

644: dissociation of the molecule with in-plane translational momentum 

645: $\hbar {\bf K}_{\|}$, by solving the wave equation 

646: \cite{Ivanov95,Ivanov98}: 

647: \begin{eqnarray}

648: &&\Big[ E_{\rm IP}({\bf p}_{\|}+{\bf K}_{\|} / 2) +  

649: E_{\rm IP}(-{\bf p}_{\|} + {\bf K}_{\|} / 2 ) 

650: \Big] \tilde{\Psi}_{\rm XX}({\bf p}_{\|},{\bf K}_{\|})

651: \nonumber \\ 

652: &&+ \ f_{\rm IP}({\bf p}_{\|},{\bf K}_{\|}) \sum_{\bf p'_{\|}} 

653: W_{\sigma^+\sigma^-}({\bf p}_{\|} - {\bf p'}_{\|}) 

654: \tilde{\Psi}_{\rm XX}({\bf p'}_{\|},{\bf K}_{\|})

655: \nonumber \\ 

656: && \ \ \ \ \ \ \ \ \ = \tilde{E}_{\rm XX}^{\rm QW}({\bf K}_{\|}) 

657: \tilde{\Psi}_{\rm XX}({\bf p}_{\|},{\bf K}_{\|}) \, . 

658: \label{BPint}

659: \end{eqnarray}

660: Here $\tilde{\Psi}_{\rm XX}$ and $\tilde{E}^{\rm QW}_{\rm XX}$ are 

661: the bipolariton (XX) wavefunction and energy, respectively, $E_{\rm IP} = 

662: \hbar \omega_{\rm IP}$ is the QW polariton energy determined by the 

663: dispersion Eq.\,(\ref{pol}), $f_{\rm IP}({\bf p}_{\|},{\bf K}_{\|}) = 

664: u_{\rm IP}^2({\bf p}_{\|}+{\bf K}_{\|} / 2) u_{\rm IP}^2(-{\bf p}_{\|}

665: +{\bf K}_{\|} / 2)$, where $u_{\rm IP}^2$ is given by Eq.\,(\ref{comp}), 

666: $\hbar {\bf p}_{\|}$ is the in-plane momentum of the relative motion 

667: of the optically-dressed constituent excitons, and $W_{\sigma^+\sigma^-}$ 

668: is the attractive Coulombic potential between $\sigma^+$- and $\sigma^-$- 

669: polarized QW excitons. The complex bipolariton energy can also be 

670: rewritten as $\tilde{E}_{\rm XX}^{\rm QW} = 2 E_{\rm X} - 

671: \epsilon_{\rm XX}^{(0)} + \Delta_{\rm XX}^{\rm QW} - i 

672: \Gamma_{\rm XX}^{\rm QW}/2$, where $\epsilon_{\rm XX}^{(0)}$ is the XX 

673: binding energy with no renormalization by the coupling with the vacuum 

674: light field. For the non-local deuteron model potential 

675: $W_{\sigma^+\sigma^-}(|{\bf p}_{\|}-{\bf p'}_{\|}|)$, which yields 

676: within the standard Schr\"odinger two-particle (two-X) equation the 

677: wavefunction $\Psi_{\rm XX}^{(0)}(p_{\|}) = 2 \sqrt{2 \pi} 

678: a_{\rm XX}^{\rm (2D)} / [(p_{\|} a_{\rm XX}^{\rm (2D)})^2 + 1]^{3/2}$ 

679: for an optically inactive molecule, the bipolariton wave 

680: Eq.\,(\ref{BPint}) is exactly-solvable \cite{Ivanov95}. The input 

681: parameters of the model are the binding energy $\epsilon_{\rm XX}^{(0)}$ 

682: and the oscillator strength $R_{\rm X}^{\rm QW}$. Thus the 

683: exactly-solvable bipolariton model simplifies the exciton-exciton 

684: interaction, but treats rigorously the (interface) polariton effect. 

685: 

686: (ii) {\it Resonant decay of QW excitonic molecules into the bulk photon 

687: modes.} The decay occurs when at least one of the constituent excitons 

688: of a QW molecule moves within the radiative zone, i.e., when 

689: $|+{\bf p}_{\|} + {\bf K}_{\|}/2| \leq p_0$ and/or $|-{\bf p}_{\|} + 

690: {\bf K}_{\|}/2| \leq p_0$. Note that the exciton-exciton resonant coherent 

691: Coulombic scattering within the molecule state intrinsically couples the 

692: X radiative and QW polariton modes. Thus the XX width, associated with 

693: the optical decay into the bulk photon modes, is given by 

694: \begin{eqnarray}

695: &&\Gamma_{\rm XX}^{\rm QW(2)}({\bf K}_{\|}\!=\!0) = {1 \over \pi}

696:  \int_0^{p_0} |\Psi_{\rm XX}^{(0)}(2p_{\|})|^2  

697: \Gamma_{\rm X}^{\rm QW}(p_{\|}) p_{\|} d p_{\|}

698: \nonumber \\  

699: && \ \ \ \ \ \ \ \ 

700: = \hbar { \sqrt{\varepsilon_b} \over c } R_{\rm X}^{\rm QW} 

701: { \sqrt{\chi} \over 4 (1 + \chi)^{5/2} } 

702: \Big[ (5 + 2 \chi) \sqrt{\chi (1 + \chi)}

703: \nonumber \\ 

704: && \ \ \ \ \ \ \ \ 

705: + 3 \ln( \sqrt{1 + \chi} + \sqrt{\chi}) \Big] \, , 

706: \label{rada}

707: \end{eqnarray}

708: where $\chi = 4 \delta^{\rm (2D)}_{\rm R} \equiv 4 (a^{\rm (2D)}_{\rm XX} 

709: p_0)^2$ and $\Gamma_{\rm X}^{\rm QW}(p_{\|})$ is given by 

710: Eq.\,(\ref{rad}). In the above integral over the QW radiative zone we 

711: approximate $\Psi_{\rm XX}^{(0)}$ by the  deuteron wavefunction. For 

712: $\chi \ll 1$ Eq.\,(\ref{rada}) yields $\Gamma_{\rm XX}^{\rm QW(2)} \simeq 

713: 2 \hbar (\sqrt{\varepsilon_b}/c) \chi R_{\rm X}^{\rm QW}$ = 

714: $8 (a^{\rm (2D)}_{\rm XX} p_0)^2 \Gamma_{\rm X}^{\rm QW}(p_{\|}$=0). 

715: However, for our reference GaAs QW with weak confinement of the electronic 

716: states one has $\chi \simeq 1.2$ so that the above simple approximation 

717: of Eq.\,(\ref{rada}) cannot be used. 

718: 

719: \begin{figure}

720: \includegraphics*[width=8cm]{figure3.eps}

721: \caption[]{ The calculated radiative decay widths of the exciton and 

722: bipolariton states versus the oscillator strength $R^{\rm QW}_{\rm X}$. 

723: The XX radiative widths associated with the decay into interface 

724: polaritons, $\Gamma_{\rm XX}^{\rm QW(1)}$, and into bulk photon modes, 

725: $\Gamma_{\rm XX}^{\rm QW(2)}$, are plotted separately. The input XX 

726: binding energy $\epsilon^{(0)}_{\rm XX} = 1.1$\,meV. The two circle 

727: symbols show $\Gamma^{\rm QW}_{\rm X}$ and $\Gamma^{\rm QW}_{\rm XX}$ 

728: inferred from the experimental data. } 

729: \end{figure}

730: 

731: In Fig.\,3 we plot the radiative widths 

732: $\Gamma_{\rm X}^{\rm QW}({\bf p}_{\|}$=0), 

733: $\Gamma_{\rm XX}^{\rm QW(1)}({\bf K}_{\|}$=0), 

734: $\Gamma_{\rm XX}^{\rm QW(2)}({\bf K}_{\|}$=0), and 

735: $\Gamma_{\rm XX}^{\rm QW(1)}({\bf K}_{\|}$=0) + 

736: $\Gamma_{\rm XX}^{\rm QW(2)}({\bf K}_{\|}$=0) against the oscillator 

737: strength of QW excitons $R_{\rm X}^{\rm QW}$. The widths are calculated 

738: with Eqs.\,(\ref{BPint}) and (\ref{rada}) for the input XX binding energy 

739: $\epsilon_{\rm XX}^{(0)} = 1.1$\,meV. As we discuss in Section III, the 

740: oscillator strength $R_{\rm X}^{\rm QW}$ of the high-quality reference QW 

741: used in our experiments is given by 

742: $\hbar^2 R_{\rm X}^{\rm QW}(d_z$=$250\,\AA) \simeq 0.035\,{\rm eV}^2\AA$. 

743: The above value, which is inferred from the experimental data, refers to 

744: the GaAs QW sandwiched between semi-infinite bulk AlGaAs barriers and is 

745: consistent with that estimated in the previous Subsection by using 

746: Eq.\,(\ref{strength}). A cap layer on top of the reference single QW 

747: modifies the evanescent field associated with interface 

748: polaritons and reduces the oscillator strength to $\hbar^2 

749: {\tilde R}_{\rm X}^{\rm QW}(d_z$=$250\,\AA) \simeq  0.028\,{\rm eV}^2\AA$  

750: (for the details see Section IV). As shown in Fig.\,3, for $\hbar^2 

751: R_{\rm X}^{\rm QW} = 0.035\,{\rm eV}^2\AA$ Eqs.\,(\ref{BPint}) and 

752: (\ref{rada}) yield $\Gamma_{\rm XX}^{\rm QW(1)}({\bf K}_{\|}$=0) $\simeq 

753: 148\,\mu {\rm eV}$ and $\Gamma_{\rm XX}^{\rm QW(2)}({\bf K}_{\|}$=0) 

754: $\simeq 33\,\mu {\rm eV}$, so that the total XX radiative width is given by 

755: $\Gamma_{\rm XX}^{\rm QW}({\bf K}_{\|}$=0) = $\Gamma_{\rm XX}^{\rm QW(1)} 

756: + \Gamma_{\rm XX}^{\rm QW(2)} \simeq 0.18$\,meV. For $\hbar^2 

757: {\tilde R}_{\rm X}^{\rm QW} \simeq  0.028\,{\rm eV}^2\AA$ one calculates 

758: $\Gamma_{\rm XX}^{\rm QW(1)}({\bf K}_{\|}$=0) $\simeq 100\,\mu {\rm eV}$,   

759: $\Gamma_{\rm XX}^{\rm QW(2)}({\bf K}_{\|}$=0) $\simeq 26\,\mu {\rm eV}$, 

760: and $\Gamma_{\rm XX}^{\rm QW}({\bf K}_{\|}$=0) = 

761: $\Gamma_{\rm XX}^{\rm QW(1)} + \Gamma_{\rm XX}^{\rm QW(2)} \simeq 

762: 0.126$\,meV. The latter value is indeed very close to the XX radiative 

763: width $\Gamma_{\rm XX}^{\rm QW} \simeq 0.1$\,meV inferred from our 

764: opical experiments with the reference QW (see Section III). 

765: 

766: The photon-assisted resonant dissociation of QW molecules into outgoing 

767: interface polaritons is more efficient than the XX optical decay into the 

768: bulk photon modes by factor 4.5 for $\hbar^2 R_{\rm X}^{\rm QW} \simeq 

769: 0.035\,{\rm eV}^2\AA$ and by factor 3.8 for $\hbar^2 

770: {\tilde R}_{\rm X}^{\rm QW} \simeq  0.028\,{\rm eV}^2\AA$, respectively. 

771: This conclusion is consistent with that of Ref.\,[\onlinecite{Ivanov97}], 

772: where for the limit of strong QW confinement ($d_z \rightarrow 0$) the 

773: relative efficiency of the two optical decay channels was estimated to be 

774: $\Gamma_{\rm XX}^{\rm QW(1)} : \Gamma_{\rm XX}^{\rm QW(2)} \simeq 25 : 1$. 

775: The latter ratio refers to the idealized case of an extremely narrow GaAs 

776: QW surrounded by infinitely thick AlGaAs barriers. The resonant optical 

777: dissociation of the QW molecules into interface polaritons is much 

778: stronger than the radiative decay into the bulk photon modes, because the 

779: constituent excitons in their relative motion move mainly outside the 

780: radiative zone, with the in-plane momenta $|\pm {\bf p}_{\|} + 

781: {\bf K}_{\|}/2| \gtrsim p_0$. In this case the excitons are optically 

782: dressed by the evanescent light field, i.e., they exist as QW polaritons 

783: and, therefore, decay mainly into the confined, QW-guided interface 

784: modes. The picture can also be justified by analyzing the joint density 

785: of states relevant to the two optical decay channels. Note that in both 

786: main equations, Eq.\,(\ref{BPint}) and Eq.\,(\ref{rada}), 

787: $\delta^{\rm (2D)}_{\rm R} = (a^{\rm (2D)}_{\rm XX} p_0)^2$ does 

788: represent the dimensionless smallness parameter of the (bipolariton) 

789: model. 

790: 

791: 

792: \subsection{ Bipolaritons in GaAs-based microcavities }

793: 

794: 

795: The bipolariton model for excitonic molecules in $\lambda$-microcavities 

796: requires to construct the XX state in terms of quasi-bound 0$\lambda$-LB, 

797: 1$\lambda$-LB, and 1$\lambda$-UB polaritons. In this case the radiative 

798: corrections to the XX state with ${\bf K}_{\|}\!=\!0$ are given by 

799: \begin{eqnarray}

800: \Delta^{\rm MC}_{\rm XX}({\bf K}_{\|}\!=\!0) &=& { 27 \over 8 } 

801: \sqrt{ \pi \over 2} \ \epsilon_{\rm XX}^{(0)} \ \mbox{Re} \Big\{ 

802: { A \over 1 + B } \Big\} \, , 

803: \nonumber \\

804: \Gamma^{\rm MC}_{\rm XX}({\bf K}_{\|}\!=\!0) &=& - { 27 \over 4 } 

805: \sqrt{ \pi \over 2} \ \epsilon_{\rm XX}^{(0)} \ \mbox{Im} \Big\{ 

806: { A \over 1 + B } \Big\} \, , 

807: \label{MCrad}

808: \end{eqnarray}

809: where

810: \begin{eqnarray}

811: A&=&{1 \over 2 \pi} \int_0^{+ \infty} \! \! \! \! \! \! \! p_{\|} dp_{\|}

812: \left[ {\tilde G}(p_{\|}) \left( \epsilon_{\rm XX}^{(0)} + {\hbar^2 

813: p_{\|}^2 \over M_x} \right) + 1 \right] \Psi_{\rm XX}^{(0)}(p_{\|}) \, , 

814: \nonumber \\

815: B&=&{ 27 \over 16 } {1 \over \sqrt{2 \pi} } \ \epsilon_{\rm XX}^{(0)} 

816: \int_0^{+ \infty} \! \! \! p_{\|} dp_{\|} \, {\tilde G}(p_{\|}) 

817: \Psi_{\rm XX}^{(0)}(p_{\|}) \, .

818: \label{mcAB}

819: \end{eqnarray}

820: In Eq.\,(\ref{mcAB}) the bipolariton Green function ${\tilde G}(p_{\|})$ 

821: is  

822: \begin{equation}

823: {\tilde G}(p_{\|}) =  \sum_{i,j} { [ u_i^{\rm MC}(p_{\|}) 

824: u_j^{\rm MC}(-p_{\|}) ]^2 \over E_{\rm XX}^{\rm MC} - 

825: \hbar \omega_i^{\rm MC}(p_{\|}) - \hbar \omega_j^{\rm MC}(-p_{\|}) 

826: + i \gamma_0 } \, , 

827: \label{Green}

828: \end{equation}

829: where $E_{\rm XX}^{\rm MC} = 2E_{\rm X}(p_{\|}$=$0) - 

830: \epsilon_{\rm XX}^{(0)}$, the MC polariton eigenfrequency 

831: $\omega_{i(j)}^{\rm MC}$ and the X component $[u_{i(j)}^{\rm MC}]^2$ with 

832: $i,j =$ 0$\lambda$-LB, 1$\lambda$-LB, and 1$\lambda$-UB are given by 

833: Eq.\,(\ref{MC}) and Eq.\,(\ref{MCcomp}), respectively, and $\gamma_0 

834: \rightarrow +0$. The XX radiative corrections, i.e., the Lamb shift 

835: $\Delta_{\rm XX}^{\rm MC}$ and the radiative width 

836: $\Gamma_{\rm XX}^{\rm MC}$, depend upon the relative motion of the 

837: constituent QW excitons over whole momentum space, i.e., 

838: Eqs.\,(\ref{MCrad})-(\ref{mcAB}) include integration over $d p_{\|}$. 

839: The change of the input XX binding energy, $\epsilon_{\rm XX}^{(0)} 

840: \rightarrow \epsilon^{\rm XX} = \epsilon_{\rm XX}^{(0)} - 

841: \Delta_{\rm XX}^{\rm MC}({\bf K}_{\|}) + (i/2)   

842: \Gamma_{\rm XX}^{\rm MC}({\bf K}_{\|})$, occurs because in their relative 

843: motion the constituent excitons move along the MC polariton dispersion 

844: curves, rather than possess the quadratic dispersion, $E_{\rm X} = \hbar 

845: \omega_t + \hbar^2 p_{\|}^2/(2M_x)$ (the latter is valid only for 

846: optically inactive excitons). 

847: 

848: The solution of the exactly-solvable bipolariton model, given by 

849: Eqs.\,(\ref{MCrad})-(\ref{Green}), includes all possible channels of the 

850: in-plane dissociation of the microcavity molecule into two outgoing MC 

851: polaritons, i.e., ``XX (${\bf K}_{\|}$=0) $\rightarrow$ 

852: $i$th-branch MC polariton ($\sigma^+, {\bf p}_{\|}$) $+$ $j$th-branch MC 

853: polariton ($\sigma^-, -{\bf p}_{\|}$)''. Note that the solution of the 

854: bipolariton wave Eq.\,(\ref{BPint}) for excitonic molecules in a single QW 

855: can be obtained from Eqs.\,(\ref{MCrad})-(\ref{Green}) by putting $i = j$ 

856: = IP and replacing $u^{\rm MC}_{i(j)}$ and $\omega_{i(j)}^{\rm MC}$ by 

857: $u_{\rm IP}$ and $\omega_{\rm IP}$, respectively. 

858: 

859: \begin{figure}

860: \includegraphics*[width=8cm]{figure4.eps}

861: \caption[]{ The radiative corrections to the excitonic molecule state, 

862: $\Gamma^{\rm MC}_{\rm XX}$ and $\Delta^{\rm MC}_{\rm XX}$, calculated 

863: against the MC detuning $\delta$ with Eqs.\,(\ref{MCrad})-(\ref{Green}) 

864: for the MC Rabi energies $\hbar \Omega^{\rm MC}_{1\lambda} = 7.76$\,meV 

865: and $\hbar \Omega^{\rm MC}_{0\lambda} = 5.60$\,meV. The input XX 

866: binding energy $\epsilon^{(0)}_{\rm XX}$ is 0.9\,meV (dash-dotted line), 

867: 1.0\,meV (solid line), and 1.1\,meV (dashed line). }

868: \end{figure}

869: 

870: The radiative width $\Gamma_{\rm XX}^{\rm MC (1)} = 

871: \Gamma_{\rm XX}^{\rm MC}(\delta)$ and the Lamb shift 

872: $\Delta^{\rm MC}_{\rm XX} = \Delta^{\rm MC}_{\rm XX}(\delta)$ calculated 

873: by Eqs.\,(\ref{MCrad})-(\ref{Green}) as a function of the MC detuning 

874: $\delta = \hbar (\omega_0 - \omega_t)$ between the 1$\lambda$ cavity mode 

875: and QW exciton are plotted in Fig.\,4 for three values of the input XX 

876: binding energy, $\epsilon_{\rm XX}^{(0)} = 0.9$\,meV, 1.0\,meV, and 

877: 1.1\,meV. By applying Eq.\,(\ref{strR}), we estimate for this plot the 

878: Rabi frequencies, $\Omega^{\rm MC}_{1\lambda}$ and 

879: $\Omega^{\rm MC}_{0\lambda}$, relevant to the used three-branch MC 

880: polariton dispersion given by Eq.\,(\ref{MC}). Namely, for $\hbar^2 

881: R^{\rm QW}_{\rm X} = 0.035\,\mbox{eV}^2 \AA$, associated with the 

882: reference QW, and $L_z = 2326\,\AA$, Eq.\,(\ref{strR}) yields 

883: $\hbar \Omega^{\rm MC}_{1\lambda} \simeq 7.76$\,meV and $\hbar 

884: \Omega^{\rm MC}_{0\lambda} \simeq 5.60$\,meV. As a result of non-ideal 

885: optical confinement in the $z$-direction by DBRs, our GaAs-based 

886: $\lambda$-microcavity (i) has a smaller value of 

887: $\Omega^{\rm MC}_{1\lambda}$, i.e., $\hbar \Omega^{\rm MC}_{1\lambda} 

888: \simeq 3.7$\,meV and (ii) with increasing $p_{\|}$ loses the strength 

889: of optical confinement for MC 1$\lambda$-mode photons of frequency 

890: $\omega^{\gamma} \simeq \omega_0$. The latter means that the MC photon 

891: radiative width $\gamma_{\rm R}$ is $p_{\|}$-dependent and smoothly 

892: increases with increasing $p_{\|}$. The DBR optical confinement is 

893: completely relaxed for $p_{\|} \sim p_0$ so that the dispersion 

894: Eq.\,(\ref{MC}) becomes inadequate, and the microcavity 0$\lambda$-LB 

895: polariton dispersion evolves towards the interface polariton dispersion, 

896: associated with the single QW and given by Eq.\,(\ref{pol}). Thus, in 

897: order to model the experimental data with 

898: Eqs.\,(\ref{MCrad})-(\ref{Green}), we use $\hbar 

899: \Omega^{\rm MC}_{1\lambda} \simeq 3.70$\,meV and $\hbar 

900: \Omega^{\rm MC}_{0\lambda} \simeq 2.67$\,meV, and replace the 

901: 0$\lambda$-LB polariton dispersion by the interface, QW polariton 

902: dispersion with $\hbar^2 R^{\rm QW}_{\rm X} = 0.035\,\mbox{eV}^2 \AA$. 

903: For this case the plot of $\Gamma_{\rm XX}^{\rm MC}$ and 

904: $\Delta^{\rm MC}_{\rm XX}$ against the detuning $\delta$ is shown in 

905: Fig.\,10 (for details see Section IV). 

906: 

907: There are two sharp spikes in the dependence $\Delta^{\rm MC}_{\rm XX} = 

908: \Delta^{\rm MC}_{\rm XX}(\delta)$ which are accompanied by the 

909: jump-like changes of the XX radiative width $\Gamma^{\rm MC}_{\rm XX} = 

910: \Gamma^{\rm MC}_{\rm XX}(\delta)$ (see Figs.\,4 and 10). The above 

911: structure is due to van Hove critical points, $M_1$ and $M_2$, in the 

912: joint density of the polariton states (JDPS) relevant to the optical 

913: decay ``MC excitonic molecule ${\bf K}_{\|}$=0 $\rightarrow$ MC polariton 

914: ${\bf p}_{\|}$ + MC polariton $-{\bf p}_{\|}$'' (for the critical points 

915: we use the classification and notations proposed in 

916: Ref.\,[\onlinecite{vanHove}]). The first critical point $M_1$ in 

917: energy-momentum space $\{ \delta, {\bf p}_{\|} \}$ refers to a negative 

918: MC detuning $\delta_1$ and deals with the condition $\mbox{Re} 

919: \{{\tilde E}^{\rm MC}_{\rm XX}({\bf K}_{\|}$=0$)\} = 2 \hbar \omega_t - 

920: \epsilon_{\rm XX}^{(0)} + \Delta^{\rm MC}_{\rm XX} = \hbar 

921: \omega^{\rm MC}_{\rm 1\lambda LB}({\bf p}_{\|}$=0$) + \hbar 

922: \omega^{\rm MC}_{\rm 1\lambda UB}(-{\bf p}_{\|}$=0). This point is 

923: marginal for the optical decay ``XX $\rightarrow$ 1$\lambda$-LB 

924: polariton + 1$\lambda$-UB polariton'': For $\delta \leq \delta_1$ 

925: the above channel is allowed, while it is absent for $\delta > 

926: \delta_1$. The critical point $M_2$ occurs at a positive detuning 

927: $\delta_2$, which corresponds to the condition $\mbox{Re} 

928: \{{\tilde E}^{\rm MC}_{\rm XX}({\bf K}_{\|}$=0$)\} = 2 \hbar \omega_t - 

929: \epsilon_{\rm XX}^{(0)} + \Delta^{\rm MC}_{\rm XX} = \hbar 

930: \omega^{\rm MC}_{\rm 1\lambda LB}({\bf p}_{\|}$=0$) + \hbar 

931: \omega^{\rm MC}_{\rm 1\lambda LB}(-{\bf p}_{\|}$=0), and is the main 

932: marginal point in the JDPS for the XX optical dissociation into two 

933: outgoing 1$\lambda$-LB polaritons. Namely, for $\delta \leq \delta_2$ 

934: the molecule can decay into two 1$\lambda$-LB polaritons, while for 

935: $\delta > \delta_2$ the optical decay of MC molecules with zero in-plane 

936: wavevector ${\bf K}_{\|}$ into $1\lambda$-LB polaritons is completely 

937: forbidden. With a very high accuracy of the order of $|\delta|/\omega_t 

938: \ll 1$, one finds from Eq.\,(\ref{MC}) that $\hbar \omega_{\rm 1\lambda 

939: UB/1\lambda LB}(p_{\|}$=$0) = \hbar \omega_t + (\delta/2) \pm (1/2) [ 

940: \delta^2 + (\hbar \Omega_{1 \lambda}^{\rm MC})^2 ]^{1/2}$. Thus from the 

941: energy-momentum conservation law we estimate the detunings $\delta_{1,2}$: 

942: \begin{eqnarray}

943: \! \! \! \! \! \! \! \!

944: \mbox{Critical point} \ M_1:&& \delta_1 = - \epsilon_{\rm XX}^{\rm MC} \, , 

945: \nonumber \\

946: \! \! \! \! \! \! \! \!

947: \mbox{Critical point} \ M_2:&& \delta_2 = { (\hbar 

948: \Omega_{1 \lambda}^{\rm MC})^2 - (\epsilon_{\rm XX}^{\rm MC})^2 \over 

949: 2 \epsilon_{\rm XX}^{\rm MC} } \, , 

950: \label{vanHove}

951: \end{eqnarray}

952: where $\epsilon_{\rm XX}^{\rm MC} = \epsilon_{\rm XX}^{(0)} - 

953: \Delta_{\rm XX}^{\rm MC}$ is the true, ``measured'' binding energy of the 

954: bipolariton state ${\bf K}_{\|}$=0, i.e., of the optically dressed 

955: molecule. 

956: 

957: \begin{figure}

958: \includegraphics*[width=8cm]{figure5.eps}

959: \caption[]{ The graphic solution of the energy-momentum conservation law 

960: for the optical decay of a MC molecule with ${\bf K}_{\|}=0$. The 

961: microcavity Rabi energies are $\hbar \Omega^{\rm MC}_{1\lambda} = 

962: 3.70$\,meV and $\hbar \Omega^{\rm MC}_{0\lambda} = 2.67$\,meV, the MC 

963: detuning is zero. The solutions are shown by the bold points $S_1$ 

964: (XX $\rightarrow$ 1$\lambda$-LB polariton + 1$\lambda$-LB polariton), 

965: $S_2$ (XX $\rightarrow$ 0$\lambda$-LB polariton + 0$\lambda$-LB 

966: polariton), and $S_{3,4}$ (XX $\rightarrow$ 0$\lambda$-LB polariton + 

967: 1$\lambda$-LB polariton). The efficiency of the last decay channel 

968: is negligible in comparison with that of the first two. 

969: The XX binding energy $\epsilon^{(0)}_{\rm XX} = 1$\,meV. } 

970: \end{figure}

971: 

972: \begin{figure}

973: \includegraphics*[width=8cm]{figure6.eps}

974: \caption[]{ The graphic solution of energy-momentum conservation for 

975: (a) the decay channels ``XX $\rightarrow$ 1$\lambda$-LB polariton + 

976: 1$\lambda$-LB polariton'' and ``XX $\rightarrow$ 1$\lambda$-LB 

977: polariton + 1$\lambda$-UB polariton'' (the MC detuning $\delta=-1$\,meV, 

978: the marginal solution $S_5$ at $p_{\|}=0$ refers to the critical point 

979: $M_1$); (b) the decay path ``XX $\rightarrow$ 1$\lambda$-LB polariton + 

980: 1$\lambda$-LB polariton'' (the MC detuning $\delta=0$); (c) the decay 

981: path ``XX $\rightarrow$ 1$\lambda$-LB polariton + 1$\lambda$-LB 

982: polariton'' (the MC detuning $\delta=6.345$\,meV, the marginal solution 

983: $S_1$ at $p_{\|}=0$ refers to the critical point $M_2$); (d) the decay 

984: channels ``XX $\rightarrow$ 0$\lambda$-LB polariton + 0$\lambda$-LB 

985: polariton'' and ``XX $\rightarrow$ 0$\lambda$-LB polariton + 

986: 1$\lambda$-UB polariton'' (this plot is practically independent of 

987: $\delta$). The MC Rabi frequencies $\Omega^{\rm MC}_{1\lambda}$ and 

988: $\Omega^{\rm MC}_{0\lambda}$, and the XX binding energy 

989: $\epsilon^{(0)}_{\rm XX}$ are the same as in Fig.\,5. } 

990: \end{figure}

991: 

992: In order to visualize the optical decay channels of MC excitonic 

993: molecules, in Figs.\,5 and 6 we plot the graphic solution of the 

994: energy-momentum conservation law, $E^{\rm MC}_{\rm XX} - \hbar 

995: \omega_i^{\rm MC}(p_{\|}) - \hbar \omega_j^{\rm MC}(-p_{\|}) = 0$ 

996: ($i,j =$ 0$\lambda$-LB, 1$\lambda$-LB, and 1$\lambda$-UB). The 

997: roots of the equation are the poles of the bipolariton Green 

998: function ${\tilde G}$ given by Eq.\,(\ref{Green}). Figure 5, which 

999: refers to the zero-detuning GaAs-based microcavity, clearly 

1000: illustrates that apart from the decay path ``XX $\rightarrow$ 

1001: 1$\lambda$-LB polariton + 1$\lambda$-LB polariton'' there are also the 

1002: decay routes which involve the 1$\lambda$-LB and 0$\lambda$-LB 

1003: microcavity polaritons with $p_{\|} \sim p_0$, i.e., ``XX $\rightarrow$ 

1004: 0$\lambda$-LB polariton + 0$\lambda$-LB polariton'' and 

1005: ``XX $\rightarrow$ 1$\lambda$-LB polariton + 0$\lambda$-LB polariton''. 

1006: The graphic solution of energy-momentum conservation for the wavevector 

1007: domain $p_{\|} \lesssim p_{\|}^{(1 \lambda)}$ is shown in a magnified 

1008: scale in Figs.\,6a-6c for $\delta = \delta_1$, 0, and $\delta_2$, 

1009: respectively. The touching points at $p_{\|}=0$ between the 

1010: 1$\lambda$-upper and 1$\lambda$-lower (see Fig.\,6a) and 

1011: 1$\lambda$-lower and 1$\lambda$-lower  (see Fig.\,6c) dispersion 

1012: branches correspond to the $M_1$ and $M_2$ critical points, respectively. 

1013: The graphic solution of the energy-momentum conservation law is shown in 

1014: Fig.\,6d for the vicinity of $p_{\|} = p_0$. According to Eq.\,(\ref{MC}), 

1015: the 1$\lambda$-LB and 0$\lambda$-LB polaritons with $p_{\|} \sim p_0$ 

1016: practically do not depend upon the MC detuning $\delta$, i.e., the plot 

1017: shown in Fig.\,6d is not sensitive to $\delta$. 

1018: 

1019: The value of the $\Gamma^{\rm MC}_{\rm XX}$-jump and 

1020: $\Delta^{\rm MC}_{\rm XX}$-spike nearby the critical point $M_1$, i.e., 

1021: at $\delta = \delta_1$, shows that the contribution of the decay path 

1022: ``XX $\rightarrow$ 1$\lambda$-UB polariton + 1$\lambda$-LB polariton'' 

1023: is rather small, about 1-2$\%$ only. This is mainly due to a small 

1024: value of the JDPS in the decay channel. The main contribution to the 

1025: XX radiative corrections in microcavities is due to the 

1026: frequency-degenerate decay routes ``XX $\rightarrow$ 1$\lambda$-LB 

1027: polariton + 1$\lambda$-LB polariton'' and ``XX $\rightarrow$ 

1028: 0$\lambda$-LB polariton + 0$\lambda$-LB polariton'' (or ``XX 

1029: $\rightarrow$ interface polariton + interface polariton'', as a result 

1030: of the relaxation of the transverse optical confinement at $p_{\|} \sim 

1031: p_0$). The JDPS associated with the first main channel is given by 

1032: \begin{eqnarray}

1033: \rho^{\rm XX \rightarrow 1\lambda LB + 1\lambda LB}_{\rm \hbar \omega 

1034: = \hbar \omega_t - \epsilon^{\rm MC}_{\rm XX}/2} &=& {\pi \over 2 \hbar 

1035: \omega_t} \left( a_{\rm XX}^{\rm (2D)} p_0 \right )^2 

1036: \nonumber \\ 

1037: \times \Bigg[ 1 &+& \left( { \Omega^{\rm MC}_{\rm 1 \lambda } 

1038: \over \epsilon_{\rm XX}^{(0)} } \right)^2 \Bigg] \Theta( \delta_2 - 

1039: \delta ) \, , 

1040: \label{JDPS}

1041: \end{eqnarray}

1042: where $\Theta(x)$ is the Heaviside step function. The above JDPS is 

1043: relevant to the calculations done by the bipolariton 

1044: Eqs.\,(\ref{MCrad})-(\ref{Green}). The appearance of the dimensionless 

1045: parameter $\delta_{\rm R}^{\rm (2D)} = ( a^{\rm (2D)}_{\rm XX} p_0 )^2$ 

1046: on the right-hand side (r.h.s.) of Eq.\,(\ref{JDPS}) is remarkable. Thus 

1047: the same control parameter $\delta_{\rm R}^{\rm (2D)}$ determines the 

1048: optical decay of excitonic molecules in the reference single GaAs QW and 

1049: in the GaAs-based microcavities. Furthermore, the JDPS given by 

1050: Eq.\,(\ref{JDPS}) depends upon the MC detuning only through the step 

1051: function $\Theta ( \delta_2 - \delta )$. The latter dependence gives 

1052: rise to the critical point $M_2$. By comparing the XX radiative 

1053: corrections for $\delta < \delta_2$ and $\delta > \delta_2$ (see 

1054: Figs.\,4 and 10), one concludes that the first main decay channel ``XX 

1055: $\rightarrow$ 1$\lambda$-LB polariton + 1$\lambda$-LB polariton'' has 

1056: nearly the same efficiency as the second one, ``XX $\rightarrow$ 

1057: 0$\lambda$-LB polariton + 0$\lambda$-LB polariton'' (or ``XX 

1058: $\rightarrow$ interface polariton + interface polariton''). Note that 

1059: the ``virtual'' decay paths, like ``XX $\rightarrow$ 1$\lambda$-UB 

1060: polariton + 1$\lambda$-UB polariton'', still contribute to the XX Lamb 

1061: shift in microcavities, according to Eqs.\,(\ref{MCrad})-(\ref{Green}). 

1062: 

1063: 

1064: \section{Experiment}

1065: 

1066: 

1067: The investigated sample consists of an MBE--grown 

1068: GaAs/Al$_{0.3}$Ga$_{0.7}$As single quantum well of the thickness $d_z = 

1069: 250\,\AA$ and placed in the center of a $\lambda$--cavity. An 

1070: AlAs/Al$_{0.15}$Ga$_{0.85}$As DBR of 25 (16) periods was grown at the 

1071: bottom (top) of the cavity. The spacer layer is wedged, in order to tune 

1072: the cavity mode along the position on the sample. Details on the growth 

1073: and sample design can be found in Ref.\,[\onlinecite{Jacob00}]. The 

1074: optical properties of the reference single QW grown under nominally 

1075: identical conditions are reported in Ref.\,[\onlinecite{LangbeinPRB2000}]: 

1076: The spectra show the ground-state heavy-hole (HH) and light-hole (LH) 

1077: exciton absorption lines separated in energy by about 2.6\,meV. In the 

1078: MC sample, the coupling of both HH and LH excitons with the 

1079: 1$\lambda$-mode cavity photons results in the formation of three 

1080: 1$\lambda$-eigenmode MC polariton dispersion branches, 1$\lambda$-LB, 

1081: 1$\lambda$-MB, and 1$\lambda$-UB \cite{Jacob00}. The 1$\lambda$-mode 

1082: polaritons have a narrow linewidth: The ratio between the HH Rabi 

1083: splitting and the polariton linewidths at zero detuning is about twenty 

1084: \cite{BorriPRB2000}. 

1085: 

1086: For the reference GaAs QW at temperature $T \lesssim 10$\,K the 

1087: homogeneous width ${\tilde \Gamma}^{\rm QW}_{\rm X}$ is dominated by 

1088: the radiative decay. The absorption linewidth, measured along the 

1089: $z$-direction and extrapolated to zero temperature, yields the HH--X 

1090: radiative width of $98 \pm 10\,\mu$eV. Note that this value is affected 

1091: by optical interference which occurs at the position of the QW, $z=0$, 

1092: due to bulk photons emitted by the QW excitons and partly reflected back 

1093: by the top surface ($z = L_{\rm cap} \simeq 499$\,nm) of a cap layer. In 

1094: this case one has a constructive interference which results in the 

1095: enhancement of the light field at $z=0$. By treating the optical 

1096: interference effect, we estimate $\Gamma_{\rm X}^{\rm QW} \simeq 60\,\mu 

1097: {\rm eV}$ for the reference QW sandwiched between semi-infinite bulk 

1098: AlGaAs barriers. This radiative width yields the intrinsic oscillator 

1099: strength of quasi-2D HH excitons $\hbar^2 

1100: R_{\rm X}^{\rm QW}(d_z$=$250\,\AA) \simeq 0.035\,{\rm eV}^2\AA$ (see 

1101: Fig.\,3). The measured characteristics of excitonic molecules in the 

1102: reference QW are consistent with those reported in 

1103: Ref.\,[\onlinecite{LangbeinPRB2000}]: The XX binding energy 

1104: $\epsilon^{\rm MC}_{\rm XX} \simeq 0.9-1.1$\,meV and the XX radiative 

1105: width $\Gamma^{\rm MC}_{\rm XX} \simeq 0.1$\,meV. The latter value is 

1106: obtained by extrapolating the measured homogeneous width 

1107: ${\tilde \Gamma}^{\rm MC}_{\rm XX} = 

1108: {\tilde \Gamma}^{\rm MC}_{\rm XX}(T)$ to $T = 0$\,K. 

1109: 

1110: The optical experiments with the MC sample were performed using a 

1111: Ti:sapphire laser source which generates Fourier-limited 100\,fs laser 

1112: pulses at 76\,MHz repetition rate. Two exciting pulses, 1 and 2, with 

1113: variable relative delay time $\tau_{12}$ propagate along two different 

1114: incident directions ${\bf p}_{1,2}$ at small angle ($\leq1^{\circ}$) to 

1115: the surface normal. Pulse 1 precedes pulse 2 for $\tau_{12}>0$. The  

1116: reflectivity spectra of the probe light and the FWM signal were analyzed 

1117: with a spectrometer and a charge-coupled device camera of 140\,$\mu$eV 

1118: FWHM resolution. The sample was held in a Helium bath cryostat at T=5\,K 

1119: for all the pump-probe measurements and at T=9\,K in the FWM experiments. 

1120: 

1121: 

1122: \subsection{Bipolariton dephasing in GaAs microcavities}

1123: 

1124: 

1125: \begin{figure}

1126: \includegraphics*[width=8cm]{figure7.eps}

1127: \caption[]{ Spectrally resolved four--wave mixing for co-circular 

1128: (dashed line), co-linear (bold dotted line), and cross-linear (solid 

1129: line) polarizations of the exciting pulses. The microcavity detuning 

1130: is $\delta = 0.76$\,meV. The pulse along ${\bf p}_1$-direction induces 

1131: only 1$\lambda$-LB polaritons, and its spectrum is shown by the dotted 

1132: line. } 

1133: \end{figure}

1134: 

1135: In order to measure the bipolariton dephasing we perform spectrally 

1136: resolved FWM. The FWM signal was detected at $2 {\bf p}_2 - {\bf p}_1$ 

1137: in reflection geometry. The spot size of both exciting beams was $\sim 

1138: 50$\,$\mu$m. In Fig.\,7 we plot the spectrally--resolved FWM signal for 

1139: different polarization configurations of the laser pulses. The positive 

1140: detuning between the cavity 1$\lambda$-eigenmode and HH exciton is 

1141: $\delta = 0.76$\,meV, and the delay time is $\tau_{12}$=1\,ps. Pulse 1 

1142: of about 500\,fs duration was spectrally shaped to {\it excite only the 

1143: 1}$\lambda$-{\it LB polaritons}, and the FWM was probed with the 

1144: spectrally broad pulse 2 at all 1$\lambda$-mode polariton resonances. 

1145: For co--linear and cross-linear polarization configurations, the 

1146: 1$\lambda$-LB polariton to excitonic molecule transition (1$\lambda$-LB 

1147: -- XX) is observed in the FWM signal (see arrow in Fig.\,7) at a spectral 

1148: position consistent with that found in our previous pump-probe 

1149: experiments \cite{BorriPRBRb00}. The XX-mediated FWM signal disappears 

1150: for co--circular polarization, in accordance with the polarization 

1151: selection rules for the two-photon generation of excitonic molecules 

1152: in a GaAs QW. 

1153: 

1154: Although the analysis of FWM in microcavities can be rather complicated 

1155: \cite{Gonokami97,ShiranePRB98}, the interpretation of our measurements 

1156: is simplified by the selective excitation of the 1$\lambda$-LB polaritons 

1157: only. The observed TI--FWM is a free polarization decay, due to the 

1158: dominant homogeneous broadening of the X lines in our high-quality 

1159: 250\,$\AA$-wide QWs \cite{LangbeinPRB2000}. At positive delays the FWM 

1160: signal is created by the following sequence. At first, pulse 1 induces 

1161: a first--order polarization associated with 1$\lambda$-LB polaritons.  

1162: The induced polarization decays with the dephasing time 

1163: $T_2^{\rm 1\lambda-LB}$ of the 1$\lambda$-LB polaritons. The dephasing 

1164: time $T_2^{\rm 1\lambda-LB}$ is dominated by the lifetime of 

1165: 1$\lambda$-mode MC photons. Pulse 2 interacts nonlinearly with the 

1166: induced polarization, and a third--order FWM signal is created with an 

1167: amplitude that decreases with increasing $\tau_{12}$, due to the decay 

1168: of the first--order polarization associated with the 1$\lambda$-LB 

1169: polaritons. The TI--FWM intensities at all probed resonances therefore 

1170: decay nearly with the time constant $T_{2}^{\rm 1\lambda-LB}$/2. At 

1171: negative $\tau_{12}$ the FWM signal stems from the two--photon coherence 

1172: of the crystal ground state to the excitonic molecule transition (0--XX) 

1173: induced by pulse 2. According to energy -- in-plane momentum conservation, 

1174: since pulse 1 is resonant with 1$\lambda$-LB polaritons only, the FWM 

1175: signal, associated with bulk photons, is emitted in the direction 

1176: $2{\bf p}_2 - {\bf p}_1$ with the energy of the 1$\lambda$-LB -- XX 

1177: transition. Thus the TI--FWM dynamics at negative time delays allows us 

1178: to study the polarization decay of the 0--XX transition \cite{BorriOECS02}, 

1179: i.e., to find ${\tilde \Gamma}^{\rm MC}_{\rm XX}$. 

1180: 

1181: \begin{figure}

1182: \includegraphics*[width=8cm]{figure8.eps}

1183: \caption[]{ Comparison between the FWM dynamics measured at the 

1184: 1$\lambda$-LB -- XX transition, when pulse 1 resonantly induces the 

1185: 1$\lambda$-mode lower-branch polaritons only, and at the 1$\lambda$-MB -- 

1186: XX transition, when pulse 1 resonantly excites only the 1$\lambda$-mode 

1187: middle-branch polaritons. Inset: The XX homogeneous linewidth 

1188: ${\tilde \Gamma}^{\rm MC}_{\rm XX}$ against the MC detuning $\delta$, 

1189: measured at $T=9$\,K with about 4\,nJ/cm$^{2}$ pump fluence. }

1190: \end{figure}

1191: 

1192: The $\tau_{12}$-dependence of the TI--FWM signals associated with the 

1193: 1$\lambda$-LB -- XX and 1$\lambda$-MB -- XX transitions is shown in 

1194: Fig.\,8. As expected, at negative $\tau_{12}$ one finds the same 

1195: dynamics for both transitions. Therefore, independently of the 

1196: 1$\lambda$-eigenmode MC polariton branch selectively excited by pulse 1, 

1197: we can infer the polarization decay rate of the 0--XX transition. The 

1198: homogeneous linewidth of the 0--XX transition 

1199: ${\tilde \Gamma}^{\rm MC}_{\rm XX}$ measured at low excitation energies 

1200: per pulse ($\sim4$\,nJ/cm$^{2}$) is potted against the MC detuning 

1201: $\delta$ in the inset of Fig.\,8. Only a weak detuning dependence of 

1202: ${\tilde \Gamma}^{\rm MC}_{\rm XX}$ is observed for the detuning band 

1203: $-2\,\mbox{meV} \leq \delta \leq 2\,\mbox{meV}$. Note that the deduced 

1204: values ${\tilde \Gamma}^{\rm MC}_{\rm XX}(T$=9\,K) $\simeq 0.3-0.4$\,meV 

1205: are by factor $1.5-2$ larger than ${\tilde \Gamma}^{\rm QW}_{\rm XX} 

1206: \simeq 0.2$\,meV measured from the reference QW at nearly the same bath 

1207: temperature T=10\,K \cite{LangbeinPRB2000} (see the dotted line in the 

1208: inset of Fig.\,8). 

1209: 

1210: 

1211: \subsection{The binding energy of bipolaritons in GaAs microcavities}

1212: 

1213: 

1214: The bipolariton energy $E_{\rm XX}^{\rm MC}$ was found by analysing the 

1215: pump-probe experiments. Pulse 1 acts as an intense pump while pulse 2 

1216: is a weak probe. The spectrum of the pump pulse is shaped and tuned in 

1217: order to excite resonantly the 1$\lambda$-LB polaritons only. The 

1218: spectrally broad probe pulse has a spot size of $\sim 40$\,$\mu$m. In 

1219: this case the in-plane spatial gradient of the polariton energy is not 

1220: significant. In order to achieve a uniform pump density over the probe 

1221: area, the cross-section of the pump pulse is chosen to be by factor 

1222: two larger than that of the probe light. 

1223: 

1224: In Ref.\,[\onlinecite{BorriPRBRb00}] we show a well-resolved 

1225: pump-induced absorption at the 1$\lambda$-LB -- XX transition 

1226: in the investigated MC sample. The 1$\lambda$-LB -- XX absorption 

1227: was observed in the reflectivity spectra at positive pump-probe delay 

1228: times and for the cross-circularly ($\sigma^+$- and $\sigma^-$-) 

1229: polarized pump and probe pulses, according to the optical selection 

1230: rules. In particularly, the induced absorption for three different 

1231: positive MC detunings was measured. Here we extend the 

1232: pump-probe experiment to study the detuning dependence 

1233: $E_{\rm XX}^{\rm MC} = E_{\rm XX}^{\rm MC}(\delta)$, including $\delta 

1234: < 0$. In Fig.\,9 the probe reflectivity spectra measured at $\tau_{12} 

1235: \simeq 0.5$\,ps for the cross-circularly polarized pump and probe 

1236: pulses is plotted. Indicated by the arrows (see Fig.\,9), a spectrally 

1237: well-resolved pump-induced absorption resonance is observed. In the 

1238: upper left-hand side (l.h.s.) part of Fig.\,9 the energy position of 

1239: the 1$\lambda$-LB, 1$\lambda$-MB, and 1$\lambda$-UB 

1240: polariton resonances and of the induced 1$\lambda$-LB -- XX absorption 

1241: are plotted against the MC detuning $\delta$. The fit done with a  

1242: three-coupled-oscillator scheme (1$\lambda$-eigenmode MC photon, 

1243: HH exciton, and LH exciton resonances) are shown by the solid lines. The 

1244: energies $E^{\rm HH}_{\rm X}$ and $E^{\rm LH}_{\rm X}$ of the HH and 

1245: LH excitons ($E^{\rm HH}_{\rm X} \simeq 1.5219$\,eV and 

1246: $E^{\rm LH}_{\rm X} \simeq 1.5245$\,eV) are inferred from the fit, and 

1247: the molecule energy $E^{\rm MC}_{\rm XX}$ is determined as the sum of 

1248: the measured 1$\lambda$-LB and 1$\lambda$-LB -- XX transition energies. 

1249: The bipolariton binding energy, evaluated as $\epsilon^{\rm MC}_{\rm XX} 

1250: = 2 E^{\rm HH}_{\rm X} - E^{\rm MC}_{\rm XX}$, is plotted against the MC 

1251: detuning $\delta$ in the lower l.h.s. part of Fig.\,9. We find that 

1252: $\epsilon_{\rm XX}^{\rm MC} \simeq 0.9-1.1$\,meV, i.e., is similar to 

1253: the value of $\epsilon_{\rm XX}^{\rm QW}$ in the reference single QW 

1254: and slightly larger than that previously reported in 

1255: Ref.\,[\onlinecite{BorriPRBRb00}]. 

1256: 

1257: \begin{figure}

1258: \includegraphics*[width=8cm]{figure9.eps}

1259: \caption[]{ The reflectivity spectra of the probe light at different 

1260: detuning values. The spectra are measured for the cross-circularly 

1261: polarized pump and probe pulses at delay time $\tau_{12} \simeq 0.5$\,ps 

1262: and $T=5$\,K. The pump fluence is about 0.1\,$\mu$J/cm$^{2}$. The arrows 

1263: indicate the 1$\lambda$-LP -- XX pump-induced absorption. Upper left 

1264: inset: The measured energy position of the 1$\lambda$-LP, 1$\lambda$-MP, 

1265: 1$\lambda$-UP resonances (filled square points) and of the induced 

1266: 1$\lambda$-LP -- XX absorption (unfilled square points) versus the MC 

1267: detuning $\delta$. The fit of the MC polariton branches, associated 

1268: with LH and HH QW excitons, is shown by the solid lines. Lower left 

1269: inset: The XX binding energy $\epsilon^{\rm MC}_{\rm XX}$ determined 

1270: as the difference between twice the bare HH exciton energy and the 

1271: sum of the 1$\lambda$-LP and 1$\lambda$-LP -- XX transition energies. }

1272: \end{figure}

1273: 

1274: 

1275: \section{Discussion}

1276: 

1277: 

1278: The optical decay of MC bipolaritons can also occur directly, through 

1279: escape of the photon component of the constituent $\sigma^+$- and 

1280: $\sigma^-$-polarized MC polaritons into the bulk photon modes. The XX 

1281: radiative width associated with this channel is given by 

1282: \begin{eqnarray}

1283: \Gamma_{\rm XX}^{\rm MC (2)} ({\bf K}_{\|}\!\!&=&\!\!0) = {\hbar \over 

1284: \pi} \gamma_{\rm R}  \int_0^{\infty} |\Psi_{\rm XX}^{(0)}(2p_{\|})|^2 

1285: \nonumber \\

1286: &\times& \Big[ \sum_i \left[ 1 - (u_i^{\rm MC})^2 \right] \Big] p_{\|} 

1287: dp_{\|} \, ,

1288: \label{MCbulk}

1289: \end{eqnarray}

1290: where $u_i^{\rm MC} = u_i^{\rm MC}(p_{\|})$ are determined by 

1291: Eq.\,(\ref{MCcomp}) and $i$ runs over 0$\lambda$-LB, 1$\lambda$-LB, and 

1292: 1$\lambda$-UB. Equation (\ref{MCbulk}) is akin to Eq.\,(\ref{rada}) and 

1293: can be interpreted in terms of optical evaporation of the MC excitonic 

1294: molecules through the DBR mirrors. Using the measured radiative 

1295: linewidth of 1$\lambda$-LB polaritons, 

1296: $\Gamma_{\rm X}^{\rm MC}(p_{\|}\!\simeq\!0.13\!\times\!10^5\,{\rm cm}^{-1}) 

1297: \simeq 0.1$\,meV, we estimate $\hbar \gamma_{\rm R} \simeq 0.3$\,meV, so 

1298: that the radiative lifetime of MC photons is given by $\tau_{\rm R} 

1299: \simeq 2.4$\,ps. In this case Eq.\,(\ref{MCbulk}) yields 

1300: $\Gamma_{\rm XX}^{\rm MC (2)} \simeq 1 - 2\,\mu$eV for 

1301: $\epsilon_{\rm XX}^{(0)} \simeq 0.9 - 1.1$\,meV and assuming that 

1302: $\gamma_{\rm R}$ is $p_{\|}$-independent. Thus 

1303: $\Gamma_{\rm XX}^{\rm MC (2)}$ is less than 

1304: $\Gamma_{\rm XX}^{\rm QW (2)}$, estimated with Eq.\,(\ref{rada}) for 

1305: the reference QW (see Fig.\,3), by more than one order of magnitude. 

1306: This is because instead of the smallness parameter 

1307: $\delta^{\rm (2D)}_{\rm R} = (a^{\rm (2D)}_{\rm XX} p_0)^2$, which 

1308: appears on the r.h.s. of Eq.\,(\ref{rada}), Eq.\,(\ref{MCbulk}) is scaled 

1309: by $(a^{\rm (2D)}_{\rm XX} p_{\|}^{(1\lambda)})^2 \ll 

1310: \delta^{\rm (2D)}_{\rm R}$. 

1311: 

1312: The radiative width $\Gamma_{\rm XX}^{\rm MC (2)}$, associated with 

1313: the decay of XXs into the bulk photon modes, is by two orders of 

1314: magnitude less than $\Gamma_{\rm XX}^{\rm MC (1)}$ calculated with 

1315: Eqs.\,(\ref{MCrad})-(\ref{Green}). Thus the resonant in-plane 

1316: dissociation of molecules into outgoing MC polaritons absolutely 

1317: dominates in the XX-mediated optics of microcavities, so that the total 

1318: XX radiative width is given by $\Gamma_{\rm XX}^{\rm MC} = 

1319: \Gamma_{\rm XX}^{\rm MC (1)} + \Gamma_{\rm XX}^{\rm MC (2)} \simeq 

1320: \Gamma_{\rm XX}^{\rm MC (1)}$ (see Figs.\,4 and 10). The extremely small 

1321: value of $\Gamma_{\rm XX}^{\rm MC (2)}$ allows us to interpret a MC 

1322: excitonic molecule as a nearly ``optically-dark'' state with respect 

1323: to its direct decay into the bulk photon modes. However it is the 

1324: resonant coupling between 1$\lambda$-mode cavity polaritons and external 

1325: bulk photons which is responsible for the optical generation and probe 

1326: of the XX states in microcavities: Our optical experiments deal only 

1327: with bulk pump, probe, and signal photons. In the meantime the 

1328: bipolariton wavefunction ${\tilde \Psi}_{\rm XX}$ is constructed in 

1329: terms of 0$\lambda$-LB, 1$\lambda$-LB, and 1$\lambda$-UB polariton 

1330: states, and umklapp between the MC polariton branches occurs through 

1331: the coherent Coulombic scattering of two constituent polaritons. 

1332: 

1333: While the interpretation of the experimental data (see Section III) does 

1334: require three-branch, 1$\lambda$-LB, 1$\lambda$-MB, and 1$\lambda$-UB, 

1335: polaritons associated with HH and LH excitons, the contribution to the 

1336: XX optics from the LH Xs is very small. This occurs because (i) the energy 

1337: $E^{\rm LH}_{\rm X}$ is well-separated from the XX-mediated resonance at 

1338: $E^{\rm HH}_{\rm X} - \epsilon^{\rm MC}_{\rm XX}/2$ (the relevant ratio 

1339: between $\epsilon^{\rm MC}_{\rm XX}/2$ and $E^{\rm LH}_{\rm X} - 

1340: E^{\rm HH}_{\rm X} + \epsilon^{\rm MC}_{\rm XX}/2$ is equal to 0.16, 

1341: i.e., is much less than unity) and (ii) because a contribution of the LH 

1342: exciton to the total XX wavefunction is unfavorable in energy, i.e., is 

1343: rather minor. We have checked numerically that by the first argument only 

1344: the LH-X resonance cannot change the XX radiative corrections for more 

1345: than 3-5$\%$. Thus the bipolariton model we develop to analyze the 

1346: optical properties of MC excitonic molecules and to explain the 

1347: experimental data deals only with 0$\lambda$-LB, 1$\lambda$-LB, and 

1348: 1$\lambda$-UB polaritons associated with the ground-state HH exciton. 

1349: 

1350: \begin{figure}

1351: \includegraphics*[width=8cm]{figure10.eps}

1352: \caption[]{ The radiative corrections to the excitonic molecule state, 

1353: $\Gamma^{\rm MC}_{\rm XX}$ and $\Delta^{\rm MC}_{\rm XX}$, calculated 

1354: versus the MC detuning $\delta$ with Eqs.\,(\ref{MCrad})-(\ref{Green}) 

1355: for the MC Rabi energy $\hbar \Omega^{\rm MC}_{1\lambda} = 3.70$\,meV 

1356: and assuming that the DBR optical confinement is completely relaxed for 

1357: $p_{\|} \geq p_{\|}^{(1\lambda)} = 10^5\,\mbox{cm}^{-1}$. In this case  

1358: the 0$\lambda$-LB dispersion is replaced by the interface polariton 

1359: dispersion with the oscillator strength $\hbar^2 R^{\rm QW}_{\rm X} = 

1360: 0.035\,\mbox{eV}^2 \AA$. The input XX binding energy 

1361: $\epsilon^{(0)}_{\rm XX}$ is 0.9\,meV (dash-dotted line), 1.0\,meV 

1362: (solid line), and 1.1\,meV (dashed line). }

1363: \end{figure}

1364: 

1365: In Fig.\,10 we plot the XX radiative corrections against the MC detuning 

1366: $\delta$, calculated with Eqs.\,(\ref{MCrad})-(\ref{Green}) by using the 

1367: MC parameters adapted to our GaAs microcavities. Namely, the 

1368: 1$\lambda$-mode cavity Rabi splitting is given by $\hbar 

1369: \Omega^{\rm MC}_{1\lambda} = 3.7$\,meV, and we assume that the DBR 

1370: optical confinement follows the step function $\Theta(p_{\|}^{(1\lambda)} 

1371: - p_{\|})$. For $p_{\|} \geq p_{\|}^{(1\lambda)}$ the 

1372: 0$\lambda$-LB is replaced by the interface polariton dispersion given by 

1373: Eq.\,(\ref{pol}). Due to the absence of the DBR transverse optical 

1374: confinement at $p_{\|} \geq p_{\|}^{(1\lambda)}$, the resonant optical 

1375: decay of the constituent excitons into the bulk photon mode is also 

1376: included in our calculations by using Eq.\,(\ref{rada}) with integration 

1377: over $dp_{\|}$ from $p_{\|}^{(1\lambda)}$ to $p_0$. From Fig.\,10 we 

1378: conclude that for the detuning band $-2\,\mbox{meV} \lesssim \delta 

1379: \lesssim 2\,\mbox{meV}$ the radiative width $\Gamma^{\rm MC}_{\rm XX}$ 

1380: is about $0.20-0.22$\,meV and indeed weakly depends upon $\delta$, in 

1381: accordance with our experimental data. A few $\mu$eV 

1382: $\Gamma^{\rm MC}_{\rm XX}$-jump, associated with the critical point 

1383: $M_1$, is too small to be detected in the current experiments. Note that 

1384: the contribution to $\Gamma^{\rm MC}_{\rm XX}$ from the decay channel 

1385: ``XX $\rightarrow$ 1$\lambda$-LB polariton + 1$\lambda$-LB polariton'' 

1386: can easily be estimated within a standard perturbation theory: 

1387: $\Gamma^{\rm MC}_{\rm XX \rightarrow 1\lambda LB + 1\lambda LB} \simeq 

1388: (\hbar \Omega^{\rm MC}_{1\lambda})^2 \rho^{\rm XX \rightarrow 1\lambda LB 

1389: + 1 \lambda LB}_{\hbar \omega = \hbar \omega_t - 

1390: \epsilon^{\rm MC}_{\rm XX}/2}$, where the JDPS is given by 

1391: Eq.\,(\ref{JDPS}). The above estimate yields $\Gamma^{\rm MC}_{\rm XX 

1392: \rightarrow 1\lambda LB + 1\lambda LB} \simeq 0.06$\,meV and is 

1393: consistent with the value of the $\Gamma^{\rm MC}_{\rm XX}$-jump around 

1394: $\delta = \delta_2$, i.e., at the $M_2$ critical point (see Fig.\,10). An 

1395: observation of $\Gamma^{\rm MC}_{\rm XX} \simeq 0.10-0.15$\,meV at 

1396: $\delta > \delta_2$, when the MC excitonic molecules become 

1397: optically dark with respect to the decay into 1$\lambda$-mode MC 

1398: polaritons, would be a direct visualization of the hidden decay path 

1399: ``XX $\rightarrow$ interface polariton + interface polariton''. 

1400: 

1401: The {\it relative} change of the XX radiative corrections is rather small 

1402: to be observed in the tested MC detuning band $|\delta| \leq 2$\,meV with 

1403: the current accuracy of our measurements: 

1404: Eqs.\,(\ref{MCrad})-(\ref{Green}) yield 

1405: $\epsilon^{\rm MC}_{\rm XX}(\delta\!\!=\!\!2\,\mbox{meV}) - 

1406: \epsilon^{\rm MC}_{\rm XX}(\delta\!\!=\!\!-2\,\mbox{meV}) \simeq - 

1407: 4\,\mu$eV and $\Gamma^{\rm MC}_{\rm XX}(\delta\!\!=\!\!2\,\mbox{meV}) - 

1408: \Gamma^{\rm MC}_{\rm XX}(\delta\!\!=\!\!-2\,\mbox{meV}) \simeq - 

1409: 5\,\mu$eV; the energy structure at $\delta = \delta_1 = - 

1410: \epsilon^{\rm MC}_{\rm XX}$, i.e., nearby the critical point $M_1$, is 

1411: also of a few $\mu$eV only (see Fig.\,10). On the other hand, the 

1412: GaAs-based microcavities we have now do not allow us to test the critical 

1413: point $M_2$ which is located in the MC detuning band 5\,meV $< \delta <$ 

1414: 8\,meV. In the latter case the relative change of 

1415: $\Delta_{\rm XX}^{\rm MC}$ and $\Gamma_{\rm XX}^{\rm MC}$ is large 

1416: enough, about $0.04-0.07$\,meV, to be detected in our experiments. 

1417: High-precision modulation spectroscopy is very relevant to observation 

1418: of the critical points, because the derivatives $\partial^n 

1419: (\Delta_{\rm XX}^{\rm MC}) / \partial \delta^n$ ($n \geq 1$) and 

1420: $\partial^n (\Gamma_{\rm XX}^{\rm MC}) / \partial \delta^n$ ($n \geq 1$) 

1421: undergo a sharp change in the spectral vicinity of $M_{1,2}$. The 

1422: modulation of $\delta$ can be done by applying time-dependent 

1423: quasi-static electric \cite{Fisher95}, magnetic \cite{Armitage97} or 

1424: pressure \cite{Zhang02} fields. Note that the measurement of the 

1425: detunings $\delta_1$ and $\delta_2$ will allow us to determine with a 

1426: very high accuracy, by using Eqs.\,(\ref{vanHove}), the XX binding 

1427: energy $\epsilon^{\rm MC}_{\rm XX}$ and the MC Rabi frequency 

1428: $\Omega_{1\lambda}^{\rm MC}$. A detailed study of the XX Lamb shift 

1429: $\Delta_{\rm XX}^{\rm MC}$ versus the MC detuning $\delta$ and, in 

1430: particular, the detection of the critical points $M_1$ and $M_2$ are the 

1431: issue of our next experiments. 

1432: 

1433: In order to estimate the radiative width $\Gamma^{\rm MC}_{\rm XX}$ from 

1434: the total homogeneous width ${\tilde \Gamma}^{\rm MC}_{\rm XX}$ measured 

1435: at $T = 9$\,K in our FWM experiment, we assume that apart from the XX 

1436: radiative decay the main contribution to 

1437: ${\tilde \Gamma}^{\rm MC}_{\rm XX}$ is due to temperature-dependent XX 

1438: -- LA-phonon scattering. Note that in the experiment we deal with a 

1439: low-intensity limit, when ${\tilde \Gamma}^{\rm MC}_{\rm XX}$ is nearly 

1440: independent of the excitation level. Thus 

1441: ${\tilde \Gamma}^{\rm MC}_{\rm XX} = \Gamma^{\rm MC}_{\rm XX} + 

1442: \Gamma^{\rm QW}_{\rm XX-LA}$, where $\Gamma^{\rm QW}_{\rm XX-LA}$ is 

1443: due to the scattering of QW molecules by bulk LA-phonons. The DBR optical 

1444: confinement does not influence the XX -- LA phonon scattering, so that 

1445: the width $\Gamma^{\rm QW}_{\rm XX-LA} = \Gamma^{\rm QW}_{\rm XX-LA}(T)$ 

1446: is the same for XXs in the reference single QW and in the microcavities. 

1447: $\Gamma^{\rm QW}_{\rm XX-LA}$ is given by 

1448: \begin{eqnarray}

1449: \Gamma^{\rm QW}_{\rm XX-LA}({\bf K}_{\|}\!=\!0) &=& 2 \pi {\hbar \over 

1450: \tau_{\rm sc}} \int_1^{\infty} d \varepsilon \varepsilon 

1451: \sqrt{ \varepsilon \over \varepsilon - 1 } 

1452: \nonumber \\

1453: &\times&

1454: | F_z(a \sqrt{ \varepsilon (\varepsilon - 1)} |^2 

1455: n^{\rm ph}_{\varepsilon} \, , 

1456: \label{XX-LA}

1457: \end{eqnarray}

1458: where $\tau_{\rm sc} = (\pi^2 \hbar^4 \rho)/(32 D_x^2 M_x^3 v_s)$, $v_s$ 

1459: is the longitudinal sound velocity, $D_x$ is the X deformation potential, 

1460: $\rho$ is the crystal (GaAs) density, $n^{\rm ph}_{\varepsilon} = 

1461: 1/[\exp(\varepsilon E_0/k_{\rm B}T) - 1]$, and $E_0 = 4 M_x v_s^2$. The 

1462: form-factor $F_z(x) = [\sin(x)/x][e^{ix}/(1 - x^2/\pi^2)]$ refers to an 

1463: infinite rectangular QW confinement potential and describes the 

1464: relaxation of the momentum conservation law in the $z$-direction. The 

1465: dimensionless parameter $a$ is given by $a = (2 d_z M_x v_s)/\hbar$. 

1466: 

1467: \begin{figure}

1468: \includegraphics*[width=8cm]{figure11.eps}

1469: \caption[]{ The temperature dependence of the homogeneous width 

1470: $\Gamma^{\rm QW}_{\rm XX-LA}$ associated with scattering of QW excitonic 

1471: molecules by bulk LA-phonons. The calculations are done with 

1472: Eq.\,(\ref{XX-LA}) for the X deformation potential $D_x = 8$\,eV 

1473: (dashed line), 10\,eV (solid line), and 12\,eV (dotted line). }

1474: \end{figure}

1475: 

1476: The values of the deformation potential $D_x$, published in literature, 

1477: disperse in the band $7\,\mbox{eV} \leq D_x \leq 18$\,eV. In Fig.\,11 we 

1478: plot $\Gamma^{\rm QW}_{\rm XX-LA} = \Gamma^{\rm QW}_{\rm XX-LA}(T)$ 

1479: calculated by Eq.\,(\ref{XX-LA}) for $D_x =$ 8, 10, and 12\,eV. The 

1480: deformation potential $D_x = 8$\,eV, which gives 

1481: $\Gamma^{\rm QW}_{\rm XX-LA}(T\!=\!9\,\mbox{K}) \simeq 0.094\,$meV 

1482: and is close to $D_x \simeq 9.6$\,eV reported for GaAs in 

1483: Ref.\,[\onlinecite{Pollak68}], fits the temperature dependence 

1484: $\Gamma^{\rm QW}_{\rm XX-LA} = \Gamma^{\rm QW}_{\rm XX-LA}(T)$ measured 

1485: for the reference QW. In particular, 

1486: $\Gamma^{\rm QW}_{\rm XX-LA}(T\!=\!10\,\mbox{K}) \simeq 0.1$\,meV is 

1487: inferred from the total ${\tilde \Gamma}^{\rm QW}_{\rm XX} \simeq 

1488: 0.2\,$meV (see the inset of Fig.\,8). Thus from our FWM measurements of 

1489: ${\tilde \Gamma}^{\rm MC}_{\rm XX}$ at $T=9$\,K we conclude that the XX 

1490: radiative width $\Gamma^{\rm MC}_{\rm XX} = 

1491: {\tilde \Gamma}^{\rm MC}_{\rm XX} - \Gamma^{\rm QW}_{\rm XX-LA}$ is about 

1492: $0.2-0.3$\,meV, i.e., is consistent with the values calculated within 

1493: the bipolariton model (see Fig.\,10). 

1494: 

1495: In order to apply the bipolariton model [see Eq.\,(\ref{BPint})] to 

1496: excitonic molecules in the reference single QW, one should take into 

1497: account that the reference QW is sandwiched between a thick substrate 

1498: and a cap layer of the thickness $L_{\rm cap} \simeq 499$\,nm. The 

1499: evanescent light field associated with the QW polaritons is modified by 

1500: the cap layer. Indeed, for the $- \epsilon_{\rm XX}^{(0)}/2$ energy 

1501: detuning from the X resonance, one estimates that $\kappa \simeq 1.4 

1502: \times 10^4\,\rm{cm}^{-1}$ so that $\exp(-\kappa L_{\rm cap}) \simeq 

1503: 0.5$ is not negligible. The estimate refers to two frequency-degenerate 

1504: outgoing interface polaritons ($\hbar \omega = E_{\rm X} - 

1505: \epsilon_{\rm XX}^{(0)}0/2$) created in the photon-assisted resonant 

1506: dissociation of the QW molecule with ${\bf K}_{\|}=0$. At $z=L_{\rm cap}$ 

1507: the initial evanescent field splits into two evanescent fields, 

1508: ``transmitted'' to air (or vacuum) and ``reflected'' back towards the QW. 

1509: The first light field very effectively decays in the $z$-direction, with 

1510: $\kappa_{\rm air} = \sqrt{ p^2_{\|} - (\omega/c)^2} \simeq 2.6 \times 

1511: 10^5\,\rm{cm}^{-1} \gg \kappa$. The ``reflected'' evanecsent light field 

1512: makes at $z$=0 a destructive superposition with the initial evanecsent 

1513: field, because the reflection coefficient of the top surface of the 

1514: cap layer is $r_{\rm cap} = (\kappa - \kappa_{\rm air})/(\kappa + 

1515: \kappa_{\rm air}) \simeq -0.9$. The destructive superposition stems from 

1516: the $\pi$-jump of the phase of the ``reflected'' evanescsent field. Thus 

1517: the effective oscillator strength relevant to the QW bipolariton wave 

1518: Eq.\,(\ref{BPint}) is given by ${\tilde R}_{\rm X}^{\rm QW} = 

1519: R_{\rm X}^{\rm QW}[1 + (r_{\rm cap}/2) \exp(- 2 \kappa L_{\rm cap})]^2$. 

1520: For our reference structure with $\hbar^2 R_{\rm X}^{\rm QW} \simeq  

1521: 0.035\,{\rm eV}^2\AA$ we estimate $\hbar^2 {\tilde R}_{\rm X}^{\rm QW} 

1522: \simeq 0.028\,{\rm eV}^2\AA$. In this case 

1523: Eqs.\,(\ref{BPint})-(\ref{rada}) yield the total radiative width 

1524: $\Gamma_{\rm XX}^{\rm QW}({\bf K}_{\|}$=0) $\simeq 0.126\,{\rm meV}$ 

1525: (see Fig.\,3), the value which is very close to $\Gamma_{\rm XX}^{\rm QW} 

1526: \simeq 0.1\,{\rm meV}$ obtained from the experimental data. 

1527: 

1528: Thus the bipolariton model, which attributes the XX radiative corrections 

1529: mainly to the in-plane dissociation of molecules into outgoing 

1530: interface/MC polaritons, reproduce quantitatively the XX radiative 

1531: widths $\Gamma^{\rm MC}_{\rm XX}$ and $\Gamma^{\rm QW}_{\rm XX}$ 

1532: estimated from the experimental data. The two main channels for the XX 

1533: decay in microcavities, ``XX $\rightarrow$ 1$\lambda$-LB polariton + 

1534: 1$\lambda$-LB polariton'' and ``XX $\rightarrow$ 0$\lambda$-LB (or 

1535: interface) polariton + 0$\lambda$-LB (or interface) polariton'' in 

1536: comparison with the one leading decay route in single QWs, 

1537: ``XX $\rightarrow$ interface polariton + interface polariton'', explain 

1538: qualitatively the factor two difference between 

1539: $\Gamma^{\rm MC}_{\rm XX}$ and $\Gamma^{\rm QW}_{\rm XX}$. The 

1540: XX-mediated optics of microcavities does require to include the 

1541: ``hidden'' 0$\lambda$-cavity (or interface, if the transverse optical 

1542: confinement is relaxated for large ${\bf p}_{\|}$) polariton mode, which 

1543: is invisible in standard optical experiments and, therefore, is usually 

1544: neglected. Furthermore, with decreasing temperature $T \lesssim 10$\,K 

1545: ${\tilde \Gamma}^{\rm MC}_{\rm XX}$ and 

1546: ${\tilde \Gamma}^{\rm QW}_{\rm XX}$ effectively approach 

1547: $\Gamma^{\rm MC}_{\rm XX}$ and $\Gamma^{\rm QW}_{\rm XX}$, respectively, 

1548: so that the dephasing of the two-photon XX polarization in the 

1549: microcavities and the reference QW occurs mainly through the optical 

1550: decay of the molecules. Thus the $T_2=2T_1$ limit holds for the 

1551: XX-mediated optics in our high-quality nanostructures and justifies the 

1552: bipolariton model. The latter interprets the XX optical response in terms 

1553: of resonant polariton-polariton scattering and requires nonperturbative 

1554: treatment of both leading interactions, exciton-exciton Coulombic 

1555: attraction and exciton-photon resonant coupling. Note that in our 

1556: calculations with the exactly solvable bipolariton model only two 

1557: control parameters of the theory, the input XX binding energy 

1558: $\epsilon_{\rm XX}^{(0)}$ and the MC Rabi frequency 

1559: $\Omega^{\rm MC}_{\rm 1\lambda}$ (or the X oscillator strength 

1560: $R^{\rm QW}_{\rm X}$ for the reference QW), are taken from the 

1561: experimental data. No fitting parameters are used in the numerical 

1562: simulations. 

1563: 

1564: The relative motion of two optically-dressed constituent excitons of the 

1565: bipolariton eigenstate (i.e., of the excitonic molecule) is affected by 

1566: the exciton-photon interaction, according to the polariton dispersion 

1567: law. The optically-induced change of the X energy occurs not only in 

1568: the close vicinity of the resonant crossover between the initial photon 

1569: and exciton dispersions, but in a rather broad band of ${\bf p}$ 

1570: (or ${\bf p}_{\|}$). For example, in bulk semiconductors the effective 

1571: mass associated with the upper polariton dispersion branch at $p=0$ is 

1572: given by 

1573: \begin{equation}

1574: M^{\rm (3D)}_{\rm UB} \simeq { M_x \over 1 + 2 (\omega_{\ell t} / 

1575: \omega_t) [(M_x c^2/\varepsilon_b)/(\hbar \omega_t)] } \, .  

1576: \label{Mass3D}

1577: \end{equation}

1578: For bulk GaAs Eq.\,(\ref{Mass3D}) yields $M_{\rm eff} = 

1579: M^{\rm (3D)}_{\rm UB}$ nearly by factor four less than the translational 

1580: mass relevant to the pure excitonic dispersion, $M_x \simeq 0.7\,m_0$. 

1581: From the microcavity dispersion Eq.\,(\ref{MC}) one estimates for $p_{\|} 

1582: \rightarrow 0$ the effective masses associated with the 

1583: 1$\lambda$-eigenmode polariton dispersion branches: 

1584: \begin{equation}

1585: M^{\rm (MC)}_{\rm 1\lambda UB/LB}(\delta) \simeq { 2 E_{\rm X} \over 

1586: (c^2/\varepsilon_b)[1 \pm \delta / (\hbar 

1587: \Omega_{\rm 1\lambda}^{\rm MC})] } \, , 

1588: \label{MassMC}

1589: \end{equation}

1590: where we assume that $|\delta| \lesssim \hbar 

1591: \Omega_{\rm 1\lambda}^{\rm MC}$. In particular, for a zero-detuning 

1592: GaAs-based microcavity Eq.\,(\ref{MassMC}) yields 

1593: $M^{\rm (MC)}_{\rm 1\lambda UB}(\delta$=$0) = 

1594: M^{\rm (MC)}_{\rm 1\lambda LB}(\delta$=$0) 

1595: = 2E_{\rm X}/(c^2/\varepsilon_b) \simeq 0.7 \times 10^{-4}\,m_0$. In the 

1596: meantime, at relatively large in-plane momenta $p_{\|} \sim 

1597: p^{(1\lambda)}_{\|} < p_0$ the $1\lambda$-LB polariton energy smoothly 

1598: approaches the exciton dispersion, i.e., $[E_{\rm X}(p_{\|}) - \hbar 

1599: \omega^{\rm MC}_{\rm 1\lambda LB}(p_{\|})] \big|_{p_{\|} \sim 

1600: p_{\|}^{(1\lambda)}} \rightarrow [\hbar 

1601: (\Omega^{\rm MC}_{\rm 1\lambda})^2 \omega_t]/[2 

1602: (c^2 p_{\|}^2/\varepsilon_b)] \propto 1/p_{\|}^2$, according to 

1603: Eq.\,(\ref{MC}). While the above difference is rather small in absolute 

1604: energy units, being compared with the in-plane kinetic energy of the 

1605: exciton, $E_{\rm kin}^{\rm X} = \hbar^2 p_{\|}^2/2M_x$, it cannot be 

1606: neglected. For example, the difference $E_{\rm X} - \hbar 

1607: \omega^{\rm MC}_{\rm 1\lambda LB}$ becomes equal to $E_{\rm kin}^{\rm X}$ 

1608: at $p_{\|} \simeq 1.35 \times 10^5\,\mbox{cm}^{-1}$. Note that for the 

1609: above value of the in-plane wavevector $p_{\|}$ the photon component, 

1610: associated with 1$\lambda$-LB polaritons, is negligible, i.e., 

1611: $(u^{\rm MC}_{\rm 1\lambda LB})^2 \simeq 1 \gg 

1612: (v^{\rm MC}_{\rm 1\lambda LB})^2$. Because it is a balance between the 

1613: positive kinetic and negative interaction energies of the constituent 

1614: excitons that gives rise to an excitonic molecule, the described 

1615: optically-induced changes of the X effective mass at $p_{\|}=0$ and the 

1616: nonparabolicity of the X dispersion at large $p_{\|}$ are responsible for 

1617: the large XX radiative corrections in quasi-2D GaAs nanostructures.

1618: 

1619: 

1620: \section{Conclusions}

1621: 

1622: 

1623: In this paper we have studied, both theoretically and experimentally, 

1624: the optical properties of QW excitonic molecules in semiconductor 

1625: (GaAs) microcavities. We attribute the main channel of the XX optical 

1626: decay to the resonant dissociation of MC molecules into outgoing MC 

1627: polaritons, so that the XX-mediated optical signal we detect is due to 

1628: the resonant radiative escape of the secondary MC polaritons through the 

1629: DBRs. The bipolariton model has been adapted to construct the XX 

1630: wavefunction ${\tilde \Psi}_{\rm XX}$ in terms of two (1$\lambda$-UB, 

1631: 1$\lambda$-LB and 0$\lambda$-LB) MC polaritons quasi-bound via Coulombic 

1632: attraction of their exciton components. The MC bipolariton wave 

1633: equation gives the radiative corrections to the XX state in 

1634: microcavities. The following conclusions summarize our results. 

1635: 

1636: (i) The radiative corrections to the excitonic molecule state in 

1637: GaAs-based microcavities, the XX Lamb shift $\Delta^{\rm MC}_{\rm XX}$ 

1638: and the XX radiative width $\Gamma^{\rm MC}_{\rm XX}$, are large (about 

1639: $0.15-0.30$ of the XX binding energy $\epsilon_{\rm XX}^{(0)}$) and 

1640: definitely cannot be neglected. 

1641: 

1642: (ii) While usually the QW exciton -- mediated optics of semiconductor 

1643: microcavities is formulated in terms of two 1$\lambda$-mode polariton 

1644: dispersion branches only (1$\lambda$-UB and 1$\lambda$-LB, according to 

1645: the terminology used in our paper), we emphasize the importance of the 

1646: 0$\lambda$-mode lower-branch polariton dispersion: The Coulombic 

1647: interaction of the constituent excitons, which is responsible for the 

1648: XX state, does couple intrinsically three relevant MC polariton branches, 

1649: (1$\lambda$-UB, 1$\lambda$-LB, and 0$\lambda$-LB). Furthermore, the XX 

1650: decay path ``XX $\rightarrow$ 0$\lambda$-LB polariton + 0$\lambda$-LB 

1651: polariton'' is comparable in efficiency with the optical decay into 

1652: 1$\lambda$-LB polariton modes, i.e., ``XX $\rightarrow$ 1$\lambda$-LB 

1653: polariton + 1$\lambda$-LB polariton''. Due do the relaxation of the DBR 

1654: optical confinement for in-plane wavevectors $p_{\|} \sim p_0 = \omega_t 

1655: \sqrt{\varepsilon_b}/c$, with increasing  $p_{\|}$ the 0$\lambda$-LB 

1656: evolves towards the interface polariton dispersion associated with QW 

1657: excitons. However, the short-wavelength LB polaritons with $p_{\|} \sim 

1658: p_0$ always contribute to the  XX-mediated optics of microcavities. 

1659: 

1660: (iii) The zero-temperature extrapolation of the experimentally found XX 

1661: dephasing width ${\tilde \Gamma}_{\rm XX}^{\rm MC}(T$=9\,K) yields 

1662: $\Gamma_{\rm XX}^{\rm MC}(T$=0\,K) $\simeq 0.2-0.3$\,meV and is in a 

1663: quantitative agreement with the result of the exactly solvable 

1664: bipolariton model, $\Gamma_{\rm XX}^{\rm MC} \simeq 0.20-0.22$\,meV. 

1665: From the analysis of the experimental data we conclude that the 

1666: bipolariton model of MC excitonic molecules, which requires $T_2 \simeq 

1667: 2T_1$ limit, is valid for our high-quality GaAs-based nanostructures 

1668: at $T \lesssim 10$\,K. For the reference GaAs QW without the DBR 

1669: transverse optical confinement we find $\Gamma_{\rm XX}^{\rm QW} = 

1670: {\tilde \Gamma}_{\rm XX}^{\rm QW}(T$=0\,K) $\simeq 0.1$\,meV. The latter 

1671: value is also quantitatively consistent with that calculated by solving 

1672: the QW bipolariton wave equation, $\Gamma_{\rm XX}^{\rm QW} = 0.126$\,meV. 

1673: The nearly factor two difference between $\Gamma_{\rm XX}^{\rm QW}$ and 

1674: $\Gamma_{\rm XX}^{\rm MC}$ clearly demonstrates the existence of the 

1675: additional decay channel for a quasi-2D excitonic molecule in 

1676: microcavities [``XX $\rightarrow$ interface polariton + interface 

1677: polariton'' in MC-free single QWs versus ``XX $\rightarrow$ 0$\lambda$-LB 

1678: (or interface) polariton + 0$\lambda$-LB (or interface) polariton'' and 

1679: ``XX $\rightarrow$ 1$\lambda$-LB polariton + 1$\lambda$-LB polariton'' 

1680: for MC-embedded QW molecules]. 

1681: 

1682: (iv) The critical van Hove points, $M_1(\delta\!=\!\delta_1)$ and 

1683: $M_2(\delta\!=\!\delta_2)$, in the JDPS of the resonant optical channel 

1684: ``XX (${\bf K}_{\|}$=0) $\leftrightarrow$ two 1$\lambda$-mode 

1685: MC polaritons'' can allow us to find accurately the molecule binding 

1686: energy $\epsilon^{\rm MC}_{\rm XX}$ and the MC Rabi frequency 

1687: $\Omega_{\rm 1\lambda}^{\rm MC}$. Thus, by using time-dependent MC 

1688: detuning $\delta = \delta(t)$, we propose to develop high-precision 

1689: modulation spectroscopies in order to detect the rapid changes of the 

1690: XX radiative corrections at $\delta = \delta_{1,2}$ [spikes in the XX 

1691: Lamb shift $\Delta^{\rm MC}_{\rm XX} = \Delta^{\rm MC}_{\rm XX}(\delta 

1692: = \delta_{1,2})$ and jumps in the XX radiatve width 

1693: $\Gamma^{\rm MC}_{\rm XX} = \Gamma^{\rm MC}_{\rm XX}(\delta = 

1694: \delta_{1,2})$] and estimate $\epsilon^{\rm MC}_{\rm XX}$ and 

1695: $\Omega_{\rm 1\lambda}^{\rm MC}$. 

1696: 

1697: 

1698: \section{Acknowledgments}

1699: 

1700: 

1701: We appreciate valuable discussions with J.~R. Jensen and J.~M. Hvam. 

1702: Support of this work by the DFG, EPSRC and EU RTN Project 

1703: HPRN-CT-2002-00298 is gratefully acknowledged. 

1704: 

1705: 

1706: %

1707: %\bibliographystyle{PRBSTY}

1708: %\bibliography{user,eiib}

1709: \begin{thebibliography}{88}

1710: 

1711: 

1712: \bibitem{Agranovich66} 

1713: V.~M. Agranovich and O.~A. Dubovskii, 

1714: Pis'ma Zh. Teor. Fiz. {\bf 3}, 345 (1966) [JETP Lett. {\bf 3}, 223 

1715: (1966)]. 

1716: 

1717: \bibitem{Nakayama85} 

1718: M. Nakayama, Solid State Commun. {\bf 55}, 1053 (1985). 

1719: 

1720: \bibitem{Andreani90} 

1721: L.~C. Andreani and F. Bassani, Phys. Rev. B {\bf 41}, 7536 (1990).  

1722: 

1723: \bibitem{Weisbuch92}

1724: C. Weisbuch, M. Nishioka, A. Ishikawa, and Y. Arakawa, 

1725: Phys. Rev. Lett. {\bf 69}, 3314 (1992). 

1726: 

1727: \bibitem{Savona94} V. Savona, Z. Hradil, A. Quattropani, and 

1728: P. Schwendimann, Phys. Rev. B {\bf 49}, 8774 (1994). 

1729: 

1730: \bibitem{SkolnickSST98}

1731: M.~S. Skolnick, T.~A. Fisher, and D.~M. Whittaker, 

1732: Semicond. Sci. Technol. {\bf 13},  645  (1998).

1733: 

1734: \bibitem{KhitrovaRMP99}

1735: G. Khitrova, H.~M. Gibbs, F. Jahnke, M. Kira, and S.~W. Koch, 

1736: Rev. Mod. Phys. {\bf 71}, 1591 (1999).

1737: 

1738: \bibitem{Gonokami97}

1739: M. Kuwata-Gonokami, S. Inouye, H. Suzuura, M. Shirane, R. Shimano, 

1740: T. Someya, and H. Sakaki, Phys. Rev. Lett. {\bf 79}, 1341 (1997).

1741: 

1742: \bibitem{Fan98}

1743: X. Fan, H. Wang, H.~Q. Hou, and B.~E. Hammons, 

1744: Phys. Rev. B {\bf 57}, R9451 (1998).

1745: 

1746: \bibitem{BorriPRBRb00}

1747: P. Borri, W. Langbein, U. Woggon, J.~R. Jensen, and J.~M. Hvam, 

1748: Phys. Rev. B {\bf 62}, R7763 (2000).

1749: 

1750: \bibitem{Jacob00}

1751: J.~R. Jensen, P. Borri, W. Langbein, and J.~M. Hvam, 

1752: Appl. Phys. Lett. {\bf 76}, 3262 (2000).

1753: 

1754: \bibitem{Saba00} M. Saba, F. Quochi, C. Ciuti, U. Oesterle, J.~L. 

1755: Staehli, B. Deveaud, G. Bongiovanni, and A. Mura, 

1756: Phys. Rev. Lett. {\bf 85}, 385 (2000). 

1757: 

1758: \bibitem{Baars01} 

1759: T. Baars, G. Dasbach, M. Bayer, and A. Forchel,

1760: Phys. Rev. B {\bf 63}, 165311 (2001).

1761: 

1762: \bibitem{Tartakovskii02} A.~I. Tartakovskii, D.~B. Krizhanovskii, 

1763: D.~A. Kurysh, V.~D. Kulakovskii, M.~S. Skolnick, and J.~S. Roberts, 

1764: phys. stat. sol. (a) {\bf 190}, 321 (2002). 

1765: 

1766: \bibitem{Neukirch00}

1767: U. Neukirch, S.~R. Bolton, N.~A. Fromer, L.~J. Sham, and D.~S. Chemla, 

1768: Phys. Rev. Lett. {\bf 84}, 2215 (2000).

1769: 

1770: \bibitem{LaRoccaJOSAB98}

1771: G.~C. La Rocca, F. Bassani, and V.~M. Agranovich, J. Opt. Soc. Am. B 

1772: {\bf 15}, 652 (1998).

1773: 

1774: \bibitem{SiehEPJ99}

1775: C. Sieh, T. Meier, A. Knorr, F. Jahnke, P. Thomas, and S.~W. Koch, 

1776: Eur. Phys. J. B {\bf 11}, 407 (1999).

1777: 

1778: \bibitem{Ivanov95}

1779: A.~L. Ivanov and H. Haug, Phys. Rev. Lett. {\bf 74}, 438 (1995).

1780: 

1781: \bibitem{Ivanov98}

1782: A.~L. Ivanov, H. Haug, and L.~V. Keldysh, 

1783: Phys. Reports {\bf 296}, 237 (1998).

1784: 

1785: \bibitem{Chemla79} D.~S. Chemla, A. Maruani, and E. Batifol,

1786: Phys. Rev. Lett. {\bf 42}, 1075 (1979).

1787: 

1788: \bibitem{Akiyama90}

1789: H. Akiyama, T. Kuga, M. Matsuoka, and M. Kuwata-Gonokami, Phys. Rev. B

1790: {\bf  42}, 5621 (1990).

1791: 

1792: \bibitem{Tokunaga99} E. Tokunaga, A.~L. Ivanov, S.~V. Nair, and Y. Masumoto,

1793: Phys. Rev. B {\bf 59}, R7837 (1999) and {\bf 63}, 233203 (2001). 

1794: 

1795: \bibitem{Mann01} 

1796: Ch. Mann, W. Langbein, U. Woggon, and A.~L. Ivanov, 

1797: Phys. Rev. B {\bf 64}, 235206 (2001).

1798: 

1799: \bibitem{Ivanov97}

1800: A.~L. Ivanov, H. Wang, J. Shah, T.~C. Damen, H. Haug, L.~N. Pfeiffer, and 

1801: L.~V. Keldysh, Phys. Rev. B {\bf 56}, 3941 (1997). 

1802: 

1803: \bibitem{Ulbrich82} C. Weisbuch and R.~G. Ulbrich, in {\it Light

1804: Scattering in Solids III}, Vol.~51 of {\it Topics in Applied

1805: Physics}, M. Cardona and G. G\"{u}ntherodt (Eds.), Springer, Berlin

1806: 1982, p.\,218.

1807: 

1808: \bibitem{LangbeinPRB2000}

1809: W. Langbein and J.~M. Hvam, Phys. Rev. B {\bf 61}, 1692 (2000). 

1810: 

1811: \bibitem{BorriPRB2000}

1812: P. Borri, J.~R. Jensen, W. Langbein, and J.~M. Hvam, 

1813: Phys. Rev. B {\bf 61}, R13377 (2000).

1814: 

1815: \bibitem{vanHove} L. Van Hove, Phys. Rev. {\bf 89}, 1189 (1953). 

1816: 

1817: \bibitem{ShiranePRB98} 

1818: M. Shirane, C. Ramkumar, Y.~P. Svirko, H. Suzuura, S. Inouye, R. Shimano, 

1819: T. Someya, H. Sakaki, and M. Kuwata-Gonokami, Phys. Rev. B {\bf 58}, 7978 

1820: (1998).

1821: 

1822: \bibitem{BorriOECS02}

1823: P. Borri, W. Langbein, U. Woggon, J.~R. Jensen, and J.~M. Hvam, phys. stat.

1824: sol. (a) {\bf 190}, 383 (2002).

1825: 

1826: \bibitem{Fisher95} T.~A. Fisher, A.~M. Afshar, D.~M. Whittaker, 

1827: M.~S. Skolnick, J.~S. Roberts, G. Hill, and M.~A. Pate, 

1828: Phys. Rev. B  {\bf 51}, 2600 (1995).

1829: 

1830: \bibitem{Armitage97} A. Armitage, T.~A. Fisher, M.~S. Skolnick, 

1831: D.~M. Whittaker, P. Kinsler, and J.~S. Roberts, 

1832: Phys. Rev. B  {\bf 55}, 16395 (1997).

1833: 

1834: \bibitem{Zhang02} J. Zhang, H. Zhang, J. Chen, Y. Deng, 

1835: Ch. Hu, L. An, F. Yang, G.-H. Li, and H. Zheng, J. Phys.: Condens. 

1836: Matter {\bf 14}, 5349 (2002). 

1837: 

1838: \bibitem{Pollak68} F.~H. Pollak and M. Cardona, Phys. Rev. 

1839: {\bf 172}, 816 (1968). 

1840: 

1841: 

1842: 

1843: 

1844: 

1845: \end{thebibliography}

1846: 

1847: 

1848: 

1849: 

1850: 

1851: 

1852: \end{document} 

1853: