1: \documentclass[prl,aps,showpacs,twocolumn]{revtex4}
2: \usepackage{epsf}
3: \usepackage[dvips]{color}
4: \newcommand {\be}{\begin{equation}}
5: \newcommand {\ee}{\end{equation}}
6: \newcommand {\bea}{\begin{eqnarray}}
7: \newcommand {\eea}{\end{eqnarray}}
8: \newcommand{\cs}{$\clubsuit$}
9: \newcommand{\sps}{$\spadesuit$}
10:
11: \begin{document}
12:
13: \title{Achieving a BCS transition in an atomic Fermi gas}
14: \author{L.~D. Carr$^1$$^*$, G.V.~Shlyapnikov$^{1,2,3}$$^{\dag}$ and Y. Castin$^1$}
15: \affiliation{\mbox{1. Laboratoire Kastler Brossel, Ecole Normale
16: Sup\'erieure, 24 rue Lhomond, 75231 Paris, France,}}
17: %\email[]{shlyap@goran.amolf.nl}
18: \affiliation{2. \mbox{FOM Institute for Atomic and Molecular
19: Physics, Kruislaan 407, 1098 SJ Amsterdam, The Netherlands}}
20: \affiliation{3. Russian Research Center, Kurchatov Institute,
21: Kurchatov Square, 123182 Moscow, Russia}
22:
23: \date{\today}
24:
25: \begin{abstract}
26: We consider a gas of cold fermionic atoms having two spin components
27: with interactions characterized by their $s$-wave scattering length
28: $a$. At positive scattering length the atoms form weakly bound bosonic
29: molecules which can be evaporatively cooled to undergo Bose-Einstein condensation,
30: whereas at negative scattering
31: length BCS pairing can take place. It is
32: shown that, by adiabatically tuning the scattering length $a$ from
33: positive to negative values, one may transform the molecular
34: Bose-Einstein condensate into a highly degenerate atomic Fermi gas,
35: with the ratio of temperature to Fermi temperature $T/T_F \sim 10^{-2}$.
36: The corresponding critical final value of $k_{F}|a|$ which leads to the BCS
37: transition is found to be about one half, where $k_F$ is the Fermi momentum.
38: \end{abstract}
39:
40: \pacs{}
41:
42: \maketitle
43:
44: Much progress has been made in the achievement of increasingly
45: degenerate regimes of trapped atomic Fermi
46: gases~\cite{jin3,truscott1,schreck1,granade2002,roati2002,ohara1,hadzibabic2003}.
47: One of the major goals of studies of these systems is to observe a
48: transition to a paired-fermion superfluid state. In a recent MIT
49: experiment on a sympathetically cooled single component Fermi gas,
50: a system with $T/T_F=0.05$ and $3\times 10^7$ fermions was
51: realized~\cite{hadzibabic2003}, where $T_F$ is the Fermi
52: temperature. As trapped atomic gases present an essentially
53: impurity free system, with density, temperature, and interaction
54: strength all free parameters, they offer great opportunities for
55: investigation of fundamental theories of superfluid states.
56:
57: An important new step in experiments has been the use of Feshbach
58: resonances, whereby one may tune the $s$-wave scattering length $a$
59: from positive to negative over many orders of
60: magnitude~\cite{vogels1}. This opens up the possibility of
61: investigating the Bose-Einstein condensate to
62: Bardeen-Cooper-Schrieffer (BEC-BCS)
63: crossover~\cite{nozieres1985,randeria1995,holland2001,ohashi2003}.
64: The BCS limit occurs for $k_F |a| \ll 1$, where $k_F$ is the Fermi
65: momentum and $a<0$. In this case the fermion pair size is much
66: larger than the interatomic spacing. The BEC limit occurs for a
67: positive scattering length much smaller than the interparticle
68: separation. Then the fermions form weakly bound dimers of a size
69: $\sim a$, and the Bose-Einstein condensation of these composite
70: bosons may be described by the well-developed theory of trapped
71: BEC's~\cite{dalfovo1}. Recently, long-lived molecules have been
72: created in a reversible, and therefore adiabatic
73: fashion from a degenerate Fermi gas by tuning the scattering
74: length~\cite{salomon2003}. This provides many
75: possibilities for evaporatively
76: cooling the molecules into the BEC regime. In contrast, a BCS transition
77: requires very low temperatures not yet obtained in two-component
78: Fermi gases.
79:
80: In this Letter, we present a straightforward theoretical
81: analysis which shows that a deeply degenerate Fermi gas
82: may indeed be created from a molecular BEC. By adiabatically
83: changing the scattering length from positive to negative values,
84: the entropy is held constant, which leads to a strong decrease in temperature.
85: This suggests a novel method of cooling a Fermi gas to extremely low
86: temperatures. One can then reach a temperature
87: $T\sim 10^{-2}T_F$ and achieve a BCS transition.
88:
89: Consider a harmonically trapped Fermi gas described by the grand
90: canonical ensemble with two equally populated spin states.
91: The entropy of the gas above the critical temperature
92: is equal to the entropy of an ideal
93: Fermi gas in a trap with mean field corrections. Omitting these corrections
94: the grand potential $\Omega$ is~\cite{diu1989}
95: \be \Omega=k_B T \int_0^{\infty} d\epsilon
96: \rho(\epsilon) \ln[1-n(\epsilon)]\, , \label{eqn:j}\ee with
97: $\rho(\epsilon)$ the density of states and \be n(\epsilon)=
98: 1/\{\exp[\beta(\epsilon-\mu)]+1\}\ee the Fermi weighting
99: factor. For a harmonic trap, the density of states for a
100: non-interacting two-component gas in the semi-classical limit
101: is \be \rho(\epsilon) =
102: \epsilon^2/(\hbar\omega)^3\equiv A \epsilon^2\,
103: .\label{eqn:dos}\ee Substituting Eq.~(\ref{eqn:dos}) into
104: Eq.~(\ref{eqn:j}) and integrating by parts, one obtains the
105: simplified integral $ \Omega=-(A/3)\int_0^{\infty} d\epsilon
106: \,\epsilon^3 n(\epsilon)$. In the degenerate regime, given by
107: $\mu\gg k_B T$, the integral may be expanded as \be
108: \Omega\simeq-\frac{A}{3}\left[\frac{\mu^4}{4}+\frac{\pi^2(k_B
109: T)^2\mu^2}{2} +\frac{7\pi^4 (k_B T)^4}{60}\right]
110: \label{eqn:j3}\, .\ee The entropy may then be obtained from the
111: relation~\cite{diu1989} \be S=-\partial_T \Omega
112: |_{\mu, \omega}\, ,\label{eqn:jtos}\ee giving \be S \simeq k_B
113: \frac{A}{3}\left[\pi^2 k_B T \mu^2 + \frac{7\pi^4 (k_B T)^3}{15}\right] \, .\ee
114: A relation between the total number of particles and the
115: chemical potential follows from the equation
116: $N=-\partial_{\mu}\Omega|_{T,\Omega}$ and reads
117: \be N \simeq (A/3)[\mu^3 +
118: \pi^2(k_B T)^2\mu] \, .\label{eqn:fmu}\ee
119: So, to lowest order, the chemical potential is $\mu=(3N/A)^{1/3}$
120: and the expression for the entropy becomes \be S=k_B
121: N \pi^2 T/T_F+{\cal O}(T^3)\, ,\label{eqn:sfermi}\ee
122: where $T_F\equiv (3N)^{1/3}\,\hbar\omega$ is
123: the Fermi temperature of the non-interacting gas. In the case of
124: an interacting Fermi gas, the lowest order mean field correction
125: in the Thomas-Fermi limit leads to the appearance of an extra
126: multiple $(1+64k_F a/35\pi^2$)
127: in the numerator of the right hand side of Eq.~(\ref{eqn:sfermi}).
128: This gives a correction of less than 10 percent
129: to the entropy for $k_F|a|\leq 1/2$. It may therefore be neglected in our calculations. Equation (\ref{eqn:sfermi}) is valid
130: for $k_B T > \hbar\omega$ in an isotropic potential. Its validity
131: extends to much lower temperatures (although much larger than the
132: inverse density of states at the Fermi energy) for incommensurable
133: trap frequencies.
134:
135: Consider the analogous calculation for a BEC of
136: weakly bound molecules at $a>0$. We assume that the condensate is in the
137: weakly interacting regime, with $\eta \equiv n_{\mathrm{mol}}^{1/3}
138: a_{\mathrm{mol}} \ll 1$ where $n_{\mathrm{mol}}$ is the molecular density
139: in the trap center and the
140: scattering length of two molecules $a_{\mathrm{mol}}$ is related to the
141: atomic scattering length $a$ by $a_{\mathrm{mol}} = 0.6
142: a$~\cite{petrov2003b}. Note that these weakly bound molecules exhibit a
143: remarkable collisional stability at large values of
144: $a$~\cite{salomon2003,petrov2003b}. This stability
145: originates from a strong decrease in the relaxation rate to deep
146: bound states with increasing $a$, and should enable efficient
147: evaporative cooling~\cite{petrov2003b}. A small value of $\eta$ may be
148: reached in experiments by reducing the trap frequencies.
149:
150: If $\eta \ll
151: 1$, the mean field shift in the condensation temperature is
152: $\delta T_{\mathrm{BEC}}/T_{\mathrm{BEC}} = -3.25 \eta^{5/4}$ \cite{dalfovo1} and the
153: quantum depletion of the condensate is negligible so that
154: Bogoliubov theory may be used. It also implies \be k_B T_{\mathrm{BEC}}/E_B = 7.78 \eta^{5/2} \ll
155: 1\label{eqn:cond1}\ee where $E_B\simeq \hbar^2/ma^2$ is the
156: molecular binding energy. This condition, together with $T<
157: T_{\mathrm{BEC}}$, ensures that the molecules may be treated as weakly
158: interacting bosons and any effect due to their thermal
159: dissociation is negligible. The grand potential is \be \Omega=k_B T
160: \int_0^{\infty} d\epsilon \rho(\epsilon)
161: \ln\left(1-e^{-\beta\epsilon}\right)\, , \label{eqn:jb}\ee where
162: $\beta$ now refers to the temperature of the Bose gas.
163: The density of states for an interacting BEC in the
164: Bogoliubov approximation may be calculated directly from the
165: Bogoliubov Hamiltonian as
166: $\rho(\epsilon)=\mathrm{Tr}[\delta(\epsilon-\hat{H}_{bog})]$. In
167: the Thomas-Fermi limit, which holds for a molecular condensate
168: with chemical potential $\mu_{\mathrm{mol}} \gg \hbar \omega$,
169: the density of states may be approximated semi-classically by \be
170: \rho(\epsilon)\simeq \int\int
171: \frac{d\vec{r}d\vec{p}}{(2\pi\hbar)^3}
172: \:\delta[\epsilon-\epsilon_{bog}(\vec{p},\vec{r})]\,
173: ,\label{eqn:dos2}\ee where the energy of Boguliubov excitations is (see~\cite{dalfovo1} and references therein) \be \epsilon_{bog} \equiv
174: \left\{\sqrt{p^2/2m_{\mathrm{mol}}\left[p^2/2m_{\mathrm{mol}}+2g_{\mathrm{mol}}n_0(r)\right]}
175: \quad|r|\leq R \atop{p^2/2m_{\mathrm{mol}}+\frac{1}{2}m_{\mathrm{mol}}\omega^2r^2
176: -\mu_{\mathrm{mol}} \quad\,\,\,\,\,\,\,\,\,\, |r|>R\, .}\right .\, \ee
177: Here $m_{\mathrm{mol}}=2m$ is the
178: mass of a molecule and
179: \be
180: n_0(r)=(\mu_{\mathrm{mol}}/g_{\mathrm{mol}})(1-r^2/R^2)\,
181: \label{eqn:tf}\ee is the
182: Thomas-Fermi density profile, with $R\equiv(2\mu_{\mathrm{mol}}/m_{\mathrm{mol}}\omega^2)^{1/2}$.
183: The coupling constant
184: for the molecule--molecule interaction is
185: $g_{\mathrm{mol}}=4\pi\hbar^2 a_{\mathrm{mol}}/m_{\mathrm{mol}}$.
186: The exact calculation of Eq.~(\ref{eqn:dos2}) yields \bea
187: \rho(\epsilon)=[\mu_{\mathrm{mol}}^2/(\pi\hbar^3\omega^3)]
188: \{2\sqrt{2}z\tanh^{-1}[\sqrt{2z}/(1+z)
189: ]\nonumber\\
190: +4z^{3/2}-\sqrt{2}z^2 [\pi+2\tan^{-1}
191: ((1-z)/\sqrt{2z})]\nonumber\\
192: +(1+z)^2 [\theta_0 -\sin(4\theta_0)/4]\}\, ,\,\,\,\,\eea where
193: $z\equiv \epsilon/\mu_{\mathrm{mol}}$ is the rescaled energy and
194: $\theta_0\equiv\cos^{-1}(1/\sqrt{1+z})$. In the limit $k_B
195: T\gtrsim \mu_{\mathrm{mol}}$, \be
196: \rho(\epsilon)=\frac{\mu_{\mathrm{mol}}^2}{(\hbar\omega)^3}\left[\frac{
197: z^2}{2} +z+\mathcal{O}(z^{0})\right]\, .\label{eqn:lim2}\ee Given
198: the density of states, the entropy can be calculated from Eq.~(\ref{eqn:jtos})
199: and~(\ref{eqn:jb}). The resulting expression is \be S=k_B
200: \frac{N_{\mathrm{mol}}}{\zeta(3)}\left(\frac{T}{T_{\mathrm{BEC}}}\right)^3
201: G(\beta\mu_{\mathrm{mol}})\, ,\label{eqn:sbose}\ee where
202: $\zeta$ is the Riemann zeta function, with $\zeta(3)=1.202\ldots$,
203: $N_{\mathrm{mol}}=N/2$ is the number of molecules and \be G(u)
204: \equiv u^3 \int_0^{\infty}dz f(z)\left[\frac{u
205: z}{e^{u z}-1}- \ln(1-e^{-u z})\right] \,
206: ,\label{eqn:G}\ee with $u\equiv\beta\mu_{\mathrm{mol}}$, and $f(z)\equiv
207: [(\hbar\omega)^3/\mu_{\mathrm{mol}}^2]\rho(z\mu_{\mathrm{mol}})$ so as to make the
208: units explicit. Equation~(\ref{eqn:G}) may be integrated
209: numerically, or explicitly in the limit of Eq.~(\ref{eqn:lim2}),
210: in which case \be S=k_B N_{\mathrm{mol}}\left(
211: \frac{T}{T_{\mathrm{BEC}}}\right)^3\left(\frac{2 \pi^4 }{45 \zeta (3)} +
212: 3\frac{\mu_{\mathrm{mol}}}{k_B T}\right)\, \label{eqn:expand}\ee This
213: expansion is accurate to within 10\% for $k_B T/\mu_{\mathrm{mol}} \geq 1/10$. In
214: the above expression for the entropy, the condensation temperature
215: for a non-interacting Bose gas, $k_B T_{\mathrm{BEC}}=\hbar\omega
216: [N_{\mathrm{mol}}/\zeta(3)]^{1/3}$ was used. Setting the entropy of the
217: molecular condensate and of the Fermi gas equal one obtains the
218: final temperature of the Fermi gas: \be
219: \left(\frac{T}{T_F}\right)_{\mathrm{final}}=\frac{G(\beta\mu_{\mathrm{mol}})}{2\pi^2\zeta(3)}
220: \left(\frac{T}{T_{\mathrm{BEC}}}\right)_{\mathrm{initial}}^3
221: \label{eq:final_temperature} \ee
222: where $\beta$ refers to the initial temperature.
223:
224:
225:
226:
227: %%%%%%%%%%% figure 1 %%%%%%%%%%%
228: %
229: \begin{figure}[t]
230: \begin{center}
231: \epsfxsize=8cm \leavevmode \epsfbox{bcs_fig1.eps}
232: \caption{\label{fig:1}
233: Final temperature of the Fermi gas (left axis)
234: and critical condition for a BCS transition (right
235: axis) as a function of the initial temperature of the Bose
236: condensed molecular gas that has been transformed to a normal
237: attractive Fermi gas via adiabatic switching of the scattering
238: length. The curves, from upper to lower, correspond to
239: $\mu_{\mathrm{mol}}/k_B T_{\mathrm{BEC}}=$ $1$, $1/2$,
240: and $1/4$, where $\mu_{\mathrm{mol}}$ is the molecular chemical
241: potential.}
242: \end{center}
243: \end{figure}
244: %
245:
246:
247: In the strongly degenerate regime, evaporative cooling of a Fermi gas becomes
248: difficult, so that until recently a maximum degeneracy of $T/T_F \sim 0.2$ appeared to be a lower
249: limit, preventing the observation of the BCS transition.
250: The adiabatic tuning of a molecular condensate suggests an
251: alternate method to cooling the fermions directly. According to
252: Eqs.~(\ref{eqn:sfermi}) and~(\ref{eqn:expand}), the entropy is
253: proportional to $T$ for fermions and $T^3$ for bosons. The limits
254: of evaporative cooling for a Thomas-Fermi Bose condensed gas are given by
255: $\mu\simeq k_B T$. Thus the final degeneracy that may be obtained
256: by adiabatic switching of an evaporatively cooled molecular BEC is
257: given by Eq.~(\ref{eq:final_temperature}),
258: with $G(\beta\mu_{\mathrm{mol}})=G(1)\simeq 8.32$.
259: Larger values of $\beta\mu_{\mathrm{mol}}$ are in principle achievable, {\it e.g.}
260: by sympathetic cooling. In Fig.~\ref{fig:1} is shown the final temperature as
261: a function of the initial temperature
262: for various interaction strengths given by the
263: ratio $\mu_{\mathrm{mol}}/k_B T_{\mathrm{BEC}}$.
264:
265: The lowest experimentally achieved temperature of a degenerate
266: Fermi gas to date is $T/T_F=0.05$~\cite{hadzibabic2003}. This
267: temperature may be obtained by adiabatic switching of a molecular
268: BEC of temperature $T=0.5\, T_{\mathrm{BEC}}$
269: with $\mu_{\mathrm{mol}}\simeq k_B T$,
270: which is routinely achieved for atomic BEC's. A temperature of
271: $T=0.25 \,T_{\mathrm{BEC}}$ with $\mu_{\mathrm{mol}}\simeq k_B T$,
272: as for example achieved
273: for an atomic Bose condensed gas in
274: Ref.~\cite{chevy2002}, would give $T/T_F=5\times 10^{-3}$.
275:
276: This offers exciting possibilities for obtaining deeply degenerate
277: Fermi regimes, such as observing the BCS transition in the weakly
278: interacting regime $k_F|a|\ll 1$. The corresponding critical condition is
279: obtained by substituting into Eq.~(\ref{eq:final_temperature})
280: the BCS critical temperature calculated in Ref.~\cite{gorkov1}:
281: \be T_{\mathrm{BCS}}/T_F = \alpha \exp(-\pi/2k_F|a|)\,
282: ,\label{eqn:bcsTc}\ee
283: where $\alpha\equiv e^\gamma\pi^{-1}(2/e)^{7/3}\simeq 0.277$,
284: $\gamma$ being Euler's constant.
285: Strictly speaking, in an isotropic harmonic potential,
286: Eq.~(\ref{eqn:bcsTc}) is only valid in the limit $k_B T_{\mathrm{BCS}}\gg
287: \hbar \omega$. However, for $k_F|a|=1/2$ or $1/3$, it gives a
288: result accurate to 30\%
289: for $k_B T_{\mathrm{BCS}}\simeq 2\hbar \omega$~\cite{baranov3}. In
290: the right axis of Fig.~\ref{fig:1} is shown the critical value $(k_F|a|)_c$
291: ensuring $T_{\rm final}=T_{\mathrm{BCS}}$, as a function
292: of $T_{\mathrm{BEC}}/T$:
293: $(k_F|a|)_c$ is on the order of one half, a regime
294: already accessible by present experiments.
295: It should be noted that our treatment is exactly valid only in the
296: regime $k_F |a| \ll 1$, as we have used BCS theory. The values of
297: $(k_F |a|)_c$ that we find are not extremely small compared to
298: unity; we therefore expect corrections to both the equation of
299: state of the normal phase and the critical temperature. Our result
300: must therefore be considered as an estimate.
301:
302: Fermi gases have been predicted to be sensitive to heating due to
303: particle losses, much more so than Bose
304: condensates~\cite{timmermans2}. One may ask what the limit is on the
305: accessible final temperature set by this heating during the
306: adiabatic tuning of $a$. Assume a loss rate
307: $\gamma_{\mathrm{loss}}$ of atoms due mainly to collisions with
308: the background gas, and let $t$ be the duration of the adiabatic
309: ramp. Using the derivation of Ref.~\cite{idziaszek2002}, and
310: assuming that $\gamma_{\mathrm{loss}} t \ll 1$ and
311: $(T/T_F)^2 \ll 1$, the temperature increase due to loss may be
312: estimated as $(T/T_F)_{\mathrm{increase}}\sim\sqrt{\gamma_{\mathrm{loss}} t/(2
313: \pi^2)}$. One may take $t\sim 1/[\gamma_{\mathrm{coll}}(T/T_F)^2]$, where
314: $\gamma_{\mathrm{coll}}=n\sigma v_F$ is the classical
315: collision rate involving the scattering cross-section
316: $\sigma\simeq 4\pi a^2$. Here $n$ is the total density,
317: $v_F\equiv \hbar k_F / m$ is the Fermi velocity,
318: and a factor of $(T/T_F)^2$ is included to account for
319: the Pauli blocking for each of the two components~\cite{ohara1}.
320: Then, supposing $(T/T_F)_{\mathrm{final}}\sim 10^{-2}$,
321: one finds the condition \be
322: \left[\gamma_{\mathrm{loss}}/(2\pi^2\gamma_{\mathrm{coll}})\right]^{1/4}\lesssim
323: (T/T_F)_{\mathrm{final}}\sim 10^{-2} \label{eqn:cri}\ee in order to avoid a
324: loss-induced increase in temperature during the switching time.
325: For the experiment of Ref.~\cite{salomon2003}, $\gamma_{\mathrm{loss}} < 10^{-2}$
326: s${}^{-1}$, and Eq.~(\ref{eqn:cri}) is satisfied for $a\gtrsim
327: 1000 a_0$ and $n\gtrsim 5\times 10^{12}$ cm$^{-3}$, where $a_0$ is the Bohr radius.
328:
329: In order to satisfy the criterion of adiabaticity,
330: thermal equilibrium must be maintained while tuning the scattering length from positive
331: to negative values.
332: This requires that the thermalization rate, {\it i.e.}, the rate of elastic collisions,
333: be much larger than the rate of change of $a$ \cite{proviso}.
334: For $a>0$, thermalization is due to collisions between non-condensed molecules.
335: As the size of the non-condensed cloud is close to the size of the condensate
336: for $k_B T \sim \mu_{\mathrm{mol}}$,
337: the effective collision rate is
338: thus
339: $\gamma_{\mathrm{eff}}=\gamma_{\mathrm{coll}}^{\mathrm{mol}}\, N_T/N_{\mathrm{mol}}$,
340: where $N_T/N_{\mathrm{mol}}$ is the non-condensed fraction and
341: $\gamma_{\mathrm{coll}}^{\mathrm{mol}}= n_{\mathrm{mol}} \sigma_{\mathrm{mol}} v_T$,
342: with $v_T$ the thermal velocity
343: and $\sigma_{\mathrm{mol}}=8\pi a_{\mathrm{mol}}^2$ the scattering
344: cross section between two molecules. For $T/T_{\mathrm{BEC}}=1/2$, the non-condensed fraction
345: is $(T/T_{\mathrm{BEC}})^3=1/8$~\cite{dalfovo1}.
346: Assuming $a\sim 1000 a_0$ and the mass of $^6$Li, even for
347: molecular densities $n_{\mathrm{mol}}$ as small
348: as $10^{12}$ cm$^{-3}$, this gives a minimum thermalization time of $\sim 30$ ms.
349: For $a<0$, the rate of thermalization is strongly reduced by Pauli blocking, and
350: one has $\gamma_{\mathrm{eff}}=\gamma_{\mathrm{coll}}(T/T_F)^2$~\cite{ohara1}.
351: At the critical value of $k_F|a|\sim 1/2$, and for a total density
352: of $5\times 10^{12}$ cm$^{-3}$, adiabaticity requires
353: that the change of the scattering length occurs on a time $t>300$ ms.
354: This time is longer than the inverse oscillation frequency
355: of lithium atoms in typical magnetic traps, so that the change in $a$ will not
356: induce macroscopic oscillations in the gas.
357:
358: It is thus far not possible to obtain a perfect balance of spin
359: states in a Fermi degenerate gas~\cite{salomon2003}. It may
360: therefore be supposed that a small fraction of fermions will
361: remain unpaired for positive scattering lengths, due to a lack of
362: partners, and will coexist with the molecular condensate.
363: %As the
364: %fermions are in a single spin state $s$-wave approximation they
365: %interact only with the molecules. The mean field plus trapping potential
366: %experienced by an unpaired fermion is
367: %$m\omega^2 r^2/2 + g_{am}\rho_{\mathrm{mol}}(r)$, where
368: %the atom-molecule coupling constant is $g_{am}=0.9
369: %g=0.9\times 4\pi\hbar^2 a/m$~\cite{petrov2003b} and the molecular mean field, in the
370: %Thomas-Fermi limit, is given by
371: %Eq.~(\ref{eqn:tf}), with
372: %$g_{\mathrm{mol}}=0.3 g$. They therefore experience an effective potential
373: %$U_{\mathrm{eff}}(r)=-2.5\,m\omega^2r^2 +3\mu_{\mathrm{mol}}$. This causes a spatial
374: %separation between atoms and molecules, with atoms pushed to the
375: %outside of the molecular mean field.
376: Fermions have an entropy
377: proportional to $T$, as opposed to $T^3$ in the case of bosons. So
378: even a small number of atoms could be expected to dominate the
379: total entropy of the atom-molecule mixture at low temperatures,
380: thereby interfering with the proposed cooling scheme.
381: This problem can
382: be avoided by simply removing the excess fermions from the system
383: for $a>0$ with a correctly tuned laser pulse, which requires that
384: their excitation frequencies are sufficiently different from those
385: of the molecules~\cite{jinPrivateCommunication2003}.
386:
387: In conclusion, we have shown that adiabatic switching
388: of a molecular BEC allows one to obtain a deeply
389: degenerate Fermi gas of temperatures on the order of $T\sim
390: 10^{-2} T_F$. This
391: suggests a way to achieve a BCS transition in the weakly
392: interacting regime. It is important to
393: note that our thermodynamic approach does not require a knowledge
394: of the equation of state of the system in the intermediate strong coupling
395: regime where $k_F|a| \gtrsim 1$.
396:
397: We thank Christophe Salomon, Deborah Jin and Carlos Lobo for useful comments.
398: This work was financially supported by NSF grant no.~MPS-DRF
399: 0104447, the French Minist\`ere des Affaires Etrang\`eres, by the
400: Dutch Foundations NWO and FOM, by INTAS, and by the Russian
401: Foundation for Fundamental Research. Laboratoire Kastler Brossel
402: is a research unit of Universit\'e Pierre et Marie Curie and Ecole
403: Normale Sup\'erieure, associated with CNRS (UMR 8552).
404:
405: {\it Note added}: Since submission of this paper, long-lived BEC's of
406: weakly bound molecules have been observed~\cite{Greiner03}, and a strongly interacting fermionic condensate in the BEC-BCS crossover regime
407: was created~\cite{regal2004}.
408:
409:
410: \vspace*{-6mm}
411:
412: \begin{thebibliography}{10}
413:
414: \bibitem[*]{byline1}
415: Present address: JILA, National Institute of Standards
416: and Technology and Physics Department, University of Colorado,
417: Boulder, CO 80309-0440
418:
419: \bibitem[$\dag$]{byline2}
420: Present address: Laboratoire de Physique Th\'eorique et Mod\`eles Statistiques,
421: Universit\'e Paris Sud, B\^at. 100, 91405 Orsay Cedex, France;
422: and Van der Waals - Zeeman Institute, University of Amsterdam,
423: Valckenierstraat 65/67, 1018 XE Amsterdam, The Netherlands
424:
425: \bibitem{jin3}
426: B. Demarco and D.~S. Jin, Science {\bf 285}, 1703 (1999).
427:
428: \bibitem{truscott1}
429: A. G. Truscott, K. E. Strecker, W. I. McAlexander, G. Partridge, and R. G. Hulet, Science {\bf 291}, 2570 (2001).
430:
431: \bibitem{schreck1}
432: F. Schreck, L. Khaykovich, K. L. Corwin, G. Ferrari, T. Bourdel, J. Cubizolles, and C. Salomon, Phys. Rev. Lett. {\bf 87}, 080403 (2001).
433:
434: \bibitem{granade2002}
435: S.~R. Granade, M.~E. Gehm, K.~M. O'Hara, and J.~E. Thomas,
436: Phys. Rev. Lett.
437: {\bf 88}, 120405 (2002).
438:
439: \bibitem{roati2002}
440: G. Roati, F. Riboli, G. Modugno, and M. Inguscio, Phys. Rev. Lett.
441: {\bf 89},
442: 150403 (2002).
443:
444: \bibitem{ohara1}
445: K. M. O'Hara, S. L. Hemmer, M. E. Gehm, S. R. Granade, and J. E. Thomas, Science {\bf 298}, 2179
446: (2002).
447:
448: \bibitem{hadzibabic2003}
449: Z. Hadzibabic, S. Gupta, C. A. Stan, C. H. Schunck, M. W. Zwierlein, K. Dieckmann, and W. Ketterle, Phys. Rev. Lett. {\bf 91}, 160401 (2003).
450:
451: \bibitem{vogels1}
452: J. M. Vogels, C. C. Tsai, R. S. Freeland, S. J. J. M. F. Kokkelmans, B. J. Verhaar, and D. J. Heinzen, Phys. Rev. A {\bf 56}, R1067 (1997).
453:
454: \bibitem{nozieres1985}
455: P. Nozi\`eres and S. Schmitt-Rink, J. Low Temp. Phys. {\bf 59},
456: 195 (1985).
457:
458: \bibitem{randeria1995}
459: M. Randeria, {\em Bose-Einstein Condensation} (Cambridge
460: University Press,
461: U.K., 1995), Chap.~15, pp.\ 355--392.
462:
463: \bibitem{holland2001}
464: M. Holland, S.~J. J. M.~F. Kokkelmans, M.~L. Chiofalo, and R. Walser,
465: Phys. Rev. Lett. {\bf 87}, 120406 (2001).
466:
467: \bibitem{ohashi2003}
468: Y. Ohashi and A. Griffin, Phys. Rev. A {\bf 67}, 033603 (2003).
469:
470: \bibitem{proviso}
471: One may wonder if the tuning time of $a$ should not also be longer than, e.g., the
472: formation time of the BCS state, which could exceed the collision time.
473: However, in our scheme, the gas is initially a Bose condensate
474: of molecules, so that no formation of a superfluid component from scratch
475: is required.
476:
477: \bibitem{dalfovo1}
478: F. Dalfovo, S. Giorgini, L.~P. Pitaevskii, and S. Stringari, Rev.
479: Mod. Phys. {\bf 71}, 463 (1999).
480:
481: \bibitem{salomon2003}
482: J. Cubizolles, T. Bourdel, S. J. J. M. F. Kokkelmans, G. V. Shlyapnikov, and C. Salomon, Phys. Rev. Lett. {\bf 91}, 240401 (2003); S. Jochim, M. Bartenstein, A. Altmeyer, G. Hendl, C. Chin, J. H. Denschlag, and R. Grimm, Phys. Rev. Lett. {\bf 91}, 240402 (2003); K.E. Strecker, G.B. Partridge, R.G. Hulet, Phys. Rev. Lett. {\bf
483: 91}, 080406 (2003); C.A. Regal, M. Greiner, and D.S. Jin, e-print
484: cond-mat/0308606 (2003).
485:
486: \bibitem{urban2003}
487: M. Urban and P. Schuck, Phys. Rev. A {\bf 67}, 033611 (2003).
488:
489: \bibitem{diu1989}
490: B. Diu, C. Guthman, D. Lederer, and B. Roulet, {\em Physique
491: Statistique}
492: (Hermann, Paris, France, 1989).
493:
494: \bibitem{petrov2003b}
495: D.~S. Petrov, C. Salomon, and G.V. Shlyapnikov, e-print cond-mat/0309010 (2003).
496:
497: \bibitem{gorkov1}
498: L.~P. Gor'kov and T.~K. Melik-Barkhudarov, Sov. Phys. JETP {\bf
499: 13}, 1018
500: (1961).
501:
502: \bibitem{baranov3}
503: M.~A. Baranov and D.~S. Petrov, Phys. Rev. A {\bf 58}, R801
504: (1998).
505:
506: \bibitem{chevy2002}
507: F. Chevy {\it et~al.}, Phys. Rev. Lett. {\bf 88}, 250402 (1997).
508:
509: \bibitem{jinPrivateCommunication2003}
510: Deborah Jin, JILA, Univ. of Colorado, private communication
511: (2003).
512:
513: \bibitem{timmermans2}
514: E. Timmermans, Phys. Rev. Lett. {\bf 87}, 240403 (2001);
515: L.~D. Carr, T. Bourdel, and Y. Castin, Phys. Rev. A in press (2004).
516:
517: \bibitem{idziaszek2002}
518: Z. Idziaszek, L. Santos, M. Baranov, and M. Lewenstein,
519: Phys. Rev. A {\bf 67}, 041403 (2003).
520:
521: \bibitem{Greiner03}
522: M. Greiner, C. Regal, and D.~S. Jin, Nature {\bf 426}, 537 (2003);
523: S. Jochim, M. Bartenstein, A. Altmeyer, S. Riedl, C. Chin, J. H. Denschlag, and R. Grimm, Science {\bf 302}, 2101 (2003);
524: M. W. Zwierlein, C. A. Stan, C. H. Schunck, S. M. F. Raupach, S. Gupta, Z. Hadzibabic, and W. Ketterle, Phys. Rev. Lett. {\bf 91}, 250401 (2003);
525: C. Salomon, private communication (2003);
526: M. Bartenstein, A. Altmeyer, S. Riedl, S. Jochim, C. Chin, J. H. Denschlag, R. and Grimm, e-print cond-mat/0401109 (2004).
527:
528: \bibitem{regal2004}
529: C.~A. Regal, M. Greiner, and D.~S. Jin, Phys. Rev. Lett. {\bf 92} 040403 (2004).
530:
531: \end{thebibliography}
532:
533: \end{document}
534:
535: