1: \documentstyle[epsfig,prl,aps,twocolumn]{revtex}

2: \newcommand{\be}{\begin{equation}}

3: \newcommand{\ee}{\end{equation}}

4: \newcommand{\bea}{\begin{eqnarray}}

5: \newcommand{\eea}{\end{eqnarray}}

6: \newcommand{\ba}{\begin{array}}

7: \newcommand{\ea}{\end{array}}

8: \newcommand{\nn}{\nonumber}

9: \newcommand{\FF}{{\cal F}_{P}}

10: \newcommand{\SS}{{\cal S}_{\vec{k} \vec{k}^{'}}}

11: \newcommand{\PP}{{\cal P}}

12: \newcommand{\N}{{\cal N}}

13: \newcommand{\LG}{{\cal L}}

14: \newcommand{\Th}{\Theta}

15: \newcommand{\al}{\alpha}

16: \newcommand{\noi}{\noindent}

17: \newcommand{\eps}{\epsilon}

18: \newcommand{\Lam}{\Lambda}

19: \begin{document}

20: \draft

21: \title{\bf Random Scattering by Atomic Density Fluctuations in Optical 

22: Lattices}

23: 

24: 

25: \author{M. Blaauboer,$^{\rm a,b}$  

26: G. Kurizki, $^{\rm a}$  V.M. Akulin,$^{\rm c}$} 

27: 

28: \address{

29: $^a$ Chemical Physics Department, Weizmann Institute of Science, 

30: Rehovot 76100, Israel\\

31: $^b$ Department of Interdisciplinary Studies, Faculty of Engineering, 

32: Tel Aviv University, Tel Aviv 69978, Israel\\

33: $^c$ Laboratoire Aim\'{e} Cotton, CNRS II, B\^{a}timent 505, 

34: Orsay Cedex 91405 France}

35: \date{\today}

36: \maketitle

37: 

38: \begin{abstract} 

39: We investigate hitherto unexplored regimes of probe scattering 

40: by atoms trapped in optical lattices: weak scattering 

41: by effectively random atomic density distributions and 

42: multiple scattering by arbitrary atomic distributions. 

43: Both regimes are predicted to exhibit a universal 

44: semicircular scattering lineshape for large density fluctuations, 

45: which depend on temperature and quantum statistics. 

46: \end{abstract}

47: 

48: \pacs{PACS numbers: 32.80.Pj, 05.40.-a, 42.50.Lc, 71.23.-k}

49: 

50: 

51: Recent advances in trapping and manipulation of cold atoms 

52: interacting with external fields have primarily been implemented 

53: thus far either in {\it single-atom} systems, such 

54: as sparse (low-density) optical lattices\cite{jess96}, or in Bose-Einstein 

55: condensates with {\it macroscopic} numbers of atoms\cite{park98}.
 

56: Between these two limiting regimes lies the scarcely investigated 

57: domain of
 processes involving a 

58: {\it finite number of interacting atoms}. 

59: Both the potential interest and the difficulties involved in studying
 

60: such processes are evident in the example of optical 

61: lattices with appreciable atomic 

62: filling factors\cite{depu99}, in
 

63: which the transition between "insulating" (localized) and "metallic" 

64: (superfluid) phases have been studied in the framework of the 

65: Bose-Hubbard model\cite{jaks97}. As confirmed by the above study, 

66: the mean-field approximation is inadequate for small numbers of 

67: interacting atoms in lattices due to the presence of large 

68: quantum fluctuations. Still more difficult is the analysis of fluctuations 

69: in systems of cold atoms coupled by long-range ($1/r$ or $1/r^3$) 

70: field-induced interactions\cite{odel99}. This leads to the intruiging 

71: question: Is there a way to circumvent the formidable 

72: task of treating the full dynamics of such systems and still infer 

73: their important characteristics, e.g., their dependence on temperature, 

74: quantum statistics (Bose or Fermi), number of atoms and lattice 

75: parameters? And: Are there universal measurable features 

76: which can be a "signature" of the statistical ensemble 

77: (distribution function) of such systems?

78: 

79: Here we consider the possibility of inferring such statistical 

80: characteristics from the spectral features of probe photons or 

81: particles that are scattered by the density fluctuations of trapped atoms, 

82: notably in optical lattices, in two hitherto unexplored scenarios: (a) The 

83: probe is weakly (perturbatively) scattered 

84: by the local atomic density corresponding to the random occupancy 

85: of different lattice sites (Fig.~\ref{fig:Dos1D} - inset a). (b) The probe 

86: is multiply scattered by an arbitrary (possibly unknown a 

87: priori) multi-atom distribution in the lattice (Fig.~\ref{fig:Dos1D} - inset b). 

88: 

89: At the heart of our analysis is the idea 

90: that the Green function of the scattered photon or particle, which 

91: embodies the relevant spectral information, can be qualitatively 

92: estimated without resorting to cumbersome perturbative calculations 

93: of the probe-multiatom interaction by replacing this 

94: interaction Hamiltonian by an 

95: equivalent random matrix. The random matrix approach, which has been 

96: successfully applied to various disordered systems\cite{meht91}, 

97: allows the evaluation of probe spectra
 

98: to {\it all orders of scattering}, expressing them by means 

99: of {\it only the first two moments} (the mean and variance) 

100: of the random interaction, averaged over the statistical 

101: ensemble of the multiatom system. The highlight of our analysis, 

102: based on this random matrix approach, 

103: is the prediction of a semicircular spectral lineshape 

104: of the probe scattering in the large-fluctuation limit of 

105: trapped atomic ensembles. Thus far, the only known case of quasi-semicircular 

106: lineshapes in optical scattering has been predicted\cite{akul93} 

107: and experimentally verified\cite{ngo94} in dielectric microspheres with 

108: randomly distributed internal scatterers. 

109: 

110: The Green function of the probe (P) at energy $\eps = \hbar \omega$ 

111: is given by 

112: \be 

113: G_{P}(\eps) = \mbox{\rm Tr}_{S} \left[ \frac{1}{\eps-\hat{H}_{P} - 

114: \hat{V}}\, \hat{\rho}_{S} \right], 

115: \label{eq:Green}

116: \ee

117: where $\hat{H}_{P}$, $\hat{V}$ and $\hat{\rho}_{S}$ are, respectively,
 

118: the unperturbed probe Hamiltonian, the probe-system interaction  

119: Hamiltonian and the density operator for the ensemble of the multiatom system (S). 

120: We shall assume that the following conditions hold. (i)  

121: There is no appreciable back-effect of the probe on  

122: the multiatom system (otherwise it is no longer a probe). 

123: (ii) The state of the  

124: multiatom system does not change during the interaction time  

125: with the probe, i.e., the multiatom system remains "frozen", as

126: is applicable for optical or atomic probing. 

127: This situation then cannot be described as Markovian relaxation  

128: (exponential decay) of the probe state into the multiatom  

129: reservoir, since the 

130: correlation time of this reservoir is now much longer  

131: than that of the probe, in contrast with the basic assumption
 

132: of relaxation. (iii) The probe spectrum is broadband, i.e., it 

133: encompasses many of its eigenstates.

134: 

135: For an ensemble "frozen" during the interaction time, the tracing 

136: in (\ref{eq:Green}) implies statistical averaging over repeated 

137: realizations of the multiatom system, every time the probe scattering 

138: is recorded, or taking the expectation value with respect 

139: to the quantum state of the system. 

140: For simplicity, let us explicitly consider elastic scattering (the extension
 

141: to inelastic scattering is straightforward), for which
 

142: \begin{mathletters} 

143: \bea 

144: \hat{V} & = & \sum_{\vec{k}} f_{\vec{k}}\, \hat{\rho}^{\dagger}_{P \vec{k}}\, 

145: \hat{\rho}_{S \vec{k}} + \mbox{\rm h.c.} \hspace*{1cm} \mbox{\rm or} 

146: \label{eq:bilinear} \\  

147: \hat{V} & = & \sum_{\vec{k}} f_{\vec{k}}\, a^{\dagger}_{\vec{k}}\, 

148: \hat{\rho}_{S \vec{k}} + \mbox{\rm h.c.} 

149: \label{eq:linear} 

150: \eea 

151: \label{eq:potential} 

152: \end{mathletters} 

153: Here $f_{\vec{k}}$ is the scattering amplitude for momentum exchange  

154: $\hbar \vec{k}$ between the probe and the system and the $\vec{k}$-mode  

155: Fourier components of the probe (system) density operators  

156: $\hat{\rho}_{P \vec{k}}$ ($\hat{\rho}_{S \vec{k}}$) are defined 

157: in terms of their respective creation and annihilation operators 

158: $\hat{\rho}_{P \vec{k}} = \sum_{\vec{q}} a^{\dagger}_{\vec{q}} 

159: a_{\vec{q} + \vec{k}}$, $\hat{\rho}_{S \vec{k}} = \sum_{\vec{q}}  

160: c^{\dagger}_{\vec{q}} c_{\vec{q} + \vec{k}}$. Equations~(\ref{eq:bilinear})  

161: and (\ref{eq:linear}) 

162: stand, respectively, for bilinear and linear probe-system coupling. 

163: For optical probes (\ref{eq:bilinear}) and (\ref{eq:linear}) correspond 

164: to Raman and single-photon scattering, respectively. For atom or 

165: neutron probes the coupling (\ref{eq:bilinear}) is appropriate.
 

166: 

167: The Green function~(\ref{eq:Green}) is obtainable, to all orders in 

168: $\hat{V}$\cite{ande61}, by solving the set of equations for its diagonal elements 

169: \be 

170: G_{\vec{k} \vec{k}}(\eps) = [ \eps - \eps_{\vec{k}} - \sum_{\vec{k}^{'}} 

171: \langle \hat{V}_{\vec{k} \vec{k}^{'}}^2 \rangle\, G_{\vec{k}^{'} 

172: \vec{k}^{'}}(\eps) ]^{-1}, 

173: \label{eq:diagonal} 

174: \ee 

175: where $\eps_{\vec{k}}$ are the probe energy eigenvalues in the absence 

176: of potential fluctuations and pointed brackets denote the expectation value. 

177: The spectral information contained in these 

178: $G_{\vec{k} \vec{k}}$ is given by the density of states (DOS) of the probe 

179: $g(\eps) = - \frac{1}{\pi} \mbox{\rm Im} \sum_{\vec{k}} G_{\vec{k} 

180: \vec{k}}(\eps)$.
 

181: 

182: In order to extract information on the system we shall make two 

183: simplifying assumptions regarding the probe and the coupling 

184: potential~(\ref{eq:potential}): (i) $f_{\vec{k}}$ is flat in $\vec{k}$ (the 

185: coupling is strongly localized in space) within a band exceeding the 

186: relevant band of the system, so that $f_{\vec{k}} \approx f$; 

187: (ii) the statistical distribution of the probe is also flat in $\vec{k}$ and 

188: its second moment in $\langle \hat{V}^2 \rangle$ is replacable by 

189: the square of its mean flux (or density) $\bar{n}_{P}^2$ for the 

190: bilinear coupling (\ref{eq:bilinear}) or by its mean flux (density) 

191: $\bar{n}_{P}$ for the linear coupling 

192: (\ref{eq:linear}). Under these assumptions 

193: we can rewrite the squared coupling potential in (\ref{eq:diagonal}) as 

194: \be 

195: \langle \hat{V}_{\vec{k} \vec{k}^{'}}^2 \rangle = 

196: \langle \hat{V}_{\vec{k} \vec{k}^{'}} \rangle^2 +

197: \FF\, \SS.

198: \label{eq:coupl}

199: \ee 

200: Here $\langle \hat{V} \rangle$ is the mean coupling potential and $\FF \sim 

201: |f|^2\, \bar{n}_{P}^2$ or $\FF \sim |f|^2\, \bar{n}_{P}$ in the case of 

202: (\ref{eq:bilinear}) and (\ref{eq:linear}), respectively. The quantity 

203: of interest for the system is the Fourier-transformed 

204: density-density correlation of the atomic system 

205: \be 

206: \SS = \langle \hat{\rho}_{S\vec{k}}^{\dagger} \hat{\rho}_{S\vec{k}^{'}} 

207: \rangle + \mbox{\rm c.c}. 

208: \label{eq:S-matrix} 

209: \ee 

210: Its diagonal element ${\cal S}_{\vec{k} \vec{k}}$ is the static 

211: structure factor ${\cal S}_{\vec{k}}$, which is the Fourier transform  

212: of the van-Hove correlation function $\langle \hat{\rho}_{S}^\dagger(\vec{r},t=0) 

213: \hat{\rho}_{S}({\vec{r}}^{'},t^{'}=0) \rangle$ for the spatial density 

214: fluctuations of the "frozen" atomic ensemble.
 

215: 

216: The difficulty of having to evaluate or measure the matrix elements 

217: $\SS$ is avoided for a spatially random density distribution of the atomic 

218: system, due to random site occupancy (Fig.~\ref{fig:Dos1D}, inset a) and 

219: short-range interaction with the probe (e.g., a neutron or thermal atom). 

220: The elements $\SS$ in (\ref{eq:coupl}) and (\ref{eq:S-matrix}) can 

221: then be replaced by the average of the structure factor over all 

222: relevant $\vec{k}$: 

223: \be 

224: \SS \rightarrow \bar{\cal S} = \int d \vec{k}\, {\cal S}_{\vec{k}} \sim 

225: \langle \hat{n}_{S}^2 \rangle - \langle \hat{n}_{S} \rangle ^2, 

226: \label{eq:struct} 

227: \ee 

228: where the right-hand side of $\bar{\cal S}$ denotes the local atomic density 

229: or number variance averaged over the ensemble. 

230: 

231: The implications of evaluating the probe DOS $g(\epsilon)$ using 

232: (\ref{eq:diagonal})-(\ref{eq:struct}) will be examined for 

233: random fluctuations about a mean scattering potential $\langle V_{S}(x) \rangle$ 

234: (corresponding to the mean atomic density distribution) that is 1D-periodic. 

235: The "unperturbed" probe dispersion associated with $\langle V_{S}(x) \rangle$

236: is $\eps_{\vec{k}} = - 2J\, \cos (k_{x}d) + A$,  

237: $J$ being the hopping frequency, $d$ the lattice period and $A$ the band  

238: energy offset. This gives rise to the following expression for the 

239: Green function (\ref{eq:diagonal})

240: \be 

241: G(\eps) = \left[ \eps - \eps_{\vec{k}} - \langle W^2 \rangle

242: \sum_{\vec{k}^{'}} (\eps - \eps_{\vec{k}^{'}} - \Lambda (\eps)

243: + i \Delta (\eps) )^{-1} \right]^{-1}.

244: \label{eq:Green2}

245: \ee

246: Here $\langle W^2 \rangle \equiv \FF \, \bar{{\cal S}}$, $\Lambda (\eps) =

247: \langle W^2 \rangle / \sqrt{(\eps - A)^2 - 4J^2}$ for 

248: $|\eps - A| > 2J$, $\Delta (\eps) = \langle W^2 \rangle /

249: \sqrt{4 J^2 - (\eps - A)^2}$ for $|\eps - A| < 2J$ and both zero

250: otherwise. Figure~\ref{fig:Dos1D} shows how the probe DOS $g(\epsilon)$ 

251: changes from that of a periodic band structure corresponding to the mean 

252: potential $\langle V_{S}(x) \rangle$ to a semicircular shape 

253: as the amount of fluctuations measured by 

254: $\langle W^2 \rangle$ increases. 

255: 

256: In the multiple-scattering scenario,
 

257: which pertains to resonantly scattered atomic probes or to intracavity 

258: optical probes (Fig.~\ref{fig:Dos1D}, inset b), semicircular lineshapes are
 

259: obtained even when the $\SS$ cannot be claimed to belong to a random 

260: distribution (Fig.~\ref{fig:bose}, inset).
 

261: In the case of strongly-interacting atoms within a lattice site or longe-range 

262: intersite density correlations\cite{odel99} the distribution may be quite 

263: intricate, corresponding to sharp peaks of $\SS$. Nevertheless, the universal 

264: spectral trends of Fig.~\ref{fig:Dos1D} can be shown
 

265: to hold in this scenario, provided $\langle \hat{V}^2 

266: \rangle^{1/2} g_{0}(\eps) \gg 1$, $g_{0}(\eps)$ denoting 

267: the "unperturbed" probe DOS. This condition allows  

268: us to estimate $G_{\vec{k} \vec{k}}$ in (\ref{eq:diagonal}) to all orders in  

269: $\hat{V}$, upon replacing the state of the atomic system by a gaussian  

270: random ensemble\cite{meht91,akul93}. The result is the following universal 

271: formula\cite{akul93} for the renormalized probe energy $\tilde{\eps}$ at a 

272: given input energy $\eps$ 

273: \be 

274: \eps = \tilde{\eps} + \langle W^2 \rangle \mbox{\rm Tr}_{P} \left (

275: \frac{1}{\tilde{\eps} - \hat{H}_{P} - i0} \right).

276: \label{eq:renorm}

277: \ee

278: The use of (\ref{eq:renorm}) leads to a semicircular 

279: lineshape similar to the one in Fig.~\ref{fig:Dos1D}, as if the 

280: potential were random.
 

281: 

282: In order to illustrate the role of temperature, quantum statistics and 

283: the mean lattice potential in producing the semicircular lineshape, we 

284: proceed to evaluate $\langle W^2 \rangle$ for several simple models:

285: \newline  

286: 1. \underline{The isolated-site limit}: The tightly-bound Bose or 

287: Fermi distributions in a lattice can be estimated by taking the 

288: potential of each site to be that of a harmonic well of depth $V_{0}$. 

289: The isolated-site approximation holds for atoms in the lowest vibrational 

290: band, when the coupling energy is much smaller than the 

291: excitation energy to the next band\cite{jaks97}, $\sqrt{\langle W^2 \rangle} 

292: \ll \hbar \omega_{\nu} = \frac{2\pi \hbar}{\lambda} \ 

293: \sqrt{\frac{2 V_{0}}{m}}$, $\lambda$ being the wavelength of the laser light. 

294: %

295: %

296: \begin{figure}

297: \centerline{\epsfig{figure=semi.eps,width=0.7\hsize}}

298: \caption[]{Density of states $g(\eps)$ of a probe scattered by bosonic atoms 

299: in a 1D optical lattice. Solid, dotted and dashed curves stand for 

300: $\Delta(\eps)$ and the thick curve stands for $\Lambda(\eps)$ 

301: (dispersion), see text. All curves 

302: are numerically computed from $G(\eps)$ and correspond to 

303: average random couplings $\langle W^2 \rangle$ = 0.4, 

304: 2 and 10 respectively. The hopping frequency $J=1$, and for all 

305: curves $\int d\eps\, g(\eps) =1$. Inset a: A probe weakly scattered by a 

306: randomly occupied lattice. Inset b: A probe multiply scattered by a regular 

307: atomic distribution. 

308: } 

309: \label{fig:Dos1D} 

310: \end{figure}

311: %

312: 

313: In the absence of additional 

314: external perturbations, the coupling $\langle W^2 \rangle$ arises  

315: because of temperature-dependent fluctuations in the site-occupancy of 

316: the optical lattice, which has an 

317: approximately gaussian distribution\cite{depu99}. 

318: The resulting random coupling energies averaged over all states 

319: yield $\langle W^2 \rangle  \sim  \FF \left\langle \langle n_{\vec{k}}, 

320: n_{\vec{k}^{'}} \rangle - \langle n_{\vec{k}} \rangle 

321: \langle n_{\vec{k}^{'}} \rangle \right\rangle_{\vec{k}\vec{k}^{'}}  

322: = \FF \left\langle \langle n_{i}, 

323: n_{j} \rangle - \langle n_{i} \rangle 

324: \langle n_{j} \rangle \right\rangle_{ij}  

325: \approx \FF \left( \langle n_{S}^2 \rangle - \langle n_{S}\rangle^2 

326: \right)$. Here $n_{\vec{k}} \equiv c^{\dagger}_{\vec{k}} c_{\vec{k}}$, $i$ and $j$ 

327: label atomic sites, and $\langle n_{S} \rangle$ is the average  

328: number of atoms per site. The last step 

329: applies whenever $n_{i} \approx n_{j}$ $\forall i,j$ 

330: and the density fluctuations are approximately site-independent. 

331: We have verified this by numerical simulation, considering 

332: 2 to 4 identical atoms on a 1D lattice with  

333: 6 sites and calculating the density fluctuations if the  

334: probability of an occupied site is 1/10 of the  

335: probability of an empty site. In all cases the maximum relative difference between 

336: $\FF \SS$ and $\langle W^2 \rangle$ was less than 10 \%.
 

337: 

338: The kinetic contribution to $\langle W^2 \rangle$ due to evaporation 

339: of atoms from the lattice is the dominant one at high temperatures, 

340: regardless of the statistics. If all the atoms are in the lowest energy  

341: band, we may adopt the rate equation used to describe the formation of 

342: electron-hole clusters in a plasma\cite{klin81} and find 

343: \be 

344: \langle W^2 \rangle_{\rm evap} = a\, \langle n_{S} \rangle\, T^2\, 

345: e^{-\beta V_{0}}. 

346: \label{eq:evap}

347: \ee

348: Here $a = k_{B} m c_{p}$, with $k_{B}$ the Boltzmann constant, $m$ 

349: the mass of the atoms and $c_{p}$ their specific heat, 

350: $T$ denotes the temperature, $\beta^{-1} \equiv k_{B} T$ and 

351: $V_{0}$ is the optical lattice potential. 

352: The influence of  

353: evaporation becomes the dominant effect for $T\sim 25\, \mu$K. 

354: Around $T\sim 300\, \mu$K 

355: these fluctuations become comparable in size to the square of the optical 

356: lattice potential ($\langle W^2 \rangle \sim V_{0}^2 \sim 100$ (neV)$^2$) 

357: and atoms then largely escape from the lattice. 

358: 

359: At low temperatures (well below 100 $\mu$K) the density-density fluctuations 

360: depend on whether the atoms in the lattice are bosons or fermions. 

361: For bosonic atoms in the lowest vibrational state we obtain\cite{reic98}

362: \be

363: \langle W^2 \rangle_{\rm stat,Bose} = \frac{z}{1-z} +

364: \left( \frac{z}{1-z} \right)^2 + \frac{d^3}{\lambda_{T}^3} 

365: \sum_{\alpha = 1}^{\infty} \frac{z^{\alpha}}{\alpha^{1/2}}.

366: \label{eq:statBose}

367: \ee

368: Here we have approximated the motion of the atoms in the potential wells 

369: by a harmonic oscillation with frequency $\omega_{v}$\cite{appr}, 

370: $z\equiv e^{- k_{B} T/\hbar \omega_{v}}$, $d$ denotes the average 

371: lattice spacing and $\lambda_{T} = (2\pi\hbar^2/m k_{B} T)^{1/2}$, the thermal 

372: wavelength, is the length scale separating quantum 

373: statistical behavior (for $\lambda_{T} \sim d$) from classical 

374: Maxwell-Boltzmann behavior (for $\lambda_{T} \ll d$). 

375: For fermionic atoms in an optical lattice\cite{andr99} one starts with 

376: the analog of the coupling~(\ref{eq:potential}) for particles
 

377: with spin, using creation and annihilation operators 

378: $c_{\vec{k}\sigma}^{\dagger}$ and $c_{\vec{k}\sigma}$ and 

379: performing an additional sum over the spin index $\sigma$, and 

380: follows the same analysis as above. One then finds  

381: \be

382: \langle W^2 \rangle_{\rm stat,Fermi} = \frac{z}{1+z} +

383: \left( \frac{z}{1+z} \right)^2 + \frac{2\, d^3}{\lambda_{T}^2 \lambda_{F}}\, 

384: \sum_{\alpha = 1}^{\infty} \frac{z^{\alpha}}{\alpha^{1/2}},

385: \label{eq:statFermi} 

386: \ee

387: with $\lambda_{F}$ the Fermi wavelength. At high temperatures $z \rightarrow 0$ 

388: and both (\ref{eq:statBose}) and (\ref{eq:statFermi}) reduce to the classical  

389: Maxwell-Boltzmann result $\langle W^2 \rangle_{\rm stat,clas} = 

390: z$.  

391: At low temperatures, 

392: fermionic fluctuations approach a constant value,  

393: whereas bosonic fluctuations become very large as $T$ decreases  

394: below $\sim 1 \mu$K, marking the Bose-Einstein condensation. 

395: 

396: In Fig.~\ref{fig:bose} we have taken typical 

397: parameters for available optical lattices\cite{appr,birk95,depu99} 

398: to show how $\langle W^2 \rangle$  

399: evolves as a function of temperature for both bosonic and 

400: fermionic Li atoms. The total density-density fluctuations consist 

401: of the sum of (\ref{eq:evap}) and either (\ref{eq:statBose}) or 

402: (\ref{eq:statFermi}), depending on the statistics. 

403: The isolated-site condition 

404: is satisfed for the entire temperature range displayed in 

405: Fig.~\ref{fig:bose}. Since the hopping frequency $J\sim V_{0}$, 

406: the random coupling for bosonic Li atoms changes from $\langle W^2 

407: \rangle/J^2 \sim\, 

408: 10$ to  $\langle W^2 \rangle/J^2 \sim\, 100$, when going from $T\sim 8\,  

409: \mu$K to $T\sim 100\, \mu$K. Simultaneously the DOS 

410: then evolves from the periodic to the semicircular shape as in 

411: Fig.~\ref{fig:Dos1D}. The behavior for Na or Rb atoms 

412: is found to be similar to that of bosonic Li atoms , apart from their

413: different mass values in (\ref{eq:evap}) and (\ref{eq:statBose}). 

414: Employing currently achievable Bragg scattering techniques\cite{birk95} 

415: with a far off-resonant laser (1 mW/cm$^2$ intensity, 5.2 MHz detuning) 

416: and a lattice with a sufficiently high atomic filling factor\cite{depu99}, 

417: the scattering spectrum is expected to evolve with T, in the microkelvin 

418: range, from the  discrete to the semicircular regime.

419: %

420: %

421: \begin{figure}

422: \centerline{\epsfig{figure=W2paper2.eps,width=0.7\hsize}\vspace{0.5cm}}

423: \caption[]{Density-density fluctuations 

424: $\langle W^2 \rangle$ (in units of 10 (neV)$^2$) as a function of temperature T for 

425: bosonic and fermionic Li atoms in an optical lattice. Thin solid line - 

426: fluctuations due to evaporation (\ref{eq:evap}) (scaled by a factor of 7.8), 

427: thin dashed line - 

428: statistical fluctuations (\ref{eq:statBose}).  

429: Thick solid line - total fluctuations  

430: $\langle W^2 \rangle$ for bosonic Li atoms; thick dashed line - 

431: their counterpart for fermionic Li atoms. 

432: Parameters used are: $V_{0}=5$ neV, $\langle n_{S} \rangle$ = 0.1, d = 0.1 $\mu$m,  

433: %$c_{p}$(Cs) $\sim (0.2)\cdot 10^{3}$J kg$^{-1}$ K$^{-1}$ and 

434: %$\omega_{v}$(Cs)$\sim 4\cdot 10^5$ s$^{-1}$; for Li: 

435: $c_{p}$(Li) $\sim (3.6)\cdot 10^{6}$ J kg$^{-1}$  

436: K$^{-1}$, $\lambda_{F}$ (Li) = $6\cdot 10^{-10}$ m and  

437: $\omega_{v}$(Li)$\sim 2\cdot 10^6$ s$^{-1}$\cite{appr}. 

438: Inset: 

439: Static structure factor vs. $\kappa$ for phonons (solid curve) and  

440: nearly-free fermions (dashed curve) in a lattice at finite T.  

441: } 

442: \label{fig:bose} 

443: \end{figure} 

444: %

445: %  

446: 2. \underline{The nearly-free limit}: A Bose or Fermi gas weakly modulated  

447: by the lattice potential yields 

448: $\SS = {\cal S}_{\vec{\kappa} = \vec{k} - \vec{k}^{'}} =  

449: | \phi_{\vec{\kappa}} |^2 {\cal S}_{\vec{\kappa}}^{(\mbox{\rm free})}$. 

450: Here $\vec{\kappa}$ is a reciprocal lattice vector, $\phi_{\vec{\kappa}}$ 

451: is the corresponding Fourier harmonic of the lattice potential (normalized to 1)  

452: and ${\cal S}_{\vec{\kappa}}^{(\mbox{\rm free})}$ is the structure factor  

453: for momentum transfer $\hbar \kappa$ in a free Bose or Fermi gas. For a  

454: Fermi gas ${\cal S}_{\vec{\kappa}}^{(\mbox{\rm free})} = \pm \Theta  

455: (k_{f} - \kappa)$, the Fourier transform of pair correlations with parallel 

456: or anti-parallel spins (which determines the sign): it is the well-known  

457: step function which vanishes for $\kappa$ larger than the Fermi wavevector $k_{f}$. 

458: At finite temperatures this distribution broadens. The 

459: replacement of the nearly-free fermionic $\SS$ by the average value 

460: (\ref{eq:struct}) is then justifiable only in the 

461: multiple-scattering scenario, while in the weak-scattering scenario the 

462: lattice potential harmonics $\phi_{\vec{k}}$ pick out well-defined  

463: ${\cal S}_{\vec{k} - \vec{k}^{'} = \vec{\kappa}}$  

464: (Fig.~\ref{fig:bose}, inset - dashed line). 

465: \newline 

466: 3. \underline{The phonon regime}: Excitations at frequencies  

467: below the chemical potential of a Bose condensate trapped in a lattice 

468: can produce collective phonon modes\cite{stam99} whose "frozen" spectrum is  

469: characterized by ${\cal S}_{\vec{\kappa}} = \sum_{\vec{q}} [ ( \langle  

470: n_{\vec{q}} \rangle + 1) \sum_{\vec{G}} \delta (\vec{\kappa} - \vec{q} - \vec{G}) 

471: +  \langle n_{\vec{q}} \rangle \sum_{\vec{G}} \delta (\vec{\kappa} + \vec{q} + 

472: \vec{G}) ]$, where $\langle n_{\vec{q}} \rangle$ is the mean number of phonons 

473: at temperature $T$ with wavevector $\vec{q}$, and $\vec{G}$ denotes the  

474: reciprocal lattice vector. The phonon mode spectrum includes quasi-local modes 

475: in the case of fluctuating atomic distributions. This naturally leads to the  

476: limit (\ref{eq:struct}) and an effectively random coupling  

477: (Fig.~\ref{fig:bose}, inset - solid line). 

478: % 

479: % 

480: 

481: To conclude, we have identified novel regimes of probe scattering by  

482: atoms trapped in optical lattices in the random-density and multiple-scattering  

483: regimes. These regimes cannot be treated by the mean-field approximation, but  

484: are characterized by a universal feature of large density fluctuations,  

485: namely, semicircular scattering lineshapes. This is the atom-optical  

486: analog of the semicircular broadening of the DOS in disordered electronic 

487: systems, which as far as we know has not yet been 

488: observed unambiguously. The observation of this atom-optical counterpart 

489: presents a nontrivial but feasible challenge for experimentalists.

490: 

491: This work was supported by US-Israel BSF , the Israel Council 

492: for Higher Education, Minerva and Arc-en-Ciel.

493: 

494: \begin{references}

495: 

496: \bibitem{jess96}

497: P. Verkerk {\it et al.}, Phys. Rev. Lett. {\bf 68}, 3861 (1992); 

498: M. Raizen {\it et al.}, Phys. Today July 1997, 30 (1997); 

499: M. Ben-Dahan {\it et al.}, Phys. Rev. Lett. {\bf 76}, 4508 (1996).

500: 

501: \bibitem{park98}

502: A.S. Parkins and D.F. Walls, Phys. Rep. {\bf 303}, 1 (1998);

503: S. Inouye {\it et al.}, Nature {\bf 392}, 151 (1998); 

504: D.M. Stamper-Kurn {\it et al.}, Phys. Rev. Lett. {\bf 80}, 2027 (1998);

505: B.D. Anderson and M.A. Kasevich, Science {\bf 282}, 1686 (1998); 

506: M.O. Scully, Phys. Rev. Lett. {\bf 82}, 3927 (1999). 

507: 

508: \bibitem{depu99} 

509: M.T. DePue et al., Phys. Rev. Lett. {\bf 82}, 2262 (1999). 

510: 

511: \bibitem{jaks97}

512: D. Jaksch {\it et al.}, Phys. Rev. Lett. {\bf 81}, 3108 (1998). 

513: 

514: \bibitem{odel99} 

515: D. O'Dell, S. Giovanazzi, G. Kurizki and V.M. Akulin, Phys. Rev. Lett.

516: {\bf 84}, 5687 (2000). 

517: 

518: \bibitem{meht91} 

519: M.L. Mehta, {\it Random Matrices}, (Academic, New York, 1991);

520: B. Velicky {\it et al.}, Phys. Rev. 

521: {\bf 179}, 747 (1968).

522: 

523: \bibitem{akul93} 

524: V.M. Akulin and G. Kurizki, Phys. Lett. A {\bf 174}, 267 (1993); 

525: V.M. Akulin, Phys. Rev. A {\bf 48}, 3532 (1993). 

526: 

527: \bibitem{ngo94}

528: D. Ngo and R.G. Pinnick, J. Opt. Soc. Am. A {\bf 11}, 1352 (1994). 

529: 

530: \bibitem{ande61} 

531: P.W. Anderson, Phys. Rev. {\bf 124}, 41 (1961).

532: 

533: \bibitem{klin81}

534: C. Klingshirn and H. Haug, Phys. Rep. {\bf 70}, 315 (1981). 

535: 

536: \bibitem{reic98} 

537: L.E. Reichl, {\it A modern Course in Statistical Physics}, 

538: (Wiley, New York, 1998). 

539: 

540: \bibitem{appr}

541: A. Kastberg {\it et al.}, Phys. Rev. Lett. {\bf 74}, 1542 (1995). 

542: 

543: \bibitem{andr99}

544: M.R. Andrews, Nature {\bf 398}, 195 (1999).

545: 

546: 

547: \bibitem{birk95}

548: G. Birkl {\it et al.}, Phys. Rev. Lett. {\bf 75}, 2823 (1995).

549: 

550: 

551: \bibitem{stam99}

552: D.M. Stamper-Kurn {\it et al.}, Phys. Rev. Lett. {\bf 83}, 2876 (1999).

553: 

554: \end{references}

555: 

556: \end{document}