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

2: \documentclass[nofootinbib,

3: showpacs,prl,twocolumn,aps,floatfix]{revtex4}

4: 

5: \usepackage{graphicx}

6: %\usepackage{amssymb}

7: \def\la{\langle}

8: \def\ra{\rangle}

9: \newcommand {\fexp} [1] {\exp \left( #1 \right)}

10: \newcommand {\fabsq}[1] {\left| #1 \right|^2}

11: \newcommand {\fabs}[1] {\left| #1 \right|}

12: \newcommand {\fnorm} [1] {\left\| #1 \right\|}

13: \newcommand {\fnormq} [1] {{\left\| #1 \right\|}^2}

14: \newcommand {\fcot} [1] {\cot\left( #1 \right)}

15: \newcommand {\fcos} [1] {\cos\left( #1 \right)}

16: \newcommand {\fsin} [1] {\sin\left( #1 \right)}

17: \newcommand {\fcosq} [1] {\cos^2\left( #1 \right)}

18: \newcommand {\fsinq} [1] {\sin^2\left( #1 \right)}

19: \newcommand {\fcosh} [1] {\cosh\left( #1 \right)}

20: \newcommand {\fsinh} [1] {\sinh\left( #1 \right)}

21: \newcommand {\fcoshq} [1] {\cosh^2\left( #1 \right)}

22: \newcommand {\fsinhq} [1] {\sinh^2\left( #1 \right)}

23: \newcommand {\fcoth} [1] {\coth\left( #1 \right)}

24: \newcommand {\ftanh} [1] {\tanh\left( #1 \right)}

25: \newcommand {\Msi}{\times 10^6\, \mbox{s}^{-1}}

26: \newcommand {\ksi}{\times 10^3\, \mbox{s}^{-1}}

27: \newcommand {\si} {/\mbox{s}}

28: \newcommand {\ms} {\, \mbox{ms}}

29: \newcommand {\mum}{\, \mu \mbox{m}}

30: \newcommand {\cms}{\, \mbox{cm/s}}

31: \newcommand {\cO} {\begin{cal} O \end{cal}}

32: \newcommand{\beq}{\begin{equation}}

33: \newcommand{\eeq}{\end{equation}}

34: \newcommand{\beqa}{\begin{eqnarray}}

35: \newcommand{\eeqa}{\end{eqnarray}}

36: \newcommand{\intf}{\int_{-\infty}^\infty}

37: \newcommand{\into}{\int_0^\infty}

38: %

39: \begin{document}

40: \title{Velocity selection 

41: of ultra-cold atoms with Fabry-Perot laser devices:

42: improvements and limits}

43: 

44: \author {A. Ruschhaupt}

45: 

46: \author {F. Delgado}

47: 

48: \author {J. G. Muga}

49: \affiliation{Departamento de Qu\'{\i}mica-F\'{\i}sica,

50: UPV-EHU,\\

51: Apartado 644, 48080 Bilbao, Spain}

52: 

53: 

54: 

55: \begin{abstract}

56: We discuss a method to select the velocities of ultra-cold atoms

57: with a modified Fabry-Perot type of device made of two

58: effective barriers and a well created, respectively, by 

59: blue and red detuned lasers.   

60: The laser parameters may be used to 

61: select the peak and width of the transmitted

62: velocity window. In particular, lowering the central 

63: well provides a peak arbitrarily close to zero velocity  

64: having a minimum but finite width.   

65: The low-energy atomic scattering off this laser device

66: is parameterized 

67: and approximate formulae are 

68: found to describe and explain its behaviour.  

69: \end{abstract}

70: \pacs{32.80.Pj, 42.50.Vk, 03.75.-b}

71: \maketitle

72: 

73: 

74: Velocity selection 

75: is a basic operation 

76: in quantum optics and atomic physics for a plethora of

77: applications. 

78: % such as  

79: %cooling, interferometry, atom lithography, atom lasers and lenses,

80: %measurements of 

81: %collision cross sections, momentum distributions, or other kinetic-energy

82: %dependent quantities.

83: There are mechanical (slotted

84: disks) and non-mechanical (optical) 

85: techniques available, useful for different

86: experimental circumstances, species, and energies.  

87: The large wavelengths achieved with laser cooling

88: have made the traditional methods no longer effective  

89: because 

90: of the increasing importance of gravity

91: and the  

92: quantum nature of translational motion. 

93: For example, the standard 

94: classical-mechanical analysis of mechanical velocity-selection

95: methods becomes 

96: invalid for small-time temporal slits, since they 

97: produce momentum spread in agreement with  

98: a time-energy uncertainty principle \cite{David}. 

99: Among the new methods, the velocity selection using

100: Doppler sensitive stimulated Raman transitions \cite{chu2}, 

101: and coherent population trapping

102: into a dark state \cite{Aspect}, provide selectivity in the 

103: ``transverse direction'' parallel to the 

104: lasers, and rely on specific internal level configurations. 

105: % ---------------- FIG. 1 BEGINS ----------------

106: \begin{figure}

107: \begin{center}

108: \includegraphics[width=.8\linewidth]{vs_fig_1}

109: \end{center}

110: \caption{\label{fig1} $m=\mbox{mass}({ }^{23}\mbox{Na})$;

111: (a) Gaussian functions, $d = 6 \mum$, $\sigma = 2 \mum$;

112: (b) square functions, $d = 5 \mum$}

113: \end{figure}

114: % ---------------- END FIG. 1 ----------------

115: %

116: Fabry-Perot (FP) cavities have been also proposed to provide coherent 

117: velocity selection or trapping for longitudinal motion, 

118: using detuned lasers perpendicular to the incident atoms \cite{WGTM93}

119: or microwave cavities \cite{LMW98,2cav}

120: to implement the 

121: partially reflecting mirrors. 

122: The velocity selection in these 

123: cavities is produced by the filtering effect of

124: resonance peaks in the

125: transmission probability. The potential of FP cavities

126: as trapping devices 

127: also stems from characteristic resonance features: 

128: high densities and large 

129: life times in the

130: interaction region.    

131: 

132: % ---------------- FIG. 2 BEGINS ----------------

133: \begin{figure}[t]

134: \begin{center}

135: \includegraphics[angle=-90,width=1.0\linewidth]{vs_fig_2}

136: \end{center}

137: \caption{\label{fig2}Transmittance versus velocity for 

138: different well

139: depths. $\widehat{V}_{b}=300 \,\hbar\si$;

140: $\widehat{V}_w = 0$ (dashed-dotted line), $\widehat{V}_w = 150\,\hbar\si$

141: (thick dotted line),

142: $\widehat{V}_w = 180.2\,\hbar\si$ (thick dashed line), $\widehat{V}_w = 

143: 180.25\,\hbar\si$

144: (solid line), $\widehat{V}_w = 180.5\,\hbar\si$ (thick solid line);

145: other parameters in Fig. \ref{fig1}a. The inset is a zoom

146: of the lower-left corner.}

147: \end{figure}

148: % ---------------- END FIG. 2 ---------------- 

149: The aim of this work is to discuss an improvement 

150: of these cavities, provide formulae to describe their

151: behaviour, 

152: and study the fundamental limitations to lower 

153: the peak width and velocity of the transmitted wave packet. 

154: The basic idea is to add a well with controllable depth 

155: between the two external barriers, see Fig. 1.

156: Effective barriers and well can

157: be implemented 

158: with blue and red detuned lasers, respectively, which do not excite 

159: the impinging ground state atom and cause only a mechanical effect.

160: The depth of the well and the barrier height can be varied with the 

161: intensities of the lasers. 

162: Making the well deeper, rather than wider, 

163: displaces the resonance peaks to lower energies

164: without diminishing the inter-resonance spacing  

165: so it is the ideal way to achieve a sharp low-energy velocity selection. 

166: The resonance peak displacement with the well-depth can be 

167: seen in Fig. 2. The velocity shift is accompanied by a peak width reduction 

168: until a minimum, non-zero width is attained when the peak reaches zero 

169: velocity at a critical ``threshold'' depth. Beyond that

170: depth the peak broadens, moves to higher velocities, and

171: its maximum decays, as shown in the inset of Fig. 2.

172: The effects of different depths are summarized in Fig. 3, 

173: which will be 

174: explained next in more detail.  

175:          

176: 

177: We shall use both a realistic model based on three Gaussians, see 

178: Fig. 1, as well as a simplified version with two square barriers 

179: and a well. The scattering off the two potential models is very similar  

180: but the later  

181: enables us to obtain analytical exact results and approximate but 

182: physically illuminating expressions. 

183: %

184: %

185: % ---------------- FIG. 3 BEGINS ----------------

186: \begin{figure}[t]

187: \begin{center}

188: \includegraphics[angle=-90,width=1.\linewidth]{vs_fig_3}

189: \end{center}

190: \caption{\label{fig3}

191: (a) Resonance velocity $v_R$ versus $\hat{V}_w$  

192: (symbols connected with dotted line).

193: Filled symbols indicate the case $\fabsq{T(v_R)} > 0.995$, and empty 

194: symbols

195: otherwise. The solid

196: lines show the approximation of Eq. (\ref{approx1}).

197: Gaussian functions, see Fig. \ref{fig1}a:

198: $\widehat{V}_b=300 \hbar\si$, $\alpha=0.65$ (diamonds),

199: $\widehat{V}_b=500 \hbar\si$, $\alpha=0.70$ (squares);

200: square functions, see Fig. \ref{fig1}b:

201: $\widehat{V}_b=300 \hbar\si$, $\alpha=0.79$ (triangles up),

202: $\widehat{V}_b=500 \hbar\si$, $\alpha=0.85$ (triangles down),

203: the circles indicate $\widehat{V}_{w,thres}$ for the square model 

204: in the 2-pole

205: approximation.

206: (b)Velocity width $\Delta v_R$ of the resonance versus

207: $\widehat{V}_w$; meaning of symbols as in (a).

208: The solid lines show the approximation of Eq. (\ref{delv}),

209: %The parameter $\beta$, in the same order of potentials used for (a)

210: %is $0.00585\,\mbox{s}/\hbar$, $0.00610\,\mbox{s}/\hbar$,

211: %$0.00424\,\mbox{s}/\hbar$, and $0.00418\,\mbox{s}/\hbar$,

212: the circles indicate $(\widehat{V}_{w,thres},

213: {\Delta v}_{R,thres})$ with the 2-pole approximation.}

214: \end{figure}

215: % ---------------- END FIG. 3 ----------------

216: %

217: Let us consider, for a single ultra-cold atom,

218: the Hamiltonian

219: %

220: \begin{eqnarray}

221: H = -\frac{\hbar^2}{2m}\,\frac{\partial^2}{\partial x^2} 

222: +  V_b (x+d) - V_w (x) + V_b (x-d),

223: \label{ham}

224: \end{eqnarray}

225: %

226: where $V_{b,w} (x)= \widehat{V}_{b,w} \;\Pi(x)$, and 

227: $\Pi$ can take the forms 

228: %

229: \begin{eqnarray*}

230: \Pi_G\!=\!\fexp{-\frac{x^2}{2 \sigma^2}},\;\;

231: \Pi_s\!=\!\left\{

232: \begin{array}{cc}

233: 1&{\rm if} -d/2<x<d/2

234: \\

235: 0&{\rm otherwise}

236: \end{array}

237: \right. 

238: \end{eqnarray*}

239: %

240: for the Gaussian and square models respectively. 

241: For simplicity we have set all Gaussians with the same width $\sigma$, and 

242: the square 

243: segments with the same length $d$.  

244: %

245: We assume  

246: that the atom impinges from the left and only initial 

247: positive velocities are considered. 

248: We are interested in the 

249: transmission amplitude $T$ 

250: and the ``transmittance'' $|T|^2$ 

251: of the scattering solutions of 

252: $

253: %\begin{eqnarray*}

254: H \phi_v (x) = E_v \phi_v (x),  

255: %\end{eqnarray*}

256: $

257: where $E_v = \frac{m v^2}{2}=\hbar^2 k^2/(2m)$.

258: Both velocity, $v$, and wavenumber, $k$, will be used,

259: the later being more appropriate for complex plane analysis and the former 

260: for presenting the physical results.     

261: For the square model,

262: %

263: %

264: %\begin{widetext}

265: \begin{eqnarray}

266: T(k)&=& -4 e^{-2idk} k k_b^2 k_w 

267: \label{trans}

268: \\

269: &\times&\Big\{ e^{id(k- k_w)}\left[

270: i k_b  (k+ k_w) C  

271: + (k k_w-k_b^2) S 

272: \right]^2

273: \nonumber\\

274: &\!-\!&\!\!e^{id(k+ k_w)}\!\!

275: \left[

276: -i k_b (k- k_w) C 

277: \!+\! (k k_w+k_b^2) S 

278: \right]^2\!\!\Big\}^{\!-1}

279: \nonumber

280: \end{eqnarray}

281: %\end{widetext}

282: %

283: where  

284: $C=\cosh(d k_b)$, $S=\sinh(d k_b)$, 

285: $k_w=(k^2+K_w^2)^{1/2}$,

286: $k_b=(K_b^2-k^2)^{1/2}$, and 

287: $K_b=(2m\widehat{V}_b)^{1/2}/\hbar$, 

288: $K_w=(2m\widehat{V}_w)^{1/2}/\hbar$. 

289: 

290: % ---------------- FIG. 4 BEGINS ----------------

291: \begin{figure}[t]

292: \begin{center}

293: %\includegraphics[angle=-90,width=0.5\linewidth]{vs_fig_4}

294: \includegraphics[width=3.5cm,height=7cm,angle=-90]{vs_fig_4}

295: \end{center}

296: \caption{\label{fig4} Connection between peaks and poles in the square model,

297: $\widehat{V}_b=400\hbar\si$, other parameters in Fig. \ref{fig1}b;

298: for the plotted parameter range, the 2-pole approximation and

299: the exact result are indistinguishable;

300: (a) resonance velocity $v_R$ (line with filled dots),

301: resonance width $\Delta v_R$ (line with unfilled diamonds)

302: versus the distance $\pm \lambda$ of the pole to the collision point

303: ($\lambda < 0$ before the collision, $\lambda > 0$ after the

304: collision); a 1-pole approximation with only $k_1$ is also plotted

305: ($v_R$ (dashed line), $\Delta v_R$ (dashed-dotted line)).

306: (b) Motion of the two poles:

307: $\widehat{V}_w=199.884\hbar\si$ (crosses),

308: $\widehat{V}_{w,coll}=199.898\hbar\si$ (coinciding squares),

309: $\widehat{V}_{w,thres}=199.901\hbar\si$ (small circles),

310: $\widehat{V}_w=199.92\hbar\si$ (big circles).}

311: \end{figure}

312: % ---------------- END FIG. 4  ----------------

313: 

314: In both models, at some critical,

315: ``threshold'' well-depths $\widehat{V}_{w,thres}$ 

316: new bound states are formed,  

317: and for well-depths close to these thresholds,

318: the description of the transmission 

319: peak is not as simple 

320: as for isolated Breit-Wigner (BW) resonances (see Fig. \ref{fig4}). At a depth

321: $\widehat{V}_{w,coll}$ 

322: slightly before threshold, 

323: a resonance-antiresonance pole collision occurs, 

324: and with further deepening, two ``virtual'' states in the complex momentum

325: plane appear. 

326: Centering our attention on the first, lowest-energy resonance at zero  

327: well-depth, one can distinguish 

328: when increasing the well-depth:

329: first an ordinary resonance regime with a  

330: BW transmittance peak; second, an intermediate pre-bound state regime  

331: near threshold, in which the second pole cannot be ignored;

332: and finally a bound-state regime. In Figure \ref{fig2} transmittance curves 

333: corresponding to the different regimes are depicted. 

334: Even though the calculations can be made exactly, it is 

335: useful for applications and physically illuminating to 

336: describe these three stages in terms of approximate expressions   

337: and dominant dependences relating:

338: poles of $T$ in the complex momentum plane;  

339: well-depth or other potential parameters;

340: and visible features such as position and width of the

341: resonance. 

342:  

343: 

344: The first stage, dominated by a BW resonance pole $k_1$      

345: in the fourth quadrant of the momentum complex plane, 

346: is the most important one for 

347: velocity selection since it allows to diminish the resonance velocity 

348: and width by deepening the well, see Figs. \ref{fig2} and \ref{fig3}a.    

349: $k_1$ is accompanied by an antiresonance at 

350: $k_2=-k_1^*$.

351: Since the barrier is symmetrical with respect to parity,

352: the (Gamow) resonance states 

353: have well defined parity. 

354: Thus they appear alternatively in one of the two eigenvalues 

355: of the $2\times2$ $S$ matrix, for symmetrical, 

356: $S_0$, or antisymmetrical scattering, $S_1$.

357: The transmission amplitude

358: is given by 

359: $

360: T=(S_{0}+S_{1})/2. 

361: $

362: We shall follow the 

363: motion of the first symmetrical resonance, 

364: the one that will become the ground state 

365: for deep enough wells,  

366: %  

367: and assume that the first antisymmetrical pole 

368: of $S_1$ is far from the origin so that a 

369: two pole approximation suffices. Retaining only two poles 

370: in the canonical pole expansion for cut-off potentials, 

371: %

372: \beq

373: S_0=-e^{-2ikr}\frac{(k-k_1^*)(k-k_2^*)}{(k-k_1)(k-k_2)},\;\;\;

374: S_1=e^{-2ikr}.

375: \label{ss1}

376: \eeq

377: %

378: Here $r$ is $3d/2$ for the square model. For the Gaussian model, 

379: we could  

380: truncate the potential at a large $r$ value and apply  

381: Eq. (\ref{ss1}).

382: In any case 

383: the phase factor does not play any role to calculate the filtering 

384: function $|T|^2$. 

385: 

386: Expressions for the two important poles can be obtained with   

387: the square model 

388: under some approximations, as we shall see later on.  

389: In the ``BW'' regime 

390: the antiresonance $k_2$ may normally be ignored if the resonance is 

391: sharp (i.e. $k_1$ is close to the real axis) and far from the origin.    

392: A decrease in the well-depth displaces $k_1$ to the left

393: and upwards, 

394: so that the transmittance curve decreases both its peak velocity and 

395: width. By inspection of the $S$ matrix, 

396: it is clear that in this regime 

397: the transmittance reaches the unitary limit $|T|^2=1$

398: close to $\Re {\rm{e}} (k_1)$.  

399: We define $E_R=mv_R^2/2$ and $v_R$ as the energy and velocity 

400: of the transmittance maximum. 

401: In the BW regime $E_R\approx \hbar^2\Re {\rm{e}}(k_1)^2/2m$.

402: Simple parameterizations of this regime are provided  

403: by perturbation or semiclassical formulae. 

404: Let $E_{R0}$ and $v_{R0}$ be the real energy and velocity 

405: of the resonance peak ``without

406: well'' ($\widehat{V}_w=0$).

407: Then the energy of the resonance with

408: non-zero well can be approximated within a perturbation theory for

409: resonance functions \cite{kukulin.book}. 

410: Up to first order in $\widehat{V}_{w}$,

411: %

412: \beq

413: E_R=E_{R0} - \alpha \widehat{V}_w,\;\;\;

414: v_R = \sqrt{v_{R0}^2 - 2\alpha \widehat{V}_w/m}.

415: \label{approx1}  

416: \eeq

417: %

418: %(Similar formulae can be obtained for perturbations of 

419: %$\widehat{V}_{b}$.) 

420: A semiclassical treatment \cite{Bohm} for opaque 

421: barriers gives $\alpha=1$, 

422: but keeping $\alpha$ as a fitting parameter the dependence 

423: of Eq. (\ref{approx1}) is valid even beyond very opaque

424: barriers or very small depths,

425: as can be seen in Fig. \ref{fig3}a. 

426: %

427: We define a velocity width $\Delta v_R$ as the width of the transmittance 

428: peak at 

429: half height. In 

430: Fig. \ref{fig3}b the BW regime corresponds to the slow decrease with 

431: well-depth up to 

432: the abrupt, almost vertical increase associated with a bound state.     

433: A semiclassical estimate for the energy-width of the resonance is 

434: given by 

435: a well-frequency factor times 

436: the WKB probability to escape through a barrier from the well

437: (see e.g. \cite{Bohm}). Retaining dominant dependences in the 

438: opaque and shallow well limit,  

439: %

440: \beq

441: \Delta v_R= \Delta v_{R0} \exp(-\beta \widehat{V}_w),

442: \label{delv}

443: \eeq

444: %

445: which, again, by keeping $\beta$ as an effective fitting parameter, 

446: describes the correct behaviour in the whole BW regime, until 

447: well-depths very near the intermediate threshold region, see Fig. \ref{fig3}b.  

448: %\begin{eqnarray}

449: %\Delta v \sim \partial \int \left[\frac{v_{Rw}}{d}\right] \frac{v_R}{v_{Rw}}

450: %\fexp{-\frac{2}{\hbar}\int_{x_1(E_R)}^{x_2(E_R)} 

451: %dx \, \sqrt{V_b (x) - E_R}}

452: %\label{eq}

453: %\end{eqnarray}

454: %

455: %with $V_0(x_1(E_R) = V_0(x_2(E_R)) = E_R$, and

456: %$v_{Rw}=[2(E_R-\widehat{V}_w)/m]^{1/2}$. 

457: %We have $\Delta E = E(v_R + \Delta v_R) - E(v_R - \Delta v_R) 

458: %= \frac{m}{2} \Delta v_R (v_{End} + v_{Start}) \approx m v_R \Delta_v$.

459: %Furthermore we approximate the integral in Eq. (\ref{eq})

460: %by $d_{eff} \sqrt{\widehat{V}_{b}-E_R}$

461: %and, by taking into account the linear dependence of $E_R$ with the well-depth, 

462: %Eq. (\ref{per}),  

463: %we arrive at the simple exponential dependence 

464: %%

465: %\begin{equation}

466: %\Delta v_R (v_R, \widehat{V}_{b}) \sim  

467: %v_{R0}\exp{-\alpha \widehat{V}_w} 

468: %\end{equation}

469: %

470: 

471: Near the threshold depth 

472: the velocity and width of the peak are

473: affected more and more 

474: by the nearby 

475: antiresonance, $k_2=-k_1^*$.

476: This intermediate regime is extremely narrow, with respect to 

477: variations of $\widehat{V}_w$, compared to the BW

478: and bound state ones (see Fig. \ref{fig4}). Nevertheless,  

479: its analysis is worthwhile since it establishes the ultimate 

480: physical lower limit of the peak velocity and width using a 

481: FP filtering device. 

482: At a critical ``collision'' depth $\widehat{V}_{w,coll}$ 

483: both poles meet at $-i\kappa_{coll}$, $\kappa_{coll}>0$,  

484: on the negative imaginary axis (see e.g. the ``square'' in Fig. \ref{fig4};

485: in one dimensional scattering,  

486: as for s-wave scattering, the collision is not at the origin

487: because bound states are not degenerate).   

488: Note that, in spite of their zero real part, 

489: the velocity peak is not at $v=0$.  

490: As the well becomes more profound the two poles  move in opposite 

491: directions, now along the imaginary axis as ``virtual''  poles

492: until the upper one arrives at the origin at the threshold

493: depth $\widehat{V}_{w,thres}$, with the lower pole at $-i\kappa_{thres}$

494: (see e.g. ``circles'' in Fig. \ref{fig4}).

495: The motion of the two poles just before the collision and even beyond threshold 

496: is well described by expanding the denominator 

497: of $S_0$ in powers of $(\widehat{V}_w-\widehat{V}_{w,coll})$

498: and retaining the first term. This gives  

499: % 

500: %\beq

501: $

502: k_{1,2}=-i\kappa_{coll}

503: \pm i \gamma (\widehat{V}_{w}-\widehat{V}_{w,coll})^{1/2},

504: $

505: %\eeq

506: %

507: with $\gamma$ real. A consequence is that, at threshold,  

508: $k_1=0$, $k_2=-i\kappa_{thres}\approx -2i\kappa_{coll}$.

509: The threshold 

510: is a singular, abnormal  

511: point

512: in which the 

513: transmission peak reaches the origin, 

514: $T(0)=1$ ($T(0)=0$ for any other well-depth).

515: Moreover, from the 2-pole approximation of $S$,  

516: %

517: %\beq

518: $

519: |T|^2={\kappa_{thres}^2}/({k^2+\kappa_{thres}^2}). 

520: $

521: %\eeq

522: %

523: Thus, the width at half height, considering

524: only positive momenta, reaches its minimum value.   

525: In wavenumber units it is just $\kappa_{thres}$,

526: and the maximum, $|T|^2=1$,

527: occurs at  $k=\kappa_{thres}$.

528: Approximate expressions for $\kappa_{thres}$ may be obtained from 

529: Eq. (\ref{trans}).  

530: We have to find zeros of the denominator of $T$. 

531: Let $\chi \equiv \frac{k}{K_b}$. 

532: We assume $\widehat{V}_b,\widehat{V}_w > 0$ and $\chi \ll 1$.

533: Neglecting $\cO(\chi^3)$ we arrive at the quadratic equation

534: %

535: %\begin{eqnarray}

536: $\alpha_2 \chi^2 +2i \alpha_1 \chi - \alpha_0 = 0$

537: %\label{eqquad}

538: %\end{eqnarray}

539: %

540: with $\alpha_0 = \fcot{\frac{d}{2} K_w} - K_w\fcoth{d K_b}/K_b$,

541: $\alpha_1 = K_w/\left(2 K_b \fsinhq{d K_b}\right)$, and

542: \begin{eqnarray*}

543: \alpha_2 &=& \frac{K_b^2\left(d K_w + \fsin{d K_w}\right)}

544: {4 K_w^2 \fsinq{d K_w/2}}\\

545: & & + \frac{K_w\left(\fcoth{d K_b} \left(\fcoshq{d K_b} - 3\right) + d K_b\right)}

546: {2 K_b \fsinhq{d K_b}},\\

547: %\nonumber

548: %\\

549: \end{eqnarray*}

550: %\begin{eqnarray*}

551: %\alpha_1 = \frac{K_w}{2 K_b\fsinhq{d K_b}}, 

552: %\alpha_0 = \fcot{\frac{d}{2} K_w}

553: %- \frac{K_w\fcoth{d K_b}}{K_b} 

554: %\nonumber

555: %\end{eqnarray*}

556: %

557: %$\alpha_1 = K_w/\left(2 K_b \fsinhq{d K_b}\right)$, and

558: %$\alpha_0 = \fcot{\frac{d}{2} K_w} - K_w\fcoth{d K_b}/K_b$.

559: The two solutions are given by

560: $\chi_{1/2} =

561: -i \alpha_1/\alpha_2 \pm \sqrt{\alpha_0\alpha_2-\alpha_1^2}/\alpha_2$.

562: At $\widehat{V}_{w,thres}\equiv\hbar^2K_{w,thres}^2/2m$,

563: a zero of the denominator of $T$ is at $k=0$, so  

564: $\alpha_0 = 0$. Note that if $K_b\gg 1$, 

565: $K_{w,thres}\approx\pi/d$.

566: The other pole is at $-i\kappa_{thres}=-2iK_b\alpha_1/\alpha_2$

567: and determines the minimal velocity width 

568: ${\Delta v}_{R,thres} = \hbar\kappa_{thres}/m$

569: (see Fig. \ref{fig3}).

570: 

571: Finally, with further well deepening, the upper pole 

572: crosses the real axis and becomes a bound state. 

573: As predicted by Eq. (\ref{ss1}),  

574: the transmittance peak broadens dramatically

575: and moves to higher positive velocities; also  

576: the peak maximum becomes smaller than one, so this 

577: regime is no longer useful for velocity filtering, 

578: see Figs. \ref{fig2} and \ref{fig3}.    

579:  

580: 

581: As an application example

582: we shall compute  

583: the transmitted velocity distributions resulting from the FP 

584: filtering of 

585: an an atomic wavepacket prepared as a  

586: Bose-Einstein condensate \cite{anglin.2002,bongs.2004}

587: in a trap.

588: The trap is moved with a certain velocity with respect to 

589: the laboratory frame and turned off suddenly at $t=0$.  

590: The condensate expands until the nonlinear interaction

591: between the atoms can be neglected and encounters the FP cavity.  

592: (Alternatively the triple potential can be moved

593: with the trap at rest.)

594: First we calculate numerically the ground state $\bar{\psi}_0$

595: (normalized to $1$) in the reference frame where the trap is at rest.

596: A harmonic trap is assumed with frequency $\omega_x$ in $x$ direction

597: and $\omega_{yz}$ in $y$ and $z$ directions.

598: Using a one-dimensional approximation

599: of the Gross-Pitaevskii equation the Hamiltonian is 

600: %%

601: %\begin{eqnarray*}

602: %$

603: $ H(\bar{\psi})\!=\!\frac{-\hbar^2}{2m}

604: \frac{\partial^2}{\partial x^2}\!+\!

605: \frac{m\omega_x^2}{2}x^2

606: \!+\! 2\hbar N a

607: \omega_{yz} {\fabsq{\bar{\psi} (x)}},$

608: %\end{eqnarray*}

609: %

610: %$

611: where $N$ is the number of atoms in the condensate

612: and $a$ the scattering length. We take  $a=2.93\times 10^{-9} \, m$. 

613: %

614: Then we change  

615: to the lab frame where the trap moves with velocity

616: $v_0$. The ground state $\psi_0$ in this reference frame is 

617: $\psi_0 (x) = e^{i x \frac{mv_0}{\hbar}} \bar{\psi}_0 (x)$.

618: %

619: At $t=0$ the trap is turned off being at position $x_{TRAP}$

620: and the velocity selection potentials are

621: switched on, i.e., the time-evolution is

622: given by the Gaussian version of the Hamiltonian of 

623: Eq. (\ref{ham}) plus a term $V'$ 

624: representing the decaying 

625: non-linear effect due to free expansion in $y$ and $z$ directions,

626: $V'= 2\hbar N a \omega_{yz}  \fabsq{\psi (x)}/({1 + \omega_{yz}^2 t^2})$.

627: %

628: %

629: % ---------------- FIG. 5 BEGINS ----------------

630: \begin{figure}[t]

631: \begin{center}

632: \includegraphics[angle=-90,width=1.\linewidth]{vs_fig_5}

633: \end{center}

634: \caption{\label{fig5} $|\psi(v, t)|^2$ for $v_0=0.0336\cms$,

635: $x_{TRAP} = -600 \mum$; $\widehat{V}_b = 300\hbar\si$,

636: $\omega_x = 5\si$, $\omega_{yz}=100\si, N = 5\times 10^4$;

637: $t=0$ (dotted line);

638: $t=0.8s$: solid line;

639: $t=8s$: $\widehat{V}_w = 140 \hbar\si$ (thick dotted line),

640: $\widehat{V}_w = 150 \hbar\si$ (dashed line),

641: $\widehat{V}_w = 160 \hbar\si$ (thick solid line);

642: the circles mark the resonance velocities

643: $v_R$.}

644: \end{figure}

645: % ---------------- END FIG. 5  ----------------

646: %

647: Fig. \ref{fig5} shows the momentum distribution at $t=0$ (ground state). At $t=0.8 \, s$ 

648: the non-linearity has practically

649: vanished, so the momentum distribution stays stable 

650: until the velocity selection. 

651: The filtered distributions 

652: at $t=8 \, s$ for several 

653: resonance velocities

654: obtained with different well-depths are also shown. 

655: 

656: 

657: We have in summary proposed an improvement 

658: of Fabry-Perot cavities to select the velocity of 

659: ultra-cold atoms using a well between the partially

660: reflecting mirrors,

661: and have provided simple formulae to explain and describe their

662: behaviour and the minimal velocity peak (zero) and width 

663: (non-zero) 

664: that can be achieved. 

665: 

666: \section*{Acknowledgments}

667: 

668: This work is supported by 

669: %CERION-II (Canadian European Research Initiative on

670: %Nanostructures), 

671: ``Ministerio de Ciencia y Tecnolog\'\i a-FEDER''

672: (BFM2003-01003), and

673: UPV-EHU (Grant 15968/2004). AR acknowledges support of the German

674: Academic Exchange Service (DAAD)

675: and Ministerio de Educaci\'on y Ciencia.

676: 

677: 

678: \begin{thebibliography}{10}

679: 

680: %\bibitem{singer.1982}

681: %S. Singer, K.F. Freed, and Y.B. Band,

682: %{\rm J. Chem. Phys.} {\bf 77}, 1942 (1982).

683: %

684: %\bibitem{band.1994}

685: %Y.B. Band and I. Tuvi,

686: %{\rm J. Chem. Phys.} {\bf 100}, 8869 (1994).

687: 

688: %\bibitem{Scoles}C. J. van den Meijdenberg, in 

689: %{\it Atomic and Molecular Beam Methods}, ed. by 

690: %G. Scoles (Oxford, New York, 1988), Chap. 13. 

691:  

692: \bibitem{David} P. Szriftgiser, D. Gu\'ery-Odelin, 

693: M. Arndt, and J. Dalibard, Phys. Rev. Lett. {\bf 77}, 4 (1996).  

694: 

695: %\bibitem{chu1} M. Kasevich, D. S. Weiss, E. Riis, K. Moler, S. Kasapi, and 

696: %S. Chu, Phys. Rev. Lett. {\bf 66}, 2297 (1991). 

697: 

698: \bibitem{chu2} K. Moler, D. S. Weiss, M. Kasevich, and S. Chu, 

699: Phys. Rev. A {\bf 45}, 342 (1992)

700: 

701: \bibitem{Aspect} A. Aspect, E. Arimondo, R. Kaiser, N. Vansteenkiste

702: and C. Cohen-Tannoudji, Phys. Rev. Lett. {\bf 61}, 826 (1998). 

703: 

704: \bibitem{WGTM93} M. Wilkens, E. Goldstein, B. Taylor, and

705: P. Meystre, Phys. Rev. A {\bf 47}, 2366 (1993). 

706: 

707: \bibitem{LMW98} L. L\"offler, G. M. Meyer, and H. Walther, 

708: Europhys. Lett. {\bf 41}, 593 (1998). 

709: 

710: \bibitem{2cav} Z. M. Zhang, S. W. Xie, Y. L. Chen, Yu  X. Xia, 

711: S. K. Zhou, Phys. Rev. A {\bf 60}, 3321 (1999). 

712: 

713: %\bibitem{vtype} W. Q. Liang, Z. Q. Niu, Z. M. Zhang, and 

714: %S. W. Xie, J. Phys. B {\bf 34}, 4427 (2001).  

715: 

716: %\bibitem{detu} T. Bastin and J. Martin, Phys. Rev. A {\bf 67},

717: %053804 (2003). 

718: 

719: \bibitem{kukulin.book}

720: V.I. Kukulin, V.M. Krasnopol'sky,

721: and J. Hor\'a\v{c}ek, {\it Theory of Resonances},

722: (Kluwer, Dortrecht, 1989).

723: 

724: \bibitem{Bohm} D. Bohm, {\it Quantum Theory}, (Dover, New York, 1951). 

725: 

726: %\bibitem{schiff.book}

727: %L.I. Schiff, {\it Quantum mechanics}, (McGraw-Hill, 1981).

728: 

729: % review BEC

730: 

731: \bibitem{anglin.2002} J.R. Angelin and W. Ketterle,

732: {\it Nature} {\bf 416}, 211 (2002).

733: 

734: \bibitem{bongs.2004} K. Bongs and K. Sengstock,

735: {\it Rep. Prog. Phys.} {\bf 67}, 907 (2004).

736: 

737: \end{thebibliography}

738: 

739: \end{document}

740: