1: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
2:
3: %\documentclass[showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
4:
5: \usepackage{graphicx}% Include figure files
6: \usepackage{dcolumn}% Align table columns on decimal point
7: \usepackage{bm}% bold math
8:
9: \newcommand{\ket}[1]{\ensuremath{| #1 \rangle}}
10: \newcommand{\bra}[1]{\ensuremath{\langle #1 |}}
11:
12:
13: \begin{document}
14:
15:
16: \title{Analysis of Photoassociation Spectra for Giant Helium Dimers.}
17:
18: \author{J. L\'eonard}
19: \email{leonard@lkb.ens.fr}
20: \author{A. P. Mosk}
21: \altaffiliation[Permanent address: ]{Dept. of science \& technology, and MESA+ institute, University of Twente,
22: Netherlands.}
23: \author{M. Walhout}
24: \altaffiliation[Permanent address: ]{Calvin College, Grand Rapids, MI, USA. }
25: \author{P. van der Straten}
26: \altaffiliation[Permanent address: ]{Utrecht University, Netherlands. }
27:
28: \author{M. Leduc}
29: \author{C. Cohen-Tannoudji}
30:
31: \affiliation{Ecole Normale Sup\'erieure and Coll\`ege de France\\
32: Laboratoire Kastler Brossel, 24 rue Lhomond, 75231 Paris Cedex 05,
33: France}
34:
35:
36: \date{\today}% It is always \today, today,
37: % but any date may be explicitly specified
38:
39: \begin{abstract}
40: We perform a theoretical analysis to interpret the spectra of purely
41: long-range helium dimers produced by photoassociation (PA) in an
42: ultra-cold gas of metastable helium atoms. The experimental spectrum
43: obtained with the PA laser tuned closed to the $2^3S_1\leftrightarrow
44: 2^3P_0$ atomic line has been reported in a previous Letter. Here, we
45: first focus on the corrections to be applied to the measured resonance
46: frequencies in order to infer the molecular binding energies. We then
47: present a calculation of the vibrational spectra for the purely
48: long-range molecular states, using adiabatic potentials obtained from
49: perturbation theory. With retardation effects taken into account, the
50: agreement between experimental and theoretical determinations of the
51: spectrum for the $0_u^+$ purely long-range potential well is very good.
52: The results yield a determination of the lifetime of the $2^3P$ atomic
53: state.
54:
55: \end{abstract}
56:
57: \pacs{34.20.Cf, 32.80.Pj, 34.50.Gb}% PACS, the Physics and Astronomy
58: % Classification Scheme.
59: \maketitle
60:
61: \section{Introduction}
62: \label{Section: Introduction}
63:
64: Photoassociation (PA) spectroscopy is a powerful technique for
65: acquiring information about the collisional properties of laser-cooled
66: atoms. It has revealed a rich array of high-resolution spectroscopic
67: data for alkali diatomic molecules \cite{Revues} and provided a means
68: of testing calculations of molecular dynamics. It has also led to good
69: estimates of the s-wave scattering length \cite{Abraham,Gardner} that
70: determines the behavior of ultra-cold dilute gases near quantum
71: degeneracy.
72:
73: The case of $^4$He atoms in the metastable $2^3S_1$ state (He$^*$) is
74: distinctive in that each atom carries a large internal energy of 20 eV.
75: Photoassociation experiments with He$^*$ were first demonstrated by
76: Herschbach {\it et al.} with atoms trapped in a magneto-optical trap
77: (MOT) \cite{Herschbach}. However, the quantitative study of pair
78: interactions has still to be completed. In particular, although Bose
79: Einstein Condensation (BEC) has been achieved in He$^*$
80: \cite{Pereira101,Robert}, the scattering length remains uncertain. What
81: is more, the accurate investigation of collisional properties
82: \cite{Sirjean} and of the dynamical behavior \cite{Leduc} of the
83: ultra-cold He$^*$ gas suffers from the uncertainty in the scattering
84: length. In order to extract quantitative information from PA
85: spectroscopy we have performed a new PA experiment starting from a
86: magnetically trapped and evaporatively cooled metastable helium gas. We
87: have thereby achieved greater state selectivity, higher density, and
88: lower temperature than were obtained previously \cite{Herschbach}.
89:
90: As a preliminary step toward the characterization of pair interactions,
91: we have reported \cite{Leonard} the observation of purely long-range
92: helium dimers produced by photoassociation of metastable helium atoms,
93: with the PA laser tuned close to the $2^3S_1 \leftrightarrow 2^3P_0$
94: atomic line (see Figure \ref{Fig: Principe_PA}). The novelty of these
95: dimers is that they are produced from two highly excited atoms and
96: therefore carry a huge internal energy of 40 eV. However, whereas one
97: might expect the molecules to decay through autoionization, the primary
98: decay mechanism is radiative. This fact allowed us to develop an
99: original, ``calorimetric'' detection method based on the strong heating
100: of the atomic cloud at resonant PA frequencies. Our preliminary model
101: for the heating accounts for the conversion of a decaying molecule's
102: vibrational kinetic energy into additional thermal energy within the
103: cloud. Autoionization appears to have a negligible effect, probably
104: because the inner turning points for these giant dimers are so far
105: apart (around 150 bohr radii). Ionization is unlikely at such
106: distances, so it is not surprising that these molecular states have not
107: been observed with the ion detectors used in MOT experiments
108: \cite{Herschbach}.
109:
110:
111: \begin{figure}[t]
112: \begin{center}
113: \includegraphics[height=7cm]{Principe_PA.eps}
114: \end{center}
115: \caption{\footnotesize a) Illustration of the principle of a
116: photoassociation (PA) experiment. A free pair of metastable atoms is
117: resonantly excited into a purely long-range $0_u^+$ molecular bound
118: state. The potential curve for the $^5\Sigma_g^+$ state is the one
119: given by \cite{StarckMeyer}, the $0_u^+$ is the one obtained by the
120: calculation described in the text. Note the change in energy and length
121: scales between the $^5\Sigma_g^+$ and the purely long-range $0_u^+$
122: potential wells.}\label{Fig: Principe_PA}
123: \end{figure}
124:
125: The present paper is meant to provide a theoretical complement to
126: reference \cite{Leonard}, which focused primarily on experimental
127: methods and results. Because $^4$He has no hyperfine structure, the
128: theoretical approach is relatively simple as compared with alkali
129: systems. Thus, giant helium dimers present an interesting case study,
130: and we have attempted to emphasize important physical concepts in
131: somewhat of a tutorial approach. In particular, a perturbative
132: description of the electronic potentials is given, which provides a
133: physical understanding of the formation of these molecules. Then, with
134: a single-channel adiabatic calculation of the effective molecular
135: potentials we find purely long-range spectra that are in excellent
136: agreement with those computed in \cite{Venturi} by more sophisticated
137: techniques.
138:
139: In Section \ref{section: shifts}, after a brief review of the
140: experiment, we relate the molecular binding energy to the measured
141: resonance frequency by subtracting shifts due to the magnetic trapping
142: potential and the non-zero temperature of the atomic cloud. In
143: particular, the free-bound character of the transitions leads to
144: temperature-induced shifts which do not exist in the case of
145: bound-bound transitions. Section \ref{section: RoVib structure}
146: describes the calculation of the long-range part of the $2^3S$ - $2^3P$
147: molecular interaction potentials, as well as the theoretical values for
148: the binding energies of the giant dimers. Our perturbative approach
149: shows how purely long-range potential wells arise from the competition
150: between the dipole-dipole interaction and the atomic fine structure.
151: Finally, we compare both the experimental and theoretical
152: determinations of the binding energies. With its high accuracy, the
153: experiment provides a clear illustration of retardation effects in the
154: electromagnetic interaction and of tiny corrections due to the
155: vibration-induced coupling between electronic and nuclear degrees of
156: freedom. Moreover, it yields a measurement of the radiative decay rate
157: $\Gamma$ of the atomic excited state $2^3P$ with an accuracy of
158: $0.2\%$.
159:
160: \section{Deriving the binding energies from PA measurements}
161: \label{section: shifts}
162:
163: \subsection{Acquisition of PA spectra}
164:
165: \begin{figure}[t]
166: \begin{center}
167: \includegraphics[height=6.5cm]{Data_v4.eps}
168: \end{center}
169: \caption{\footnotesize Detection of the resonant formation of giant
170: dimers in the $v=4$ vibrational state of the $0_u^+$ potential well.
171: After the PA laser pulse and further thermalization, the remaining
172: atoms are detected optically: a) atom number, b) temperature in $\mu$K
173: and c) peak optical density versus the PA laser detuning from the
174: atomic $D_0$ line. Each point represents a new evaporated cloud after
175: PA pulse illumination, thermalization and ballistic expansion. The
176: curves in graphs a) and c) indicate the averaging of data over 5
177: adjacent points. The curve in graph b) is a Lorentzian fit to the data
178: with a width of 2.8 MHz. Strong heating of the atomic cloud is
179: observed when the PA laser is resonant with a molecular transition.}
180: \label{Fig: Data_v4}
181: \end{figure}
182:
183: We perform PA experiments with a cold metastable helium gas confined in
184: a magnetic trap. The atomic cloud is cooled by RF-induced evaporation
185: to a temperature in the $\mu$K range, just above the BEC transition
186: \cite{Pereira2}. The cloud is illuminated for a short period (0.1 to
187: 10 ms) by a low-intensity PA laser beam and then allowed to thermalize
188: for a few hundred ms. It is subsequently released and then detected
189: optically after a few-ms expansion time. Giant helium dimers are
190: produced when a free (unbound) pair of cold atoms absorbs a PA photon
191: and is excited into a bound state of the purely long-range potential.
192: This free-bound transition occurs when the PA laser is tuned red of the
193: $2^3S_1 \leftrightarrow 2^3P_0$ ($D_0$) atomic line (see Figure
194: \ref{Fig: Principe_PA}). Several resonance lines appear in the
195: recorded temperature data, indicating that the formation of transient
196: molecules results in the deposition of energy in the surrounding atomic
197: cloud. Figure \ref{Fig: Data_v4} illustrates the typical data obtained
198: when we tune the PA laser in the vicinity of a bound state in the
199: $0_u^+$ potential well. Although few atoms are lost (Figure \ref{Fig:
200: Data_v4}-a), a strong increase in temperature (Figure \ref{Fig:
201: Data_v4}-b) and consequently a strong decrease in peak optical density
202: (Figure \ref{Fig: Data_v4}-c) are monitored. Since the cloud is very
203: cold (typically $5 \mu$K), the excitation of relatively few molecules
204: is enough to cause significant heating. Thus, the atomic cloud serves
205: as a sensitive calorimeter capable of detecting the position of the
206: molecular lines with an accuracy of 0.5 MHz. The quantitative study of
207: the heating mechanism is in progress and will be published in a
208: separate paper.
209:
210: \subsection{Discussion of the various line shift mechanisms}
211:
212:
213: Acquiring experimental spectra consists in measuring the PA laser
214: detunings at which molecular lines are resonantly excited in the
215: magnetically trapped atomic cloud. For an accurate
216: interpretation of the data, we need to take into account the correct lineshape
217: function, which may include shifts and/or asymmetric broadening due
218: to various mechanisms. We do so on the basis of the following calculation
219: of the molecular binding energy, which emerges straightforwardly from the
220: conservation of energy and momentum.
221:
222: \subsubsection{Conservation of energy for a free-bound transition}
223:
224: The energy $E_{i}$ of a pair of trapped atoms in the initial unbound state
225: can be written:
226: \begin{equation}\label{eq: Initial Energy}
227: E_{i}(\vec{r_1},\vec{r_2},\vec{P},\vec{p}_{rel})=\frac{\vec{P}^{\;2}}{4m}+\frac{
228: \vec{p}_{rel}^{\;2}}{m} - \vec{\mu}\cdot\left( {\vec{B}(\vec{r_1}) +
229: \vec{B}(\vec{r_2})} \right),
230: \end{equation} where $m$ is the mass of the He atom, $\vec{P}=\vec{p_1}+\vec{p_2}$
231: is the momentum of the pair's center of mass, $ \vec{p}_{rel}
232: =(\vec{p_1}-\vec{p_2})/2$ is the relative momentum,
233: ${\vec{B}(\vec{r_1})}$ and ${\vec{B}(\vec{r_2})}$ are the magnetic
234: field at the location of each atom, and ${\vec{\mu}}$ is the magnetic
235: dipole moment of an atom in the $2^3S_1$ state (the Land\'e factor
236: being 2, we define $\mu =- 2 \mu_B$, with the Bohr magneton $\mu_B<0$).
237: In expression (\ref{eq: Initial Energy}), we neglect any interaction
238: energy between the two atoms. This will be justified below.
239:
240: After the pair of atoms absorbs a photon with momentum $\hbar\vec{k}$
241: and frequency $\nu_L$, the binding energy $hb<0$ of the resulting
242: molecule can be inferred from the conservation law for energy and
243: momentum:
244: \begin{eqnarray}\label{eq:conservation}
245: E_i(\vec{r_1},\vec{r_2},\vec{P},\vec{p}_{rel})+ h \nu_L & =&
246: \frac{\vec{P}_M^{\;2}}{4m} + h(\nu_0+b)\ \\
247: \mbox{with }\ \vec{P} +\hbar \vec{k}& =& \vec{P}_M\ ,\nonumber
248: \end{eqnarray}
249: where $\vec{P}_M$ is the final momentum of the molecule and $h\nu_0$ is
250: the energy of the $D_0$ line (for an isolated atom in a zero magnetic
251: field). The difference between the molecular binding energy and the PA
252: laser detuning $\delta=\nu_L-\nu_0<0$ is thus given by:
253: \begin{eqnarray} \label{eq:binging energy}
254: h(b-\delta) &=& - \hbar \vec{k}\cdot
255: \frac{\vec{P}}{2m} - \frac{\hbar^2k^2} {4m} \nonumber\\
256: &&- \vec{\mu}\cdot\left( {\vec{B}(\vec{r_1}) + \vec{B}(\vec{r_2})}
257: \right) + \frac{ \vec{p}_{rel}^{\;2}}{m} ,
258: \end{eqnarray}
259: Any dependence of the molecular level energy on the magnetic field
260: (Zeeman effect) or on the density (mean field interaction of the
261: molecule with the surrounding atomic and/or molecular cloud) is {\it a
262: priori} included in $b$, which may therefore also depend on the
263: position of the molecule.
264:
265: Note that the relative kinetic energy term $\vec{p}_{rel}^{\;2}/m$ in
266: Equation (\ref{eq:binging energy}) would not appear in the case of a
267: bound-bound transition, since it would be implicitly included in the
268: initial binding energy. As it is always positive, it contributes an
269: {\it asymmetric} lineshape, and consequently a mean shift
270: \cite{Napolitano}. Also, the harmonic magnetic trapping potential
271: contains quadratic terms which contribute to the inhomogeneous,
272: asymmetric broadening and shift of the lines. However, the temperature
273: is low enough that the asymmetric broadening terms remain much smaller
274: than the natural lorentzian width. Thus, the only effect is a shift of
275: the peak position of the lines, which can be calculated by averaging
276: Equation (\ref{eq:binging energy}) over the distribution function for
277: the initial pair of free atoms.
278:
279:
280: \subsubsection{Initial distribution function of the free pair}
281:
282: The distribution function for the pairs {\it that undergo the PA
283: transition} is the thermal distribution for a pair of trapped atoms
284: {\it multiplied} by the transition probability. According to the
285: Franck-Condon overlap principle, the latter is proportional to the
286: square of the overlap between the initial and final radial wave
287: functions. Since the excited state is a bound state, the overlap is
288: peaked at the Condon radius $R_C$ close to the classical outer turning
289: point. According to Table \ref{tab: results} in Section \ref{section:
290: RoVib structure}, the transition occurs mainly for an internuclear
291: distance $R_C=||\vec{r_1}-\vec{r_2}|| \lesssim 50 $ nm, which is much
292: smaller than the size of the atomic cloud ($\sim 100\ \mu$m at $T\sim
293: 10\ \mu$K). This allows us to use the approximation
294: $\vec{r_1}\simeq\vec{r_2}\simeq\vec{r}$ in Equations (\ref{eq: Initial
295: Energy}) and (\ref{eq:binging energy}), where $\vec{r}$ is the center
296: of mass of the pair. Furthermore, because the temperature is so low,
297: the collision between two atoms occurs in the s-wave scattering regime,
298: for which the relative angular momentum $\vec{p}_{rel}=\hbar \vec{q}$
299: has no component orthogonal to the internuclear axis. Thus, the
300: vectorial character of $\vec{p}_{rel}$ can be ignored, since there is
301: only one degree of freedom for the relative motion of the colliding
302: atoms. For internuclear distances $R$ close to $R_C$, the radial part
303: $u(R)$ of the ground state wave function can be approximated as
304: $u(R)\propto \sin(q(R-a))\propto q$ since $qR_C \ll 1$ (with $a$
305: representing the s-wave scattering length; see {\it e.g.}
306: \cite{Revues}). Finally, the distribution function for a pair of
307: trapped atoms in the s-wave scattering regime absorbing a PA photon is
308: found to be proportional to:
309: \begin{eqnarray}\label{eq:distrib}
310: q^2 \delta(\vec{r_1}-\vec{r_2})\times
311: \exp(-E_{i}(\vec{r_1},\vec{r_2},\vec{P},q)/k_BT)\ .
312: \end{eqnarray}
313:
314:
315: \subsubsection{Mean frequency shifts}
316:
317: i)Average over the center-of-mass momentum.\\
318:
319: The first term in the right-hand side of Equation (\ref{eq:binging
320: energy}) is responsible for the Doppler profile. It produces no average
321: shift, since there is \textit{a priori} no correlation between the
322: momenta of the two atoms and of the photon: $\langle
323: \vec{k}.\vec{P}\rangle=0$. However, it is responsible for a symmetric
324: broadening of the lines, which scales like $\sqrt{T}$ ($T$, the
325: temperature of the cold gas). In the microK range of temperature, this
326: Doppler broadening turns out to be small compared with the natural
327: lifetime broadening of the molecular states probed.
328:
329: The second term in Equation (\ref{eq:binging energy}) is the recoil
330: energy of the molecule after absorbing the photon. In units of $h$, its
331: numerical value is $\sim 21$ kHz, which is well below our experimental
332: accuracy. Therefore we neglect the corresponding shift.\\
333:
334: ii) Average over the center of mass position.\\
335:
336: Using expression (\ref{eq:distrib}), the average over the positions
337: $\vec{r_1}$ and $\vec{r_2}$ turns out to be an average over the
338: position $\vec{r}$ of the center of mass of the pair. The shift induced
339: by the external trapping potential is thus calculated to be:
340: \begin{equation}
341: \left\langle - \vec{\mu}\cdot\left( {\vec{B}(\vec{r_1}) +
342: \vec{B}(\vec{r_2})} \right) \right\rangle = 2 \mu B_0 +
343: \frac{3}{2}k_BT,
344: \end{equation}
345: where $2 \mu B_0$ is twice the Zeeman shift of one atom at the center
346: of the trap, and $3 k_BT/2$ is the average of the harmonic trapping
347: potential energy, according to the equipartition theorem for quadratic
348: energy terms.
349:
350: As already noted, the binding energy $hb$ {\it a priori} also depends
351: on the center of mass position, and should therefore be averaged as
352: well. However, we neglect this position dependence, since the effect of
353: both the inhomogeneous magnetic field (molecular Zeeman effect) and
354: density (atom-molecule interaction) turn out to be small compared with
355: our experimental accuracy, as discussed below.
356: \\
357:
358: iii) Average over the relative momentum.\\
359:
360: Making use of expression (\ref{eq:distrib}), we find the average of the
361: relative kinetic energy term:
362: \begin{eqnarray}
363: \left\langle \frac{\hbar^2 q^2}{m} \right\rangle =
364: \frac{\int\frac{\hbar^2 q^2}{m}\ q^2 \exp (-\frac{\hbar^2
365: q^2}{mk_BT})\;dq}{\int q^2 \exp (-\frac{\hbar^2
366: q^2}{mk_BT})\;dq}=\frac{3}{2}k_BT,
367: \end{eqnarray}
368: where the denominator normalizes the distribution function. Let us
369: mention that while there is only one degree of freedom for the relative
370: momentum (in the s-wave scattering regime), our inclusion of the pair
371: distribution function leads us coincidentally to the same $3k_BT/2$
372: that one finds when treating three classical degrees of freedom.
373: \\
374:
375:
376: iv) Other shift mechanisms.\\
377:
378: The mean-field interaction due to the surrounding medium on both the
379: initial and final states of the transition can cause density-dependent
380: shifts of the lines. As far as the initial pair of free atoms is
381: concerned, the mean field interaction energy is $4\pi\hbar^2\times
382: na/m$, where the atomic density $n<10^{14}$ cm$^{-3}$, the s-wave
383: scattering length $a<20$ nm \cite{Pereira101,Robert}, and $m\sim
384: 6.68\times 10^{-27}$ kg. In units of $h$, the upper bound for this
385: mean-field interaction is less than $\sim 60$ kHz, which is below our
386: experimental accuracy and therefore negligible. The mean field energy
387: shift of the final molecular state, which would appear as a
388: density-dependent term in the experimental binding energy, has not been
389: detected experimentally.
390:
391: Finally, light-induced line shifts are completely negligible, since the
392: spectra were measured with PA laser intensities well below the atomic
393: saturation intensity.
394: \\
395:
396: v) Summary.\\
397:
398: In our experiment, each molecular line produces a resonant increase in
399: temperature as a function of PA detuning $\delta$. Each resonance line
400: is fit by a Lorentzian. The fit's center frequency $\delta_v$ is taken
401: to be the resonant frequency for excitation to vibrational level $v$.
402: Accounting for the corrections described above, we infer the molecular
403: binding energy $hb_v$ of this vibrational level to be:
404: \begin{equation}
405: h b_v \simeq h \delta_v + 2\mu B_0 + 3k_BT. \label{eq:e3}
406: \end{equation}
407:
408: \subsection{Experimental checks for the lineshifts}
409:
410: We have measured $\delta_v$, $B_0$ and $T$ for the lines $v=0$ through
411: $v=4$ in the $0_u^+$ potential well, for $B_0=0.1$ to $\sim10 $ Gauss
412: and for $T= 1.5$ to 30 $\mu$K. The temperature of the gas was varied by
413: changing the final RF frequency of the evaporation ramp above the
414: critical temperature. Consequently, the atomic density was also varied
415: from $n\sim0.5\times 10^{13}$ to $\sim 8 \times 10^{13}$ at$/$cm$^{3}$.
416:
417:
418: \begin{figure}[b]
419: \begin{center}
420: \includegraphics[height=6cm]{B0_dependence.eps}
421: \end{center}
422: \caption{\footnotesize Experimental determination of the binding
423: energy in the $0_u^+$ potential well for the vibrational level $v=3$:
424: illustration of the dependence of the measured detuning $\delta_v$ on
425: the magnetic field $B_0$, after correction from the temperature-induced
426: shift (see Equation (\ref{eq:e3})). The slope of the linear fit is
427: compatible with the expected dependence in $B_0$ (see in the text).}
428: \label{Fig: B0_dependence}
429: \end{figure}
430:
431:
432: In Equation \ref{eq:e3} the most important correction is due to $B_0$.
433: Figure \ref{Fig: B0_dependence} shows the dependence on $B_0$ of the
434: measured detuning $\delta_v$ of the $v=3$ line, after it is corrected
435: for the temperature-induced effect ($3k_BT$). If the magnetic field is
436: measured in units of $\mu B_0$, a linear fit to the data gives a slope
437: of $-2.02\pm0.02$. Given Equation (\ref{eq:e3}), the contribution of
438: the initial pair of free cold atoms (the ``ground" state), should be
439: exactly $-2 \mu B_0$. A deviation from this value could be attributed
440: to the contribution of the mean Zeeman effect of the molecular bound
441: (``excited") state. As the $0_u^+$ electronic state is non degenerate,
442: the molecule cannot have any magnetic dipole moment except one induced
443: by the molecular rotation, which is expected to be of the order of the
444: nuclear magneton, or about three orders of magnitude smaller than
445: $\mu_B$. Given the experimental accuracy and the range of magnetic
446: field explored, the correspondingly small Zeeman effect would be
447: difficult to measure. But our data permit us to set an upper bound of
448: $0.02\; \mu = 0.04\;|\mu_B|$ on the molecular magnetic dipole moment.
449: This result justifies neglecting the molecular Zeeman effect in the
450: calculation of the mean line shifts.
451:
452:
453:
454: \begin{figure}[b]
455: \begin{center}
456: \includegraphics[height=6cm]{T_n_dependence.eps}
457: \end{center}
458: \caption{\footnotesize Experimental determination of the binding
459: energies in the $0_u^+$ potential well: illustration, in the case of
460: the vibrational level $v=4$, of the dependence of the measured detuning
461: $\delta_v$ on the temperature and on the density, after correction from
462: the magnetically-induced shift (see Equation (\ref{eq:e3})). Data are
463: displayed before (squares) and after (circles) applying the
464: temperature-dependent correction. Error bars include uncertainty in the
465: measurements of $\delta$, $B_0$, and $T$.} \label{Fig: T_n_dependence}
466: \end{figure}
467:
468: Figure \ref{Fig: T_n_dependence} displays the measured position of the
469: $v=4$ line, corrected for the magnetically-induced shift ($2\mu B_0$),
470: as function of the atomic cloud density. Data with (circles) and
471: without (squares) the additional temperature-dependent correction are
472: shown. The uncorrected data has been displayed in order to illustrate
473: the importance of the temperature effect (up to 2 MHz at $\sim 30\
474: \mu$K) as compared to the experimental accuray (0.5 MHz). For this set
475: of data, the density was increased simply by further evaporative
476: cooling of the gas. Thus, higher density is associated with lower
477: temperature, and the temperature-induced shift indicated by the squares
478: nearly vanishes for large density. It should be noted here that the
479: size of the molecules ($917\ a_0 \sim 50$ nm, see Table \ref{tab:
480: results}) is not vanishingly small compared with mean inter-atomic
481: distance in the cloud ($\sim 260$ nm at $6 \times 10^{13}$
482: at/cm$^{3}$). Under these conditions, one might expect to find a
483: density- dependent shift due to the mean field interaction between the
484: molecule and the surrounding atomic medium. However, no such shift is
485: evident in our data after we apply the corrections for temperature and
486: magnetic field. The error bars include experimental uncertainty in
487: $\delta$, $B_0$ and $T$. Additional scatter of about $ 0.3$ MHz can be
488: attributed to the uncertainty in the PA laser frequency lock. We have
489: studied the stability of the experiment and the possible sources of
490: systematic error in all achievable parameter ranges (accumulating many
491: more data than are shown in Figure \ref{Fig: T_n_dependence}). We
492: conclude that the binding energy for $v=4$ is $-18.2\pm 0.5$ MHz, in
493: units of $h$.
494:
495: Finally, from Figure \ref{Fig: T_n_dependence} and from the 0.5 MHz
496: uncertainty, we can infer that the density-induced energy shift of the
497: molecules must be smaller than $\sim 100$ kHz per $10^{13}$ cm$^{-3}$
498: of density. Actually, the atomic Bose gas surrounding the molecule is
499: near resonance and therefore has a permittivity that differs from the
500: vacuum value. For an ideally homogeneous medium, the permittivity would
501: enter in the resonant dipole potential \cite{Dip-Dip_in_medium},
502: leading to a density-dependent term in the binding energy which would
503: be at least a factor two above our upper bound. Since we do not detect
504: this effect, we conclude our gas can not be considered as an
505: homogeneous medium on the size scale of a molecule. This point may
506: deliver important information about the three-particle correlation
507: function in the atomic gas and would require further study, but it has
508: not been investigated so far.
509:
510: Similar data were registered for the other vibrational lines that we
511: were able to measure. The experimental results for the binding energies
512: are reported in Table \ref{tab: comparison}, Section \ref{section:
513: RoVib structure}.
514:
515:
516: \section{Ro-vibrational structure of the giant dimers}
517: \label{section: RoVib structure}
518:
519: In order to interpret the measurements described above, we now develop
520: the calculation of the long-range interaction of one atom in the
521: $2^3S_1$ state, and another one in the $2^3P_{J=0,1,2}$ state. It
522: happens that some of the resulting potential energy curves have minima
523: at very large internuclear distance and support purely long-range bound
524: states. In particular, the calculated spectrum of five vibrational
525: states in the $0_u^+$ potential will be shown to be in excellent
526: agreement with our measurements.
527:
528: \subsection{Electronic potential curves for the $2^3S+2^3P$ system with fixed nuclei}
529: \label{section: adiabatic potentials}
530:
531: \subsubsection{Hamiltonian}
532:
533: The general task for calculating molecular potentials in $^4$He
534: consists in solving the following Schr\"odinger equation
535: \cite{Lefebvre-Brion}:
536: \begin{eqnarray}
537: \label{Hamiltonian_General}
538: &&\hat{H}\ket{\psi_{\alpha}}=(\;\hat{T}_n+\hat{T}_e+\hat{V}+\hat{H}_{rel}\;)\ket{\psi_{\alpha}}=E_{\alpha}\ket{\psi_{\alpha}}\\
539: &&\mbox{where }\hat{T}_n=\displaystyle \sum_{k=1}^2
540: \frac{\hat{\textbf{p}}_\textbf{k}^2}{2M}\mbox{ , }\hat{T}_e
541: =\displaystyle \sum_{i=1}^4
542: \frac{\hat{\textbf{p}}_\textbf{i}^2}{2m}\mbox{ ,} \nonumber\\
543: &&\mbox{and }\hat{V}=\hat{V}(\hat{\textbf{r}}_k,\hat{\textbf{r}}_i)
544: \mbox{ , } \hat{H}_{rel} =
545: \hat{H}_{rel}(\hat{\textbf{r}}_i,\hat{\textbf{s}}_i).\nonumber~
546: \end{eqnarray}
547: Here, $\ket{\psi_{\alpha}}$ is a stationary solution corresponding to a
548: set of quantum numbers $\{\alpha\}$ to be detailed later. The
549: hamiltonian written above appears as the sum of four terms $\hat{T}_n$,
550: $\hat{T}_e$, $\hat{V}$ and $\hat{H}_{rel}$ which represent respectively
551: the kinetic energy of the two nuclei, the kinetic energy of the four
552: electrons, the non relativistic interaction between the six charged
553: particles, and the relativistic part of the hamiltonian. This operator
554: is written as function of the positions of the nuclei
555: $\hat{\textbf{r}}_k$, and of the electrons $\hat{\textbf{r}}_i$, and as
556: function of the spin coordinates $\hat{\textbf{s}}_i$ of the four
557: electrons. The $^4$He nuclei have no spin. To solve this very
558: complicated problem, we adopt a perturbative approach, in which we
559: consider the internuclear distance large enough that the interaction
560: potential $\hat{V}$ can be treated as a perturbation of the system of
561: two independent atoms $A$ and $B$. Thus the hamiltonian
562: (\ref{Hamiltonian_General}) is approximated as follows:
563: \begin{eqnarray}
564: \label{Hamiltonian_approx} \hat{H}=\;\hat{T}_n + \hat{H}_{0}(A) +
565: \hat{H}_{0}(B) + \hat{H}_{fs}(A) + \hat{H}_{fs}(B) + \hat{U}(R)
566: \end{eqnarray}
567: where $\hat{H}_{0}$ and $\hat{H}_{fs}$ are respectively the
568: non-relativistic and relativistic part of the hamiltonian for one
569: isolated atom, and $\hat{U}(R)$ stands for the long-range electrostatic
570: interaction between the two atoms, whose leading term is the retarded
571: dipole-dipole interaction.
572:
573: To describe long-range molecular interactions, we expand the molecular
574: state in linear combinations of (entangled) atomic states (LCAO
575: approximation). Moreover, according to the usual Born-Oppenheimer
576: approximation we first consider only the electronic degrees of freedom
577: while keeping the nuclei (more precisely, the atomic
578: centers of mass) fixed. We then treat both the dipole-dipole
579: interaction and the atomic fine structure as perturbations of the
580: non-relativistic hamiltonian for two independent atoms. We write the
581: two interactions in the basis set of states formed by the tensorial
582: product of isolated non relativistic atomic states: $\{\ket{\mbox{atom
583: } A: L_A,M_{LA};S_A,M_{SA}}\otimes \ket{\mbox{atom } B:
584: L_B,M_{LB};S_B,M_{SB}}\}$. Considering one atomic orbital in the $2^3S$
585: state and another one in the $2^3P$ state, the space of states is of
586: dimension 54. As the two nuclei are identical, the hamiltonian is
587: unchanged under the inversion $\hat{I}_e$ of all the electrons with
588: respect to the center of mass \cite{Herzberg}. The operator $\hat{I}_e$
589: commutes with the hamiltonian (\ref{Hamiltonian_approx}) and has two
590: eigenvalues $\omega=\pm 1$ with eigenstates labeled {\it gerade} ($g$)
591: and {\it ungerade} ($u$) respectively.
592:
593: \subsubsection{Retarded dipole-dipole interaction}
594: The dipole-dipole interaction $\hat{U}(R)$, first, only couples the
595: orbital angular momenta of the two independent non-relativistic atoms.
596: It is diagonal in the Hund's case (a) basis set labelled
597: $\ket{^{2S+1}\Lambda_{u/g}}$ (see e.g. \cite{Herzberg,Dashevskaya}).
598: These states can be written as follows in the atomic basis:
599: \begin{eqnarray}
600: \label{Base_du_cas_a}
601: \ket{^{2S+1}\Lambda_{u/g}}=\frac{1}{\sqrt{2}}\;(\;1+\omega\hat{I}_e\;)\ket{A:0,0;B:1,M_L}\otimes \ket{S,M_S}\nonumber~\\
602: =\frac{1}{\sqrt{2}}(\ket{A:0,0;B:1,M_L}-\omega(-1)^S\ket{A:1,M_L;B:0,0})\nonumber~\\
603: \otimes\ket{S,M_S}.\nonumber~
604: \end{eqnarray}
605: Here, $S$ is the total electronic spin of the molecule ($S=$0, 1 or 2),
606: $\Lambda$ is the projection onto the molecular axis of the electronic
607: orbital angular momentum of the molecule. In the Hund's case (a) basis,
608: the retarded dipole-dipole interaction is respectively given by
609: \cite{Dashevskaya,Meath}:
610: \begin{subequations}\label{eq:_Retarded_Dipole_All}
611: \begin{equation}\label{eq:_Retarded_Dipole_1}
612: -2\omega (-1)^S C_{3}/R^{3} \times\left(\cos(kR)+kR\sin(kR)\right),
613: \end{equation}
614: \begin{equation}\label{eq:_Retarded_Dipole_2}
615: \omega (-1)^S C_{3}/R^{3}
616: \times\left(\cos(kR)+kR\sin(kR)-(kR)^2\cos(kR)\right)\mbox{,}
617: \end{equation}
618: \end{subequations}
619: for $^{2S+1}\Sigma_{u/g}$ states (\ref{eq:_Retarded_Dipole_1}), and
620: $^{2S+1}\Pi_{u/g}$ states (\ref{eq:_Retarded_Dipole_2}). The
621: coefficient $C_3$ is related to the atomic dipole matrix element
622: $d=<2^3P|\hat{d_z}|2^3S>$, and thus to the radiative life time $1/
623: \Gamma$ of the atomic transition:
624: \begin{equation}\label{eq: C3}
625: C_3=\frac{|d|^2}{4 \pi \varepsilon_0}=\frac{3}{4}\; \hbar\Gamma \left(
626: \frac{\lambda}{2\pi}\right)^3,
627: \end{equation}
628: with $\varepsilon_0$ the vacuum permittivity. The fine structure
629: splitting is small enough that we assume the three atomic lines of
630: interest ($2^3S_1 \leftrightarrow 2^3P_{J=0,1,2}$) have the same
631: wavelength $\lambda = 1083.3$ nm within 0.1 nm. The radiative
632: decay rate $\Gamma = 2\pi \times 1.6248$ MHz can be inferred from
633: $\lambda^2$ and from an accurate calculation of the oscillator strength
634: of the atomic transition \cite{Drake}. Finally, $C_3$ is found to be
635: $C_3=6.405$ atomic units, within a relative uncertainty of $5 \times
636: 10^{-4}$.
637:
638: \subsubsection{Fine structure coupling}
639: We next consider the relativistic part of the hamiltonian,
640: $\hat{H}_{fs}(A) + \hat{H}_{fs}(B)$, which is diagonal in the Hund's
641: case (c) basis (by definition of Hund's case (c), see e.g.
642: \cite{Herzberg}), with three eigenvalues corresponding to the three
643: states $2^3S_1 + 2^3P_{J=0,1,2}$. The eigenstates can only be
644: characterized by the projection $\Omega$ of the total electronic
645: angular momentum (orbital and spin) on the molecular axis
646: \cite{Herzberg}. In $^4$He the atomic fine structure can be modeled
647: using the following operator:
648: \begin{eqnarray}
649: \hat{H}_{fs}=\alpha\vec{L}.\vec{S}+\beta(\vec{L}.\vec{S})^{2},
650: \end{eqnarray}
651: where $\vec{L}$ and $\vec{S}$ are the atomic orbital and spin angular
652: momenta. In addition to the usual spin-orbit coupling, spin-spin
653: magnetic dipole interaction between the two electrons is an important
654: effect in helium \cite{Bethe_Salpeter}, leading to a non equidistant
655: splitting of the fine structure levels. In our model, the constants
656: $\alpha$ and $\beta$ are determined phenomenologically, in order to
657: reproduce the fine structure splittings which have been measured
658: \cite{FineStructSplit1,FineStructSplit2} very accurately:
659: \begin{equation}
660: \alpha=-\frac{\Delta_{J=2\leftrightarrow1}}{2\hbar^2}\ \ \ \mbox{ and
661: }\ \ \ \beta=\frac{2 \Delta_{J=1\leftrightarrow0} -
662: \Delta_{J=2\leftrightarrow1}} {6\hbar^4}\ \ , \nonumber \end{equation}
663: \begin{equation}\mbox{with }\left\{
664: \begin{tabular}{l}
665: $\Delta_{J=2\leftrightarrow1} = h \times 2.291175 $ GHz\\
666: $\Delta_{J=1\leftrightarrow0} = h \times 29.616950$ GHz
667: \end{tabular}
668: \right.\nonumber
669: \end{equation}
670:
671:
672: \begin{figure}[t]
673: \begin{center}
674: \includegraphics[height=13cm]{Non_Rotating_ungerade.eps}
675: \end{center}
676: \caption{\footnotesize Ungerade electronic potential curves (in GHz)
677: for fixed nuclei for the $2^3S+2^3P$ system versus the internuclear
678: distance $R$ (in atomic units; 1 $a_0 \sim 0.0529$ nm). The potential
679: curves result from the numerical diagonalization of the hamiltonian
680: (\ref{eq: non_rotating_electronic_hamiltonian}). Three arrows indicate
681: the three purely long-range potential wells in which bound states are
682: determined numerically.} \label{fig:Non_Rotating}
683: \end{figure}
684:
685:
686:
687: \subsubsection{Potential curves with fixed nuclei}
688:
689: According to the Movre-Pichler approach \cite{MovrePichler}, both
690: retarded dipole-dipole interaction and atomic fine structure coupling:
691: \begin{eqnarray}\label{eq: non_rotating_electronic_hamiltonian}
692: \hat{H}_{fs}(A) + \hat{H}_{fs}(B) + \hat{U}(R)
693: \end{eqnarray}
694: should be considered simultaneously as a perturbation of the non
695: relativistic hamiltonian for two independent atoms $\hat{H}_{0}(A) +
696: \hat{H}_{0}(B)$. Only the projection $\Omega$ of the total electronic
697: angular momentum on the molecular axis is a good quantum number. States
698: of different $u/g$ symmetry are uncoupled and two sets of potential
699: curves can be determined independently for {\it gerade} and {\it
700: ungerade} states. Since we do photoassociation experiments in a
701: magnetically trapped atomic cloud, the initial quasi-molecular state is
702: $^5\Sigma_g^+$, and {\it gerade} states are not accessible by
703: single-photon excitation. Thus we focus only on {\it ungerade} states.
704: Figure \ref{fig:Non_Rotating} shows the results of the calculated {\it
705: ungerade} eigenvalues of the operator (\ref{eq:
706: non_rotating_electronic_hamiltonian}) as a function of $R$. Here, the
707: electronic states are determined with fixed nuclei. Also, the potential
708: curves describe only the long-range part of the molecular interactions
709: as a consequence of the perturbative description. For the $\Omega=0$
710: space, the reflection symmetry (in a plane containing the molecular
711: axis) leads to a relevant additional label $+/-$, which distinguishes
712: two states with different energies. For $\Omega\neq 0$ states, this
713: symmetry can be defined as well but the two resulting states have the
714: same energy.
715:
716:
717: \begin{figure*}[t]
718: \begin{center}
719: \includegraphics[height=6cm]{Ilustration_AntiCrossings.eps}
720: \end{center}
721: \caption{\footnotesize Eigenvalues of the restriction of the
722: hamiltonian (\ref{eq: non_rotating_electronic_hamiltonian}) to the
723: $0_u^+$ subspace. Energies are given in GHz, distances are in atomic
724: units. a) The fine structure coupling is neglected: the eigenstates are
725: pure Hund's case(a) states. b) The fine structure coupling is partly
726: included: couplings between the repulsive state and the attractive ones
727: are neglected. After diagonalization, one attractive and one repulsive
728: states cross each other. c) Finally, including all the fine structure
729: coupling terms leads to an anti-crossing and a purely long-range
730: potential well. Note that graph b) is only for illustration and that
731: the neglected terms are not small.} \label{fig:
732: Ilustration_AntiCrossings}
733: \end{figure*}
734:
735:
736:
737: \subsubsection{Physical origin of the purely long-range molecules}
738:
739:
740: The hamiltonian (\ref{eq: non_rotating_electronic_hamiltonian}) is
741: block diagonal with each block corresponding to a given
742: $\Omega_{u/g}^{(\pm)}$ subspace. As an example, let us consider the
743: subspace $0_u^+$. It is of dimension four. Figure \ref{fig:
744: Ilustration_AntiCrossings} illustrates the physical reason why a purely
745: long-range well arises in this subspace of states. If we consider only
746: the dipole-dipole interaction, one eigenvalue is purely repulsive,
747: while the three others are purely attractive, two of them being
748: identical (Figure \ref{fig: Ilustration_AntiCrossings}-a). They all
749: have the same asymptote. The four corresponding eigenstates are pure
750: Hund's case (a) states. Let us consider separately the repulsive state
751: and the manifold of attractive states. If we ``turn on" the fine
752: structure coupling inside each of these two subspaces of states, while
753: neglecting the couplings between them, then the potential curves repel
754: each other and the asymptotes no longer coincide. Of course, since the
755: neglected couplings are not small, the four asymptotes have no
756: straightforward physical meaning. However, the important point is that
757: a crossing shows up between the repulsive curve and one attractive
758: curve (Figure \ref{fig: Ilustration_AntiCrossings}-b). Finally, if we
759: turn on the neglected fine structure terms, we couple the subspaces
760: corresponding to the two crossing states, and an anti-crossing appears
761: (Figure \ref{fig: Ilustration_AntiCrossings}-c). The resulting
762: potential well is thus a consequence of the fine-structure mixing of
763: long-range molecular interactions, which links the inner, repulsive
764: dipole-dipole curve with an outer, attractive one. What is remarkable
765: about this well is that {\it even the repulsive part occurs at very
766: long-range}, in a region where the asymptotic dipole-dipole expression
767: remains a very good approximation. That is why the perturbative
768: approach used here is very well suited to describe the bound states
769: lying in this kind of well, or the so-called purely long-range
770: molecular states \cite{Stwalley}.
771:
772: \begin{figure}[b]
773: \begin{center}
774: \includegraphics[height=6cm]{Projections_Ouplus.eps}
775: \end{center}
776: \caption{\footnotesize Eigenstate for the $0_u^+$ purely long-range
777: potential well connected to the $2^3S+2^3P_0$ asymptote within the
778: fixed-nuclei approximation. The electronic state is given with its
779: decomposition in the Hund's case (a) basis set: the weights are the
780: squares of the projection on the different subspaces of Hund's case (a)
781: states. Distances are in atomic units.} \label{fig: Projections_Ouplus}
782: \end{figure}
783:
784:
785: Due to the competition between the dipole-dipole and the fine structure
786: interactions, not only the potential curves, but also the electronic
787: states explicitly depend on $R$. As an illustration, the $0_u^+$ purely
788: long-range electronic eigenstate is shown in Figure \ref{fig:
789: Projections_Ouplus}. The eigenstate is given with its projections over
790: the Hund's case (a) basis set. It evolves from the pure Hund's case (a)
791: $^5\Pi_u$ at short range, where the dipole-dipole interaction
792: dominates, to a pure Hund's case (c) for asymptotically large values of
793: $R$ where the dipole-dipole interaction vanishes like $1/R^3$.
794: Consequently, the fixed-nuclei approximation must be corrected by an
795: accounting of the coupling between the electronic and nuclear degrees
796: of freedom.
797: \\
798:
799:
800: The discussion just presented can also be applied to all the other
801: $\Omega_{u/g}^{(\pm)}$ subspaces. Figure \ref{fig:Non_Rotating} shows
802: three purely long-range {\it ungerade} potential wells. One is
803: connected to the $2^3S_1+2^3P_0$ asymptote and belongs to the $0_u^+$
804: subspace; it has been presented above. The two others are connected to
805: the $2^3S_1+2^3P_1$ asymptote and belong to the $0_u^-$ and $2_u$
806: subspaces. Within the fixed nuclei approximation the calculated $0_u^+$
807: well is $2.130$ GHz deep, the $2_u$ one is 0.321 GHz deep, and the
808: $0_u^-$ one is 0.054 GHz deep. We will examine these wells more closely
809: in the following discussion.
810:
811: \subsection{Description of the motion of the nuclei}
812:
813: So far the dynamics of the electrons has been treated independently
814: from the dynamics of the nuclei. In our perturbative model, the
815: coupling between the two comes from the kinetic energy operator for the
816: relative motion of the nuclei:
817: \begin{eqnarray}\label{eq: nuclear_kinetic_energy}
818: \hat{T}_n(R,\theta,\varphi)=-\frac{\hbar^2}{2\mu}\left(
819: \frac{1}{R}\frac{\partial^2}{\partial R^2}R -
820: \frac{\vec{\ell}^{\;2}}{\hbar^2R^2}\right)\cdot
821: \end{eqnarray}
822: In this expression $(R,\theta,\varphi)$ are the spherical coordinates
823: of the fictitious particle of reduced mass $\mu$ associated with the
824: pair of nuclei, and $\vec{\ell}$ is the orbital angular momentum associated
825: with its rotation.
826:
827: \subsubsection{Effect of the rotation}
828: \label{parag: Effect of rotation}
829:
830: \begin{figure}[t]
831: \begin{center}
832: \includegraphics[height=13cm]{Effect_Of_Rotation_b.eps}
833: \end{center}
834: \caption{\footnotesize Influence of the nuclear rotation on the
835: electronic potential energy for the three {\it ungerade} purely
836: long-range wells shown in Figure \ref{fig:Non_Rotating}. The dotted
837: lines are the result of the fixed nuclei approximation. The full lines
838: are the potential used to calculate the binding energies presented in
839: Table \ref{tab: results}. Note that the horizontal and vertical scales
840: are different for each graph.} \label{fig: Effect_Of_Rotation}
841: \end{figure}
842:
843:
844: First, the effect of the rotation of the nuclei on the electronic
845: states calculated above can be found if we add the last term of
846: (\ref{eq: nuclear_kinetic_energy}) to the hamiltonian (\ref{eq:
847: non_rotating_electronic_hamiltonian}). The operator to be diagonalized
848: becomes:
849: \begin{eqnarray}\label{eq: Rotating_electonic_states}
850: \hat{\mathcal{H}}=\hat{H}_{fs}(A) + \hat{H}_{fs}(B) + \hat{U}(R) +
851: \frac{\vec{\ell}^{\;2}}{2\mu R^2}
852: \end{eqnarray}
853: Now, the space of states has to be extended to the rotational degrees
854: of freedom. Only the total angular momentum
855: $\vec{J}=\vec{L}+\vec{S}+\vec{\ell}$ has to be conserved
856: \footnote{Here, $\vec{L}$ and $\vec{S}$ represent the {\it molecular}
857: orbital and spin angular momenta.}, so we must consider the set of
858: states $\ket{\phi_{J,\Omega_{u}^{\pm}}}$ defined by the product of
859: electronic states determined above $\ket{\Omega_{u}^{(\pm)}}$ and of
860: rotational states $\ket{J,M,\Omega}$ \cite{Hougen} :
861: $\ket{\phi_{J,\Omega_{u}^{\pm}}} =\ket{\Omega_{u}^{(\pm)}} \otimes
862: \ket{J,M,\Omega}$. The quantum number $M$ is the projection of
863: $\vec{J}$ onto a lab-fixed frame. Since the molecule is linear,
864: $\vec{\ell}$ is orthogonal to the molecular axis, which means
865: $\ell_z=0$ and $J_z=L_z+S_z$. Thus the electronic quantum number
866: $\Omega$ represents the projection of $\vec{J}$ onto the molecular axis
867: and it is recalled as a parameter in the notation for the rotational
868: state. In this basis, $\vec{\ell}$ can be written as
869: $\vec{\ell}=\vec{J}-\vec{L}-\vec{S}$, the square of which is given by:
870: \begin{eqnarray}\label{eq:Detail l2}
871: \hat{\vec{\ell}}^{\;2}&=&\hat{\textbf{J}}^2+\hat{\textbf{L}}^2+\hat{\textbf{S}}^2-2\;\hat{J_z}^2+2\;\hat{L_z}\hat{S_z} +\;(\hat{L}_+\hat{S}_-+\hat{L}_-\hat{S}_+) \nonumber~\\
872: &\;&-\;(\hat{J}_+\hat{L}_-+\hat{J}_-\hat{L}_+)-(\hat{J}_+\hat{S}_-+\hat{J}_-\hat{S}_+)
873: \end{eqnarray}
874:
875: In Equation (\ref{eq:Detail l2}), the first line contains terms that
876: couple electronic states with each other {\it inside} each
877: $\Omega_{u}^{(\pm)}$ block. The second line contains the terms that
878: couple states belonging to different $\Omega$ subspaces, due to the
879: action of $\hat{J}_{\pm}$ which obeys {\it anomalous} commutation rules
880: \cite{Zare} and couples $\Omega$ to $\Omega \mp 1$. These off-diagonal
881: coupling terms become important where potential curves belonging to
882: different $\Omega$ subspaces cross each other; they produce
883: anti-crossings. For the three purely long-range wells of interest, such
884: crossings appear far enough in the classically forbidden region that
885: the off-diagonal coupling terms can be neglected in the calculation of
886: the binding energy. Thus, in the following calculation, only the terms
887: coupling states within a given $\Omega$ subspace (first line in
888: Equation (\ref{eq:Detail l2})) are kept in the expression of the
889: rotation of the nuclei.
890:
891: Figure \ref{fig: Effect_Of_Rotation} shows the change in the three {\it
892: ungerade} potential wells resulting from the inclusion of the rotation
893: of the nuclei in the hamiltonian. The minimum possible value for $J$ is
894: $J=\Omega$. For higher values of $J$ the contribution of the
895: centrifugal barrier due to the rotation of the nuclei increases.
896: Bose-Einstein statistics dictates that $J$ should be odd for $0_u^+$,
897: and even for $0_u^-$ (see e.g. \cite{Hougen}). There is no restriction
898: on $J$ for the $2_u$ state, since it is doubly degenerate.
899:
900:
901: \subsubsection{Effect of the vibration}
902: \label{section effect of vibration}
903:
904: Next, since the electronic states depend on $R$ (Figure \ref{fig:
905: Projections_Ouplus}), the vibration of the nuclei also influences the
906: electronic degrees of freedom. This effect is described by the radial
907: part of the kinetic energy of the nuclei, namely the first term in
908: parenthesis in Equation (\ref{eq: nuclear_kinetic_energy}). This final
909: addition to the hamiltonian leads to the following equation:
910: \begin{eqnarray}\label{eq: bound states hamiltonian}
911: \hat{H}\ket{\psi} = \left\{-\frac{\hbar^2}{2\mu}\
912: \frac{1}{R}\frac{\partial^2}{\partial R^2}R
913: +\hat{\mathcal{H}}\right\}\ket{\psi}=E\;\ket{\psi}
914: \end{eqnarray}
915: where the eigenstates $\ket{\psi}$ are written using a basis with
916: separable variables: $\ket{\psi}=\ket{\chi_v} \otimes
917: \ket{\phi_{J,\Omega_{u}^{\pm}}}$, with $\ket{\chi_v}$ the vibrational
918: part, and $\ket{\phi_{J,\Omega_{u}^{\pm}}}$ the electronic and
919: rotational part. With these notations,
920: $\ket{\phi_{J,\Omega_{u}^{\pm}}}$ are the $R$-dependent eigenstates of
921: the hamiltonian $\hat{\mathcal{H}}$, with the eigenvalues
922: $V_{J,\Omega_{u}^{\pm}}(R)$ determined previously and given in Figure
923: \ref{fig: Effect_Of_Rotation}.
924:
925:
926: Because the crossings between electronic potential curves lie far
927: enough in the classically forbidden region, the action of
928: $\partial^2/\partial R^2$ on the electronic part should be considered
929: as a diagonal correction and we neglect the off-diagonal terms of this
930: operator. This is the so-called adiabatic approximation
931: \cite{Lefebvre-Brion}, and Equation (\ref{eq: bound states
932: hamiltonian}) reduces to a set of independent radial equations:
933: \begin{widetext}
934: \begin{eqnarray}\label{eq: radial equation}
935: \left\{-\frac{\hbar^2}{2\mu}\;\left(\frac{d^2}{dR^2}+\left\langle
936: \phi_{J,\Omega_{u}^{\pm}} \left| \frac{\partial^2}{\partial R^2}
937: \right| \phi_{J,\Omega_{u}^{\pm}} \right\rangle \right)+
938: V_{J,\Omega_{u}^{\pm}}(R) -E_{J,\Omega_{u}^{\pm},v}\right\}\;u(R)=0 \
939: \mbox{,}
940: \end{eqnarray}
941: \end{widetext}
942: where the vibrational part of the wave function has been written
943: $\bra{\vec{R}}\chi_v\rangle=u(R)/R$, and $E_{J,\Omega_{u}^{\pm},v}$ is
944: the binding energy for the ro-vibrational level $(J,v)$ in the
945: $\Omega_{u}^{\pm}$ potential well. Finally, the vibration of the nuclei
946: is described through a single effective potential well which is the sum
947: of $V_{J,\Omega_{u}^{\pm}}(R)$ (which already takes into account the
948: rotation of the nuclei) and of the correction coming form the
949: dependence in $R$ of the eigenstates of $\hat{\mathcal{H}}$.
950:
951:
952: \subsection{Calculation and comparison with the experimental spectrum}
953: \label{section: Theory Binding energies}
954:
955: Table \ref{tab: comparison} provides the comparison between the
956: experimental results obtained for the $0_u^+$ potential well (column
957: A), and the calculated binding energies from the adiabatic approach
958: developed above (column B). In column (A), the measured binding
959: energies include the corrections discussed in Section \ref{section:
960: shifts}. Within the experimental accuracy, the agreement between our
961: measurement and our predictions for $J=1$ is remarkably good (except
962: for the $v=5$ line, which is too close to the atomic resonance to be
963: observed). Note that the $v=0$ line was probed with a different laser
964: set up, so its measured binding energy is less precise than the others
965: (see \cite{Leonard}). Also, the $(0_u^+,J=3)$ progression is too weak
966: to be observed in our experiment.
967:
968:
969: The effect of retardation on the calculated energy is illustrated by
970: the quantity $\epsilon^{Ret}$ (Table \ref{tab: comparison}, column C).
971: It increases the depth of the well, and therefore the binding energies
972: as well. Compared with the non retarded calculation ($k \rightarrow 0$
973: in the expressions \ref{eq:_Retarded_Dipole_1} and
974: \ref{eq:_Retarded_Dipole_2}), retardation is a correction proportional
975: to $R^2$ in relative value, but to $1/R$ in absolute value. Therefore,
976: it becomes very important in relative values for very elongated states
977: (up to $\sim 30 \%$ for $v=5$), and it is more important in absolute
978: values for less elongated states ($\epsilon^{Ret}=-6.6 MHz$ for $v=0$).
979: Given the experimental accuracy of 0.5 MHz, this work is a
980: demonstration of the retardation effect, which has to be taken into
981: account to reproduce the measured binding energies. This effect has
982: been already demonstrated for sodium atoms in 1996 \cite{Retard_Na}.
983:
984: The correction to the electronic potential due to the vibration of the
985: nuclei is illustrated by the quantity $\epsilon^{Rad}$ in Table
986: \ref{tab: comparison}, column (D). Practically $\epsilon^{Rad}$ is the
987: difference between the binding energy calculated with and without the
988: term $\bra{\phi_{J,\Omega_{u}^{\pm}}}
989: \partial^2/\partial R^2 \ket{\phi_{J,\Omega_{u}^{\pm}}}$ in Equation
990: (\ref{eq: radial equation}). This term is part of the kinetic energy of
991: the system. Thus it brings a positive contribution to the effective
992: electronic potential and it moves the bound states upward in the wells.
993: Its contribution is non vanishing in the region where the electronic
994: state changes its character with $R$ due to the anti-crossings
995: discussed previously, that is to say in the vicinity of the bottom of
996: the potential well. Therefore the correction is stronger for the
997: deepest states, as they don't extend very far from this region. Weakly
998: bound states extend much farther into regions where the electronic
999: state does not depend strongly on $R$ (pure Hund's case c), and the net
1000: effect is less pronounced.
1001:
1002:
1003: \begin{table}[t]
1004: \caption{\footnotesize Comparison between experimental and theoretical
1005: binding energies in the case of the $0_u^+$ purely long-range potential
1006: well. Column (A) gives the experimental results, after the corrections
1007: discussed in Section \ref{section: shifts} are applied. Column (B)
1008: gives the binding energy $E_{v,J}$ calculated from equation \ref{eq:
1009: radial equation} within the adiabatic approximation. For each bound
1010: state, $\epsilon^{Ret}$ is an estimate of the contribution to $E_{v,J}$
1011: of the retardation effect. $\epsilon^{Ret}$ comes from the comparison
1012: with the non-retarded calculation. Similarly, $\epsilon^{Rad}$ is the
1013: calculated estimate of the term $\bra{\phi_{J,\Omega_{u}^{\pm}}}
1014: \partial^2/\partial R^2 \ket{\phi_{J,\Omega_{u}^{\pm}}}$ (see Equation
1015: (\ref{eq: radial equation})). Note that the binding energies presented
1016: in column (B) already implicitly contain the contributions
1017: $\epsilon^{Ret}$ and $ \epsilon^{Rad}$. All the energies are given in
1018: units of $h$, in MHz.}
1019: \begin{ruledtabular}
1020: \begin{tabular}{cc|ccc}
1021: & (A) & (B) & (C) & (D) \\
1022: $v$ & Experiment & $E_{v,J}$ & $\epsilon^{Ret}$ & $ \epsilon^{Rad} $\\
1023: \hline
1024: & & & & \\
1025: 5 & - & -2.487 & -0.78 & +0.053 \\
1026: 4 & -18.2 $\pm 0.5$ & -18.12 & -1.6 & +0.28 \\
1027: 3 & -79.6 $\pm 0.5$ & -79.41 & -2.6 & +0.95 \\
1028: 2 & -253.3 $\pm 0.5$ & -252.9 & -3.9 & +2.4 \\
1029: 1 & -648.5 $\pm 0.5$ & -648.3 & -5.2 & +5.3 \\
1030: 0 & -1430 $\pm 20$ & -1418 & -6.6 & +10.3 \\
1031: & & & & \\
1032:
1033:
1034: \end{tabular}
1035: \end{ruledtabular}
1036: \label{tab: comparison}
1037: \end{table}
1038:
1039:
1040: Finally, the high accuracy of the data and the good agreement between
1041: the experimental and calculated spectra lead to an experimental
1042: determination of the $C_3$ coefficient, which describes the
1043: dipole-dipole interaction. In our calculations, changing $C_3$ by
1044: $0.1\%$ changes the binding energies by at most 0.3 MHz, which is of
1045: order of our experimental accuracy. Therefore, the present results
1046: confirm the theoretical value used for the $C_3$ coefficient to within
1047: $0.2\%$. As a consequence of Equation \ref{eq: C3}, we can infer that
1048: the atomic radiative decay rate is $\Gamma=2\pi \times (1.625 \pm
1049: 0.003)$ MHz. As far as we know, this is the most accurate experimental
1050: determination for the helium $2^3P$ decay rate.
1051:
1052:
1053: \subsection{Other {\it ungerade} giant dimers}
1054: \label{section: Other wells}
1055:
1056: \begin{table}[b]
1057: \caption{\footnotesize Results of the calculation detailed in the text
1058: for the three purely long-range {\it ungerade} wells. Column (A) gives
1059: the binding energy $E_{v,J}$ calculated within the adiabatic
1060: approximation. The three last columns illustrate the unusual size of
1061: the dimers. $R_{min}$ and $R_{max}$ are classical inner and outer
1062: turning points, $\langle R \rangle$ is the mean internuclear distance.
1063: All the energies are given in MHz, and the lengths in atomic units.}
1064: \begin{ruledtabular}
1065: \begin{tabular}{ccc|rrr}
1066: & & (A) & (B) & (C) & (D) \\
1067: & $v$ & $E_{v,J}$\footnote{Binding energies are given with respect to the asymptote of the potential considered.} & $R_{min}$ & $R_{max}$ & $\langle R \rangle$ \\
1068: \hline
1069: & & & & & \\
1070: $0_u^+$, $J=1$& 5 & -2.487 & 147.6 & 2182 & 1797 \\
1071: & 4 & -18.12 & 147.7 & 1122 & 917 \\
1072: & 3 & -79.41 & 148.1 & 689 & 560 \\
1073: & 2 & -252.9 & 149.5 & 467 & 379 \\
1074: & 1 & -648.3 & 152.9 & 336 & 276 \\
1075: & 0 & -1418 & 162.5 & 246 & 213 \\
1076: & & & & & \\
1077: \hline
1078: & & & & & \\
1079: $0_u^-$, $J=2$& 0 & -7.304 & 461.7 & 970 & 824 \\
1080: & & & & & \\
1081: \hline
1082: & & & & & \\
1083: $2_u$, $J=2$& 3 & -4.584 & 320.5 & 2097 & 1712 \\
1084: & 2 & -21.41 & 322.5 & 1231 & 999 \\
1085: & 1 & -72.32 & 329.3 & 808 & 659 \\
1086: & 0 & -191.5 & 351.1 & 558 & 477 \\
1087: & & & & & \\
1088:
1089: \end{tabular}
1090: \end{ruledtabular}
1091: \label{tab: results}
1092: \end{table}
1093:
1094: Bound states in {\it ungerade} potential wells other than $0_u^+$ have
1095: not been explored. However, the calculation presented above can also be
1096: applied for those. Table \ref{tab: results} presents the theoretical
1097: results for the molecular binding energies and characteristic sizes in
1098: the three {\it ungerade} purely long-range potential wells. Column (A)
1099: gives the results obtained when one solves Equation \ref{eq: radial
1100: equation}. Experimentally, bound states are produced by driving an
1101: electric dipole transition from the electronic state $^5\Sigma_g^+$
1102: with $J=2$, so only $J=1$, 2 or 3 are accessible. In Table \ref{tab:
1103: results}, the results are given for one relevant value of $J$, taking
1104: into account the Bose-Einstein statistics already mentioned in
1105: paragraph \ref{parag: Effect of rotation}.
1106:
1107: The purely long-range character of these molecules arises from the very
1108: large distance at which their inner classical turning points lie (Table
1109: \ref{tab: results}, column B). The outer turning points (column C) and
1110: mean sizes $\langle R \rangle =\bra{\chi_v}R\ket{\chi_v}$ (column D)
1111: are also particularly large, leading to an unusual type of ``giant"
1112: dimer for which asymptotic calculations allow an accurate description.
1113: At such large distances, the next order term $C_6/R^6$ in the
1114: electromagnetic interaction can clearly be neglected. The $C_6$
1115: coefficient has never been published for this system, but one can
1116: estimate that it is smaller than the value of $C_6=3265$ a.u. for the
1117: $2^3S-2^3S$ interaction \cite{Starck} and calculate the order of
1118: magnitude of the neglected term. For internuclear distances larger than
1119: 150 $a_0$, which is the range of interest for these purely long-range
1120: molecules (see Table \ref{tab: results}), $C_6/R^6<C_3/R^3\times 1.5\
1121: 10^{-4}$. So neglecting this term leads to an error smaller than the
1122: one due to the uncertainty on $C_3$.
1123:
1124:
1125:
1126: While writing the present article we were informed that Venturi {\it et
1127: al.} \cite{Venturi} had submitted for publication the result of a
1128: multichannel calculation, which is also in very good agreement with our
1129: experimental results. Their method is more elaborate and allows for a
1130: direct solution of the full set of equations (\ref{eq: bound states
1131: hamiltonian}). However, the binding energies obtained by both methods
1132: are equal to within 0.5 MHz for all the bound states presented in Table
1133: \ref{tab: results}. We have also performed a multi-channel resolution
1134: of Equation \ref{eq: bound states hamiltonian} with the use of a mapped
1135: Fourier grid method. Our results \cite{Proceeding_Australie} are
1136: comparable to those of reference \cite{Venturi} to within 100 kHz. The
1137: main reason why the adiabatic approach is efficient and the
1138: multi-channel calculation required is that there is no crossing between
1139: the adiabatic potential wells of interest and the other potential
1140: curves. This allows for a single-channel calculation that leads to
1141: Equation \ref{eq: radial equation} and is accurate enough to reproduce
1142: the experimental spectrum.
1143:
1144: \section{Summary and conclusion}
1145:
1146:
1147: In a previous Letter \cite{Leonard}, we reported an accurate
1148: measurement of the binding energies of purely long-range helium dimers
1149: in the $0_u^+$ potential well connected to the $2^3S_1+2^3P_0$
1150: asymptote. The present paper reports theoretical calculations which
1151: complement the experimental results in order to interpret the spectra
1152: measured.
1153:
1154:
1155: The experiment consists in measuring the PA laser detunings for which a
1156: strong heating of the atomic cloud is observed. The heating is assumed
1157: to be a consequence of the resonant excitation of a bound state in the
1158: $0_u^+$ potential well. To infer the corresponding binding energy, the
1159: measured PA laser detunings must be corrected from a mean shift of the
1160: molecular lines due to the non-zero magnetic field $B_0$ at the center
1161: of the trap, and also to the non-zero temperature of the cold gas.
1162: Since the detunings are measured with high accuracy, a simple
1163: calculation shows that the temperature-induced shift must be
1164: considered, given the range of temperature explored (2 - 30 $\mu$K).
1165: This calculation does not include the exact shape and width of the
1166: lines but only gives in a mean correction. The binding energies deduced
1167: after correction are independant of the density, and no magnetic dipole
1168: moment is detectable for the excited state. Apart from the symmetric
1169: and asymmetric broadening mechanisms discussed in Section \ref{section:
1170: shifts}, the lineshapes are actually also influenced by the dynamics of
1171: the heating mechanism. Indeed the temperature curves are an indirect
1172: measurement of the lineshape which relies on the efficiency of the
1173: thermalization of the cloud. An incomplete thermalization can lead to
1174: another source of broadening of the lines, but no additional shift. The
1175: calorimetric detection scheme and its implications on the lineshape
1176: will be discussed in a separate paper.
1177:
1178: Here we have presented an approximate solution of the Schr\"odinger
1179: equation that is well suited for asymptotically large internuclear
1180: distances. The adiabatic approach allows for accurate calculations of
1181: the binding energies in the case of purely long-range potential wells.
1182: The calculation can easily be extended to other purely long-range
1183: potential wells which can in principle also be observed in our
1184: experimental conditions, namely $0_u^-$ and $2_u$.
1185:
1186: Finally, the comparison between the experimental and theoretical
1187: determination of the binding energies in the $0_u^+$ potential well is
1188: very good if retardation effects are taken into account. As a
1189: consequence, an accurate measurement of the radiative decay rate for
1190: the excited atomic state $2^3P$ can be inferred. The accuracy of the
1191: experimental data allows for a test of retardation effects as well as
1192: of tiny vibration-induced couplings between electronic and nuclear
1193: degrees of freedom.
1194:
1195: Thus, the excellent agrement between our perturbative calculation and
1196: our experiment suggests a good understanding of the purely long range
1197: system. This work is a first step towards a better knowledge of pair
1198: interactions in ultra-cold metastable helium. Further developments will
1199: follow in order to measure the s-wave scattering length for two atoms
1200: interacting through the $^5\Sigma_g^+$ electronic potential.
1201:
1202: {\bf Acknowledgements :} The authors thank the group of F.
1203: Masnou-Seeuws, at Laboratoire Aim\'e Cotton in Orsay, for fruitful
1204: discussions.
1205:
1206: \begin{thebibliography}{30}
1207:
1208: \bibitem{Revues} see e.g. review articles W.C. Stwalley,H. Wang, J.
1209: Mol. Spec. {\bf 195}, 194 (1999), J. Weiner, V. S. Bagnato, S. Zilio,
1210: P. S. Julienne, Rev. Mod. Phys. {\bf 71}, 1 (1999), F. Masnou-Seeuws,
1211: P. Pillet, Advances in Atomic, Molecular, and Optical Physics {\bf 47},
1212: 53 (2001), and references therein.
1213:
1214: \bibitem{Abraham} E. R. I. Abraham, W. I. McAlexander, C.A. Sackett, R. G. Hulet, Phys. Rev. Lett. {\bf 74}, 1315 (1995).
1215:
1216: \bibitem{Gardner} J. R. Gardner, R. A. Cline, J. D. Miller, D. J.
1217: Heinzen, H. M. J. M. Boesten, B. J. Verhaar, Phys. Rev. Lett. {\bf 74},
1218: 3764 (1995).
1219:
1220: \bibitem{Herschbach} N. Herschbach, P. J. J. Tol, W. Vassen, W. Hogervorst, G. Woestenenk, J.W. Thomsen, P. van der
1221: Straten, A. Niehaus, Phys. Rev. Lett. {\bf 84}, 1874, (2000).
1222:
1223: \bibitem{Pereira101} F. Pereira Dos Santos, J. L\'eonard, Junmin
1224: Wang, C. J. Barrelet, F. Perales, E. Rasel, C. S. Unnikrishnan, M.
1225: Leduc, C. Cohen-Tannoudji, Phys. Rev. Lett. {\bf 86}, 3459 (2001).
1226:
1227: \bibitem{Robert} A. Robert, O. Sirjean, A. Browaeys, J. Poupard, S.
1228: Nowak, D. Boiron, C. I. Westbrook, A. Aspect, Sci. Mag. {\bf 292}, 463
1229: (2001).
1230:
1231: \bibitem{Sirjean} O. Sirjean, S. Seidelin, J. Viana Gomes, D. Boiron, C. I. Westbrook, A. Aspect, G.V.
1232: Shlyapnikov, Phys. Rev. Lett. \textbf{89}, 220406 (2002).
1233:
1234: \bibitem{Leduc} M. Leduc, J. L\'eonard, F. Pereira dos Santos, E. Jahier, S. Schwartz and C.
1235: Cohen-Tannoudji, Acta Phys. Pol. \textbf{B33}, p. 2213 (2002).
1236:
1237: \bibitem{Leonard} J.L\'eonard, M.Walhout, A.P.Mosk, T.Mueller, M.Leduc, C.Cohen-Tannoudji, Phys. Rev.
1238: Lett. \textbf{91}, 073203 (2003).
1239:
1240: \bibitem{StarckMeyer} J. St\"arck, W. Meyer, Chem. Phys. Lett. {\bf 255}, p.229 (1994).
1241:
1242: \bibitem{Pereira2} F. Pereira dos Santos, J. L\'eonard, Junmin Wang, C.
1243: J. Barrelet, F. Perales, E. Rasel, C. S. Unnikrishnan, M. Leduc, C.
1244: Cohen-Tannoudji, Eur. Phys. J. D, {\bf 19}, 103 (2002).
1245:
1246: \bibitem{Napolitano} R. Napolitano, J. Weiner, C. J. Williams, P. S.
1247: Julienne, Phys. Rev. Lett. {\bf 73}, 1352 (1994).
1248:
1249: \bibitem{Dip-Dip_in_medium} Ho Trung Dung, Ludwig Kn\"oll, and Dirk-Gunnar Welsch, Phys. Rev. A \textbf{66}, 063810
1250: (2002).
1251:
1252: \bibitem{Lefebvre-Brion} H. Lefebvre-Brion, R. W. Field, ``Perturbations
1253: in the spectra of diatomic molecules", Acamdemic Press (1986).
1254:
1255: \bibitem{Herzberg} G. Herzberg, "Spectra of Diatomic Molecules", 2$^{nd}$ edition (1950),
1256: D. Van Nostrand Company INC.
1257:
1258: \bibitem{Dashevskaya} E.I. Dashevskaya, A.I. Voronin, E.E. Nikitin,
1259: Can. J. Phys. {\bf 47}, 1237 (1969).
1260:
1261: \bibitem{Meath} W. J. Meath, J. Chem. Phys. {\bf 48}, 227 (1968).
1262:
1263:
1264: \bibitem{Drake} G. W. F. Drake in {\it Atomic, Molecular and Optical
1265: Physics Handbook}, edited by G. W. F. Drake, AIP Press, Chap.11 (1996).
1266:
1267: \bibitem{Bethe_Salpeter} H. A. Bethe, E. E. Salpeter, ``Quantum
1268: mechanics of one- and two-eletcron atoms", Springer-Verlag (1957).
1269:
1270: \bibitem{FineStructSplit1} M.C. George, L.D. Lombardi, E.A. Hessels,
1271: Phys. Rev. Lett. {\bf 87}, 173002, (2001), and references therein.
1272:
1273: \bibitem{FineStructSplit2} J. Castillega, D. Livingston, A. Sanders, D.
1274: Shiner, Phys. Rev. Lett. {\bf 84}, 4321, (2000), and references
1275: therein.
1276:
1277:
1278: \bibitem{MovrePichler} M. Movre and G. Pichler, J. Phys. B, {\bf 10}, p. 2631 (1977).
1279:
1280: \bibitem{Stwalley} W.C. Stwalley, Y.-H. Uang, G. Pichler, Phys. Rev.
1281: Lett. {\bf 41}, p. 1164 (1978).
1282:
1283: \bibitem{Zare} see for example R. N. Zare, ``Angular Momentum", Wiley and Sons (1988).
1284:
1285: \bibitem{Hougen} J. T. Hougen, ``The calculation of rotational energy
1286: levels and rotational line intensities in diatomic molecules", Nat.
1287: Bur. of Standards, momograph {\bf 115}, (1970).
1288:
1289: \bibitem{Starck} J. St\"arck, W. Meyer, Chem. Phys. Lett. {\bf 255},
1290: p.229 (1994).
1291:
1292: \bibitem{Retard_Na} K. M. Jones, P. S. Julienne, P. D. Lett, W. D. Phillips, E. Tiesinga and
1293: C. J. Williams, Europhys. Lett., \textbf{35}, 85 (1996).
1294:
1295: \bibitem{Venturi} V. Venturi, P. J. Leo, E. Tiesinga, C. J. Williams, I. B. Whittingham, Phys. Rev.
1296: A. 68, 022706 (2003).
1297:
1298: \bibitem{Proceeding_Australie} M. Leduc, M. Portier, J. L\'eonard, M. Walhout, F. Masnou-Seeuws, K. Willner, A.
1299: Mosk, ``Laser Spectroscopy XVI, Proceedings of the XVI International
1300: Conference on Laser Spectroscopy", Palm Cove, Australia (14-18 Juillet,
1301: 2003), Eds. P. Hannaford, H.-A. Bachor, K.G. Baldwin and A.I. Sidorov
1302: (World Scientific, New Jersey, London, Singapore, Hong Kong).
1303:
1304: \end{thebibliography}
1305:
1306: \end{document}
1307: