Experimental realization of

BCS-BEC crossover physics

with a Fermi gas of atoms
by
Cindy Regal1
A thesis directed by Dr. Deborah Jin1 and submitted to the Faculty of the
Graduate School of the University of Colorado in partial fulfillment of the
requirements for the degree of Doctor of Philosophy, Department of Physics
1

JILA, National Institute of Standards and Technology Quantum Physics

Division and University of Colorado, and Department of Physics, University of
Colorado, Boulder, CO 80309-0440, USA

December 2005

Abstract
This thesis presents experiments probing physics in the crossover between BoseEinstein condensation (BEC) and BCS superconductivity using an ultracold gas
of atomic fermions. Scattering resonances in these ultracold gases (known as
Feshbach resonances) provide the unique ability to tune the fermion-fermion interactions. The work presented here pioneered the use of fermionic Feshbach resonances as a highly controllable and tunable system ideal for studying the cusp
of the BCS-BEC crossover problem. Here pairs of fermionic atoms have some
properties of diatomic molecules and some properties of Cooper pairs. I present
studies of a normal Fermi gas at a Feshbach resonance and the work required
to cool the gas to temperatures where superfluidity in the crossover is predicted.
These studies culminated in our observation of a phase transition at the cusp
of the BCS-BEC crossover through condensation of fermionic atom pairs. I also
discuss subsequent work that confirmed the crossover nature of the pairs in these
condensates.

Acknowledgements
I could not have hoped to find a better place to pursue my graduate studies than JILA. I had the opportunity to work directly with and learn from an
array of great scientists. Debbie provided me with an exceptional model of a researcher. Debbie is the ultimate debugger; from constructing a plot that captures
the essence a technical problem to sorting out the important physics from a mess
of effects, she always knows how to resolve a problem. As a graduate student you
expect to be an underling, but as Debbies student I always felt that I had a say
in what I was working on and that there was always room for creativity. For this
opportunity I am extremely grateful.
I also had the privilege to work with one of the new great minds of atomic
physics, Markus Greiner. Markus brought an amazing store of technical knowledge
to our lab. From Markus I learned innumerable facts about optics but also gained
a very physical understanding of light. My successors, Jayson Stewart and John
Gaebler, also contributed to the work in this thesis. Although my time working
with them was short, I am sure they will do great work in the years ahead.
I also thank Jon Goldwin and Shin Inouye for always pausing to answer my
many questions and Michele Olsen for allowing me to help bolt together her UHV
system. Jianying Lang, Travis Nicholson, Tyler Cumby, and Margot Phelps all
deserve my respect for having the ability to discern whether my advice was useful
or plain wrong. Everyone in the machine shop and in the electronics shop never
left my questions unanswered in addition to building amazing components for our
experiment.
As with every thesis from the JILA ultracold atom groups, Eric and Carl
played a crucial role in my graduate work. I am indebted to Eric for letting me
work in his lab in the summer of 2000, where I came to understand what a great
place JILA is. Carl has been an amazing second advisor; I thank him for his
honest advice and for the many bagels I have stolen from his group on Saturdays.
As a graduate student, presenting work to Eric and Carl is always the ultimate
challenge; after going through their interrogation the questions I received after an
invited talk were never harder.
I have also had the unique opportunity to collaborate on a variety of problems

ii

with many theorists: John Bohn, Chris Ticknor, Chris Greene, Brett Esry, Murray
Holland, Stefano Giorgini, and Marilu Chiofalo. The amazing aspect of JILA is
that all these people are at or were visiting JILA during my graduate career, and
they always had time to answer my peculiar breed of theory questions.
To Scott Papp I owe the happiness that I have enjoyed as a graduate student.
I could not begin to convey how important Scott has been in my life over the past
five years. He is my best friend, and he always makes me laugh. He taught me
that my world would not fall apart if I did not quite finish Jackson E&M exercise
4(e) perfectly. He is an exceptional scientist, and I doubt there is anything he
cannot build. It has been incredibly useful in my graduate work to walk home
everyday with a talking electronics encyclopedia. Our many Sundays spent on
the trails and roads of Colorado are times I will never forget.
JILA is full of so many other people that I have come to know well, I could
not begin to name them all. However I do want to single out Josh Zirbel as a
great friend. I thank him for the many hours he spent helping us with homework
and his patience in tolerating Scott and me for so long. The members of the
Longmont symphony have been invaluable for keeping the violin and music in my
life. I also thank Colorado for being such a beautiful state; it is rare to see a sight
more breathtaking than a Rocky Mountain alpine meadow.
The thanks for reaching this point is not relegated to those with a direct role
in my years at JILA. My parents have always been there for me but still made
me make all my own decisions, an invaluable gift. In my family a common motto
was Never say die, up man and try, and I always took it as given that the only
honorable path was working towards a goal I felt was important to the world.
Still coming from a Minnesota family whose favorite pastimes include felling dead
trees and sawing them apart with handsaws, keeps one grounded, literally.
Most responsible for my entry to the world of physics were my professors at
Lawrence University. John Brandenberger, David Cook, Matt Stoneking, and Jeff
Collett filled their corner of the science hall with a genuine love of science and
of teaching physics. Their hard work in building an accomplished small physics
department attracted many students I looked up to and learned a great deal from.
I must especially thank John Brandenberger for convincing me that I could become
a physicist and that, above all, physics can be more interesting than biochemistry.

iii

The generosity of the Fannie and John Hertz foundation has provided me with
constant support throughout my graduate career. Thanks!

Contents
1 Introduction
1.1 Historical perspective . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 BCS-BEC crossover physics
2.1 Pairing in a Fermi gas of atoms
2.2 Varying interactions . . . . . .
2.3 Simple theory . . . . . . . . . .
2.4 Beyond T = 0 . . . . . . . . . .
2.5 Modern challenges . . . . . . .

1
1
8

.
.
.
.
.

11
11
13
15
18
19

3 Feshbach resonances
3.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 A specific example . . . . . . . . . . . . . . . . . . . . . . . . . .

20
20
23

4 Cooling a Fermi gas

4.1 Cooling 40 K . . . . . . . . . . . . . . . . . . . .
4.2 Measuring the temperature of a Fermi gas . . .
4.2.1 An optically trapped Fermi gas . . . . .
4.2.2 Momentum distribution thermometry . .
4.2.3 Impurity spin-state thermometry . . . .
4.2.4 Temperature in the BCS-BEC crossover

.
.
.
.
.
.

28
28
31
32
37
40
43

5 Elastic scattering at Feshbach resonances

5.1 Measuring the collision cross section . . . . . . . . . . . . . . . . .
5.2 Anisotropic expansion . . . . . . . . . . . . . . . . . . . . . . . .

45
45
47

iv

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.

.
.
.
.
.
.

CONTENTS

5.3
5.4

Measuring the mean-field interaction . . . . . . . . . . . . . . . .

K Feshbach resonance summary . . . . . . . . . . . . . . . . . .

40

6 Creating molecules
6.1 Magnetic-field association . . . . . . . . . .
6.2 Detecting weakly bound molecules . . . . . .
6.2.1 Rf spectroscopy . . . . . . . . . . . .
6.2.2 Magnetic-field modulation . . . . . .
6.3 Understanding molecule conversion efficiency
6.4 A precise measurement of B0 . . . . . . . .

49
52

.
.
.
.
.
.

54
55
57
57
61
63
65

.
.
.
.

67
68
69
71
72

8 Condensates in the crossover

8.1 A molecular condensate . . . . . . . . . . . . . . . . . . . . . . .
8.2 Observing condensates in the crossover . . . . . . . . . . . . . . .
8.3 Measurement of a phase diagram . . . . . . . . . . . . . . . . . .

74
75
80
85

9 Momentum distributions in the crossover

9.1 Measuring the momentum distribution . .
9.2 Extracting the kinetic energy . . . . . . .
9.3 Comparing the kinetic energy to theory . .
9.4 Temperature dependence . . . . . . . . . .

.
.
.
.

88
90
92
94
95

.
.
.
.
.
.

97
97
103
106
112
115
116

7 Inelastic collisions
7.1 Expected inelastic decay processes
7.2 Lifetime of Feshbach molecules . .
7.3 Three-body recombination . . . .
7.4 Comparison of 40 K and 6 Li . . .

10 The
10.1
10.2
10.3
10.4
10.5
10.6

apparatus
Making an ultracold 40 K gas
Diode laser system . . . . .
A stable optical dipole trap
Large magnetic-field control
Fast magnetic-field control .
Rf delivery . . . . . . . . . .

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.
.
.

CONTENTS

vi

10.7 Absorption imaging . . . . . . . . . . . . . . . . . . . . . . . . . . 118

11 Conclusions and future directions
122
11.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
11.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
A Journal articles

140

Chapter 1
Introduction
1.1

Historical perspective

The phenomenon of superconductivity/superfluidity has fascinated and occupied

physicists since the beginning of the 20th century. In 1911 superconductivity was
discovered when the resistance of mercury was observed to go to zero below a critical temperature [1]. Although liquid 4 He was actually used in this discovery, the
superfluid phase of liquid 4 He was not revealed until the 1930s when the viscosity
of the liquid below the point (2.17 K) was measured [2, 3]. Much later, 3 He,
the fermionic helium isotope, was also found to be superfluid at yet a much colder
temperature than 4 He [4]. Relatively recently in 1986, high-temperature superconductors in Copper-oxide materials further enlarged the list of superconducting
materials [5].
These super systems, which I will refer to in general as superfluids, are listed
in Fig. 1.1, but they are only classic examples. There are many other physical
systems that display superfluid properties from astrophysical phenomena such
as neutron stars, to excitons in semiconductors, to atomic nuclei [6]. Although
the physical properties of these systems vary widely, they are all linked by their
counterintuitive behaviors such as frictionless flow and quantized vorticity. The
manifestation of these effects depends upon, for example, whether the system in
question is electrically charged (superconductors) or neutral (superfluids). Besides these intriguing properties, there are many practical reasons for the intense
research in this field; arguably the most useful super-systems are superconduc1

CHAPTER 1. INTRODUCTION

tors, and if a robust room-temperature superconductor were created it would be

an amazing discovery.

Tc / TF

100

Alkali BEC
4
Superfluid He
BCS-BEC crossover regime
High-Tc superconductors
3
Superfluid He
Superconductors

10-2
10-4
10-6

1010

105
100
2Egap/kBTF

10-5

Figure 1.1: Classic experimental realizations of superfluidity/superconductivity

arranged according to the binding energy (twice the excitation gap, Egap ) of
the constituent fermions. The vertical axis shows the corresponding transition
temperature, Tc , to a superfluid/superconducting state compared to the Fermi
temperature, TF . (Figure reproduced with permission from Ref. [7].)

Some of the first attempts to understand the phenomenon of superfluidity were

in the context of Bose-Einstein condensation (BEC) of an ensemble of bosonic
particles [8]. BEC is a consequence of the quantum statistics of bosons, which
are particles with integer spin, and it results in a macroscopic occupation of a
single quantum state (Fig. 1.2) [9, 10, 11]. Fritz London proposed in 1938 that
superfluid 4 He was a consequence of Bose-Einstein condensation of bosonic 4 He
[12]. (4 He is a boson because it is made up of an even number of 1/2 integer
spin fermions - electrons, protons, and neutrons.) Physicists such as Blatt et al.
pushed a similar idea in the context of superconductors in proposing that at
low temperature, charge carrying bosons occur, e.g., because of the interaction of
electrons with lattice vibrations [13]. For the case of tightly bound bosons, such
as 4 He, Londons hypothesis turned out to be correct. However, the very strong
interactions in 4 He made it difficult to verify the existence of condensation [14],
and for many years 4 He studies did not mention BEC.

CHAPTER 1. INTRODUCTION

In the case of superconductors, discussion of BEC was overshadowed by the

amazing success in 1957 of the Bardeen-Cooper-Schrieffer (BCS) theory of superconductivity [15, 16]. In 1956 Cooper found that a pair of fermions in the
presence of a filled Fermi sea (Fig. 1.2) will form a bound pair with an arbitrarily
small attractive interaction [17]. The BCS theory solved this problem in the case
where many pairs can form in the Fermi sea. The result predicted (among other
things) the formation of a minimum excitation energy, or energy gap, in the conductor below a critical temperature Tc . A great many properties of conventional
superconductors can be understood as consequences of this energy gap.
Figure 1.1 sorts the classic superfluid systems according to the strength of
the interaction between the fermions. A key aspect of the classic BCS theory is
that it applies to the perturbative limit of weak attractive interactions and hence
is only an exact theory for the far right side of Fig. 1.1. The theory perfectly
described conventional superconductors for which the attraction between fermions
is 10,000 times less than the Fermi energy, EF . The BCS ground state was also
able to accurately describe superfluid 3 He and many (although arguably not all
[18]) aspects of high-Tc superconductors. The theory in its original form, however,
did not at all apply to the case of the tightly bound boson, 4 He.

bosons:
integer spin

fermions:
half-integer spin

EFermi
spin
spin
Figure 1.2: Quantum statistics: Bosons versus fermions with weak interactions at
T = 0. Bosons form a BEC in which all of the bosons macroscopically occupy a
single quantum state. Due to the Pauli exclusion principle fermions form a Fermi
sea in which each energy state up to the Fermi energy is occupied.

In 1995 a completely new system joined 4 He on the left side of Fig. 1.1.

CHAPTER 1. INTRODUCTION

Here the composite bosons were alkali atoms, such as 87 Rb, that had been cooled
as a gas down to nanoKelvin temperatures via laser cooling and evaporative
cooling [19, r
20, 21]. At these temperatures the thermal deBroglie wavelength
2

2
h
) of the particles becomes on order of the interparticle spacing
(deBroglie = mk
bT
in the gas and a Bose-Einstein condensate is formed (Fig. 1.3(a)) [22, 23]. In
contrast to 4 He the alkali BEC that was created was weakly interacting, making
the condensation stunningly visible, as shown in Fig. 1.3(b). Experiments also
observed that condensates behave as coherent matter waves [24] and verified the
superfluid nature of a condensate [25, 26]. In this way both the BEC and superfluid properties could be clearly seen in one system and understood extremely
well theoretically [27]. However, the fact that the dilute gas BEC was found to be
a superfluid was not at all a surprise. Although a long and complicated history
was required, physicists now understand the basic connection between BEC and
superfluidity [28]. It is commonly accepted that superfluidity is always intimately
connected to the macroscopic occupation of a quantum state.
Besides providing the first clear evidence for BEC, ultracold alkali gases opened
a world of possibilities for studying superfluid systems. Many of the initial experiments with alkali BEC could be perfectly described by existing theories. However,
recent work in the field of BEC has developed techniques to reach a regime that
is more relevant for the outstanding theoretical questions in condensed matter
physics, which are most commonly in strongly correlated systems. For example,
experiments achieved BEC with much stronger interatomic interactions than typical alkali gases; furthermore, these interactions could even be controllably tuned
[29, 30]. A phase transition to the highly-correlated Mott insulator state was observed through studies of quantum gases in optical lattice potentials [31]. These
bosonic systems require theory that goes beyond mean-field interactions; yet they
have a controllability rarely found in solid state materials.
At the same time as the creation of the first strongly interacting Bose gases, the
techniques used to create alkali BECs were applied to the other class of quantum
particles, fermions. To create a Fermi gas of atoms experimenters applied the
same cooling techniques as those used to achieve BEC, simply replacing a bosonic
atom, such as 87 Rb or 23 Na, with an alkali atom with an odd number of electrons,
protons, and neutrons. (The two such stable alkali atoms are 40 K and 6 Li.) Still,

CHAPTER 1. INTRODUCTION

(a)

(b)

T<Tc

Tc: d ~ ldeBroglie
T>Tc

Figure 1.3: Bose-Einstein condensation in a dilute gas of 87 Rb atoms. (a) Phase

space density criterion for condensation. (b) Momentum distributions of 87 Rb
atoms at three values of the temperature compared to the critical temperature.
(Figure adapted by M. R. Matthews from data in Ref. [22].)

evaporatively cooling fermions required ingenuity. Due to the quantum statistics

of fermions the s-wave collisions required for evaporative cooling are not present at
ultracold temperatures in a gas of spin-polarized, identical fermions. The solution
to this problem was to introduce a second particle for the evaporative cooling,
either another state of the fermionic atom or another species entirely. The first
gas of fermionic atoms to enter the quantum degenerate regime was created at
JILA in 1999 using 40 K [32]. The observation in these experiments was not a
phase transition, as in the Bose gas, but rather the presence of more and more
energy than would be expected classically as the Fermi gas was cooled below the
Fermi temperature. Many more Fermi gas experiments, using a variety of cooling
techniques, followed [33, 34, 35, 36, 37, 38, 39, 40, 41].
The next goal after the creation of a normal Fermi gas of atoms was to form a
superfluid in a paired Fermi gas. In conventional superconductors s-wave pairing
occurs between spin-up and spin-down electrons. The hope was that s-wave pairing could similarly occur with the creation of a two-component atomic gas with
an equal Fermi energy for each component. Such a two-component gas can be realized using an equal mixture of alkali atoms in two different hyperfine spin states.
The simplistic view was that a BCS state would appear if the temperature of this
two-component gas were cold enough and the interaction between fermions attractive and large enough. However, for typical interactions the temperatures required

CHAPTER 1. INTRODUCTION

to reach a true BCS state were far too low compared to achievable temperatures
(at that point) to imagine creating Cooper pairs. Stoof et al. noted that the
interaction between 6 Li atoms was large compared to typical values (|a| 2000
a0 ), as well as attractive, bringing the BCS transition temperature closer to realistic temperatures [42, 43]. It was then recognized that a type of scattering
resonance, known as a Feshbach resonance, could allow arbitrary changes in the
interaction strength. Theories were developed that explicitly treated the case
where the interactions were enhanced by a Feshbach resonance [44, 7, 45].
In these proposals, however, increasing the interactions beyond the perturbative limit of BCS theory meant that the physical system would not be simply a
BCS state, but rather something much more interesting. It was predicted that
a Feshbach resonance would allow the realization of a system with an excitation
gap on the order of the Fermi energy and provide the ability to continually tune
within this region (green box in Fig. 1.1) [7]. This system, if achievable, would
be the experimental realization of a theoretical topic that dated back to the late
1960s, the BCS-BEC crossover. In a theory originally put forth by Eagles and
later by Leggett, it was proposed that the BCS wavefunction was more generally
applicable than just to the weakly interacting limit [46, 47]. As long as the chemical potential is found self-consistently as the interaction is increased, the BCS
ground state can (at least qualitatively) describe everything from Cooper pairing to BEC of composite bosons made up of two fermions, i.e., the fundamental
physics behind all of the systems in Fig. 1.1 [46, 47, 48, 49, 50, 51]. After nearly a
century of 4 He and superconductors being considered separate entities, an experimental realization of a superfluid in the BCS-BEC crossover regime would provide
a physical link between the two. More recent interest in crossover theories has
also come in response to the possibility that they could apply to high-Tc superconductors. These superconductors differ from normal superconductors both in their
high transition temperature and the apparent presence of a pseudogap, which are
both characteristics expected to be found in a Fermi gas in the crossover [8, 18].
Thus, starting in about 2001 (the year I started work on this thesis) a major
goal in dilute Fermi gas experiments was to achieve a superfluid Fermi gas at a
Feshbach resonance, often referred to as a resonance superfluid [7]. However,
achievement of this experimental goal was a number of years and many steps

CHAPTER 1. INTRODUCTION

away. The existence of Feshbach resonances had been predicted by atomic physicists both in the 6 Li and in the 40 K systems [52, 53], and the first step was to
locate these resonances [54, 55, 56, 57]. Subsequent experimental studies appeared
at an amazingly fast rate with contributions from a large number of groups, in
particular those of R. Grimm (Innsbruck), R. Hulet (Rice), D. Jin (JILA), W.
Ketterle (MIT), C. Salomon (ENS), and J. Thomas (Duke). Experimenters discovered interesting properties of the normal state of a strongly interacting Fermi
gas [58, 59, 60, 61, 62]. Then Fermi gases were reversibly converted to gases of
diatomic molecules using Feshbach resonances [63, 64, 65, 66]. The observation
that these molecules were surprisingly long-lived created many opportunities for
further study [64, 66, 65, 67]. Condensates of diatomic molecules in the BEC
limit were achieved [68, 69, 70, 71, 72]; then these condensates were found to exist
in the crossover regime [73, 74], signalling the existence of a phase transition in
the BCS-BEC crossover regime. Collective excitations [75, 76, 77] and thermodynamic properties [39, 71, 78, 79] were also measured, and the nature of the pairs
was probed in a variety of ways [80, 81, 72]. Most recently a vortex lattice was
even created in the crossover [82].
Developing these techniques to access and probe the BCS-BEC crossover was
a challenging adventure for the field. Experiments in the crossover are inherently
difficult because the strong interactions make probing difficult. Some of the techniques used in the end were borrowed from those developed for alkali BEC, while
others were taken from condensed matter physics. Some were new inventions
that relied on the unique ability to tune the interaction at arbitrary rates using a
Feshbach resonance.
So far the experiments that have been carried out with dilute Fermi gases
near Feshbach resonances have been qualitatively consistent with classic BCSBEC crossover theory. The excitation gap is on the order of the Fermi energy;
the system crosses a phase transition to a superfluid state. Quantitatively though
there is much work to be done. In tandem with these experiments, sophisticated
theories of the crossover have been developed that are too numerous to list here,
but are actively being pursued in groups such as those of A. Bulgac, J. Carlson,
M. Chiofalo, S. Giorgini, A. Griffin, H. Heiselberg, T.-L. Ho, M. Holland, K.
Levin, S. Strinati, B. Svistunov, and E. Timmermans. In time it is expected that

CHAPTER 1. INTRODUCTION

the BCS-BEC crossover system provided by dilute Fermi gases should be able to
rigorously test these theories.
The power to test these many-body theories comes from the ability to create
a very clean strongly interacting Fermi system. In principle the density and
two-fermion interaction in the sample can be known precisely and the s-wave
pairing fully characterized as a function of the interaction strength. The end
result could be a fully understood physical system that connects the spectrum
of pairing from BCS to BEC, uniting the basic physics surrounding super
systems. On the other hand, the complicated materials physics involved in, for
example high-Tc superconductors, cannot be elucidated in these clean crossover
experiments. Still the hope is that an understanding of the basic physics will
provide a solid foundation for studies of real materials.

1.2

Contents

In this thesis I will present some of the first experimental work studying fermions
in the BCS-BEC crossover in a dilute atomic system. The contents come from
experiments co-workers and I performed in the group of D. Jin at JILA between
2001 and 2005 using 40 K atoms. As outlined above this was an exciting time in the
field of Fermi gases. In 2001 in our lab at JILA the technology existed to create
two-component Fermi gases at temperatures around 0.25 TF [83]. Predictions had
been made for the existence of a superfluid state near a Feshbach resonance for
a Fermi gas on the order of this temperature [44, 7, 45]. Yet in the early days of
this work many physicists were skeptical about the feasibility of experimentally
realizing such a state. The Feshbach resonances that had been observed in Bose
gases were associated with extremely fast inelastic decay of the trapped gas [29, 84,
85]. These decay processes, which most often stem from three-body collisions, can
quickly turn a hard-earned quantum gas into a classical gas of atoms [86, 87, 88,
89]. Carl Wiemans group at JILA produced the only experiments studying BEC
near a Feshbach resonance over long time scales [30]. In this group 85 Rb BECs
were studied at very low densities, where two-body elastic collisions dominate over
three-body collisions. For two-component Fermi gases three-body decay processes
were expected to be suppressed compared to the Bose case [90, 91], due to the

CHAPTER 1. INTRODUCTION

Pauli exclusion principle. Still there was a fair amount of contention about the
degree of this suppression.
Further difficulties stemmed from the fact that not all researchers in the atomic
physics community were familiar with theories from condensed matter physics,
such as BCS-BEC crossover theory. Even when these theories were understood
there was significant confusion about the relation between classic crossover theory
and the physics at Feshbach resonances in 6 Li and 40 K. An additional challenge
was creating a sufficiently cold Fermi gas near a Feshbach resonance. To be certain
of achieving the predicted phase transition, temperatures well below the predicted
Tc would have to be achieved. A technical challenge was that the states required
to access Feshbach resonances in the 40 K and 6 Li gases could not be confined in
a magnetic trap, which was the most proven trap in studies of ultracold gases up
to that point. Instead the experiments would have to achieve an ultracold Fermi
gas in an alternative trap, such as an optical dipole trap.
Much of the work to create this thesis involved overcoming these difficulties
and sources of confusion, and as we shall see there are now clear answers to
many of these questions. These answers were found through careful studies of
the normal state of a Fermi gas near a Feshbach resonance and significant work
to cool a 40 K gas to the coldest temperatures possible. In late 2003 we observed
a phase transition in the crossover and since then have been able to study BCSBEC crossover physics with a Fermi gas of atoms. In this thesis I will not include
every aspect of the experiments we completed; instead I will concentrate on our
key contributions to the understanding of BCS-BEC crossover physics in Fermi
gases. The full range of experiments is outlined in Appendix A through a list of
published articles.
In the first few chapters I discuss theory that is necessary to understand experiments presented in later chapters. The goal is not to rigorously derive modern
theories, but rather to convey the mindset that many experimentalists currently
use to think about the crossover problem with atomic Fermi gases. Chapter 2
presents the ideas of BCS-BEC crossover physics through simplified theory. Chapter 3 introduces Feshbach resonances and discusses how well the Feshbach resonance systems reproduce conditions for the classic universal BCS-BEC crossover
problem.

CHAPTER 1. INTRODUCTION

10

The main body of the thesis presents experiments using an ultracold gas of
K atoms. Chapter 4 describes cooling methods and temperature measurement
techniques. Chapter 5 contains experiments that probe the presence of Feshbach
resonances in the 40 K system and study their ability to tune atomic scattering
properties. Chapter 6 introduces the creation of molecules from a Fermi gas of
atoms, which is the analog of the BCS-BEC crossover in the normal state. Chapter 7 concentrates on the stability of fermionic atoms and pairs against inelastic
collisions in the presence of a Feshbach resonance; this is a subject of practical
importance to the ability to study the BCS-BEC crossover with a Fermi gas of
atoms. Chapter 8 describes the first experiments to observe a phase transition in
a Fermi gas of atoms in the BCS-BEC crossover regime. Chapter 9 focuses on
a measurement of the momentum distribution of the pairs in the crossover; this
measurement is an important probe of the nature of the pairs in the crossover.
Chapter 10 discusses some of the most relevant additions and changes to the apparatus of Ref. [32] that were required to access the BCS-BEC crossover physics
described in this thesis.
40

Chapter 2
BCS-BEC crossover physics
To understand the experimental work presented in this thesis it is helpful to be
familiar with BCS-BEC crossover physics. In this chapter I present the theory
of the BCS-BEC crossover, first from a purely qualitative point of view and then
from a slightly more quantitative perspective. In this quantitative perspective it
is not my goal to present the most sophisticated theory, but rather a theory that
introduces important parameters and illustrates the key differences between the
crossover problem and the BCS limit. Lastly I discuss some of the unanswered
questions related to the crossover problem; this conveys a sense of the importance
of experimental studies of crossover physics.

2.1

Pairing in a Fermi gas of atoms

As we have seen, superfluids are fundamentally associated with the quantum

properties of bosons. Since all visible matter is made up of fermions, creating
a superfluid most often requires pairing of fermions. The simplest (although
historically not the most famous) way to imagine pairing fermions is to create
a two-body bound state of the two fermions. Two half-integer spin fermions
when paired will produce an integer spin particle, which is a composite boson. In
the case of experiments discussed in this thesis the fermionic particles are atoms;
this makes such a two-body bound state a diatomic molecule. Below a critical
temperature an ensemble of these diatomic molecules will form a BEC. The left
side of Fig. 2.1 represents a superfluid containing these type of pairs. The two
11

12

CHAPTER 2. BCS-BEC CROSSOVER PHYSICS

colors represent fermions in two different spin states; two states are required if
the fermions are to pair via s-wave (l = 0) interactions.

BEC

diatomic molecules

BCS

strongly interacting pairs

Cooper pairs

Figure 2.1: Cartoon illustration of the continuum of pairing in the BCS-BEC

crossover.
In other pairing mechanisms, such as Cooper pairing, the underlying fermionic
nature of the system is much more apparent. Cooper considered the problem of
two fermions with equal and opposite momentum outside a perfect Fermi sea [17].
The energy of the two fermions turns out to be less than the expected value of 2 EF
for arbitrarily weak attractive interactions. This result is in surprising contrast to
the result of the problem of two fermions in vacuum; in this case there will not be
a bound state until the interaction reaches a certain threshold. The key difference
between the two situations arises from Pauli blocking, which in the Cooper pair
case prevents the two fermions under consideration from occupying momentum
states k < kF , where kF is the Fermi wavevector [92].
Considering only one pair of electrons as free to pair on top of a static Fermi
sea is not a sufficient solution to the pairing problem. All fermions should be
allowed to participate in the pairing, and we expect that pairs should form until
an equilibrium point is reached. At this equilibrium point the remaining ensemble
of fermions is disturbed enough from a Fermi sea configuration to no longer lead
to a bound state at the given interaction strength [93]. The BCS state is an
approximate solution to this many-body problem. A description of the full BCS
theory is beyond the scope of this current discussion, but is presented in the
original papers [15, 16] and discussed in numerous books, for example Refs. [92,
93]. Qualitatively the BCS state consists of loose correlations between fermions

CHAPTER 2. BCS-BEC CROSSOVER PHYSICS

13

across the Fermi surface in momentum space (Fig. 2.1 right side). Spatially
the pairs are highly overlapping and cannot simply be considered to be composite
bosons. In the BCS limit the momentum distribution only changes from the usual
Fermi sea in an exponentially small region near the Fermi surface.
It is interesting to consider what happens if diatomic molecules become more
and more weakly bound, to the point where the binding energy of the molecules,
Eb , becomes less than the Fermi energy, EF . One could also consider increasing
the interaction energy of a Cooper paired state until it is close to EF . The essence
of the BCS-BEC crossover is that these two sentences describe the same physical
state. As the interaction between fermions is increased there will be a continual
change, or crossover, between a BCS state and a BEC of diatomic molecules. The
point where two fermions in vacuum would have zero binding energy is considered
the cusp of the crossover problem, and pairing in such a state is represented in the
middle of Fig. 2.1. These pairs have some properties of diatomic molecules and
some properties of Cooper pairs. Many-body effects are required for the pairing,
as with the BCS state, but there is some amount of spatial correlation, as with
diatomic molecules. The pair size is on the order of the spacing between fermions,
and the system is strongly interacting.

2.2

Varying interactions

It is instructive to consider a physical situation that will allow the realization

pairing throughout the crossover (Fig. 2.1) [47]. Suppose we start with an attractive potential between two atomic fermions in vacuum, such as the square well
potential with characteristic range r0 shown in Fig. 2.2. If this potential is very
shallow there is a weak attractive interaction between the fermions. If we make
this potential deeper the interaction between fermions becomes stronger, and for
a strong enough attraction a bound molecular state will appear. This molecule
will become more and more deeply bound as the potential becomes deeper.
The interaction in this system can be characterized by the s-wave scattering
length a. The quantity a comes out of studying two-body, low-energy scattering
and is related to the s-wave collision cross section through = 4a2 . The top
of Fig. 2.2 shows a pictorial representation of a. Just before the bound state

14

CHAPTER 2. BCS-BEC CROSSOVER PHYSICS

a >0

r
a <0

r0

a >0
repulsive

r0

a <0
attractive
-1/a

Eb

Figure 2.2: Scattering wavefunctions in the presence of an attractive potential

(right) and a more deeply attractive potential (left), in a regime where a weakly
bound state of the potential (dashed line) is near threshold. r here describes the
relative position of two distinguishable fermions. The scattering length diverges
as the bound state moves through threshold.

CHAPTER 2. BCS-BEC CROSSOVER PHYSICS

15

appears a is large and negative, corresponding to a strong attractive interaction.

As the bound state moves through threshold a diverges and then becomes large
and positive, corresponding to a strong repulsive interaction. When a is much
larger than r0 , the interaction is independent of the exact form of the potential,
and a > 0 is universally related to the binding energy of the two-body bound

h2
state through Eb = ma
2 , where m is the mass of a fermion [94].
Now if we consider an ensemble of many fermions under the situation in Fig.
2.2, we have a system that can be tuned from BCS to BEC simply by tweaking the
attractive potential. To the far right we have a small negative a and thus the BCS
limit. In the opposite limit we have an ensemble of diatomic molecules, hence the
BEC limit. It is important to note that although the interaction between fermions
in pairs is strongest in the BEC limit, from the point of view of collisions in the
gas the BEC limit is actually weakly interacting because the molecule-molecule
interaction is weak.1 The most strongly interacting gas from the point of view
of collisions occurs near the divergence of a. Here many-body calculations are
difficult because there is no small expansion parameter. The precise point at
which a diverges is known as unitarity. Here the only length scale in the problem
is 1/kF , giving this point many unique properties [96, 97, 98, 99].

2.3

Simple theory

BCS theory was originally applied in the limit where the interaction energy is
extremely small compared to the Fermi energy. In this case the chemical potential,
, can be fixed at EF , and many calculations become reasonably simple. Leggett
pointed out that if the BCS gap equation is examined allowing to vary, the
gap equation actually becomes precisely the Schrodinger equation for a diatomic
molecule in the limit where dominates [47]. Below we briefly illustrate the
key steps in an application of BCS theory to the entire crossover. This gives
qualitatively correct results for the entire spectrum of pairing. The structure of
the crossover theory below originates in the work of Nozieres and Schmitt-Rink
(NSR) in Ref. [48] and of Randeria et al. in Ref. [49].
1

In the limit a r0 the molecule-molecule scattering length amm is predicted to be 0.6aaa ,

where aaa is the scattering length for atoms scattering above threshold [95].

CHAPTER 2. BCS-BEC CROSSOVER PHYSICS

16

Let us consider a homogeneous Fermi system in three dimensions in an equal

mixture of two different states at T = 0. Application of usual BCS theory results
in the gap equation
X
k
k =
Ukk
(2.1)
2Ek
k
q

2 2

where Ek = k2 + 2 , k = k , and k = h2mk . Ukk < 0 is the attractive

interaction for scattering of fermions with momenta k and k to k and k. We
also obtain the number equation
hNtot i =

X
k

(1

k
)
Ek

(2.2)

where Ntot is the total number of fermions in both states.

To solve Eqn. 2.1 in the BCS limit the standard approach is to assume that
the potential is constant at a value U < 0, which implies that the gap is constant
as well, i.e., k = . In this case the gap equation (Eqn. 2.1) becomes

X 1
1
=
.
U
k 2Ek

(2.3)

One will find, however, that this equation diverges. For BCS superconductors
this problem is resolved because the interaction can be limited to within the
Debye energy, h
D , of EF . This is a result of the nature of the phonon-mediated
interaction between the electrons that gives rise to the attractive interaction.
Further simplifications in the BCS limit are that = EF and that, since h
D
EF , the density of states is constant at the value N ( = 0). The gap equation
then becomes
Z hD
d
1
2
.
(2.4)
=

N (0)U

hD 2 + 2
Solving Eqn. 2.4 produces the BCS result 2
hD e1/N (0)|U | .
To extend this calculation to the crossover in atomic systems we can no longer
apply the h
D cutoff. The solution to the divergence problem in this case is
nontrivial and requires a renormalization procedure, the full description of which
we will not present here (but see, for example, Ref. [8] and references therein).

17

CHAPTER 2. BCS-BEC CROSSOVER PHYSICS

The result of such a procedure is a renormalized gap equation

m
1 X 1
1

)
(
2 =
V k 2Ek 2k
4
ha

(2.5)

where the interaction is now described by the s-wave scattering length a instead
of U and V is the volume of the system. Furthermore, in the crossover we cannot
assume = EF ; instead we must solve the gap equation (Eqn. 2.5) and number
equation (Eqn. 2.2) simultaneously for and the gap parameter . We will solve
for these parameters as a function of the dimensionless parameter kF a, where

h. As pointed out in a useful paper by M. Marini et al. this can

kF = 2mEF /
actually be done analytically using elliptic integrals [100].

2.0

BEC limit
BCS limit

m/EF

D/EF

1.5
1.0
0.5
0

1/kFa

-1

-2

1
0
-1
-2
-3
-4
-5
-6

-1

-2

1/kFa

Figure 2.3: Gap parameter, , and chemical potential, , of a homogeneous Fermi

gas at T = 0 as determined through NSR theory. The red and blue lines show the
BCS and BEC limits of the theory. Note that the limiting theories only deviate
significantly from the full theory in approximately the range 1 < kF1 a < 1.
The black lines in Fig. 2.3 show the result of this calculation of and .
We also plot the values of both and as they would be calculated in the BCS
and BEC limits to find that the crossover occurs in a relatively small region of
the parameter 1/kF a, namely from approximately 1 < 1/kF a < 1. In typical
crossover experiments with 40 K or 6 Li, this regime corresponds to varying a from
2000 a0 through and to 2000 a0 , where a0 is the Bohr radius (0.0529 nm).

18

CHAPTER 2. BCS-BEC CROSSOVER PHYSICS

It is useful to explicitly understand the value and meaning of both and

in the two limits. is EF in the BCS limit and Eb /2 = ( kF1 a )2 EF in the

16 1
in the BEC limit
BEC limit. is e kF |a| EF in the BCS limit and EF 3
kF a
[100]. Although referred to as the gap parameter, only has meaning as the
excitation gap, i.e., the smallest possible energy that can create a hole (remove a
fermion) in the superfluid,
in the BCS limit.
In general the excitation energy is

2 2

( h2mk )2 + 2 [8]. This is when is positive (as

in the BCS limit), but becomes 2 + 2 when is negative.

Egap = min Ek = min

2.4

Beyond T = 0

The phase transition temperature, Tc , is an important parameter for any superfluid system. In the BCS-BEC crossover the transition temperature increases as
the interaction is increased, i.e., it is lowest in the perturbative BCS regime and
highest in the BEC limit (Fig. 1.1). In a homogeneous system, in the BCS limit

Tc /TF = 8 e2 e 2kF |a| where = 0.58 [51], and in the BEC limit Tc /TF = 0.22
[101]. Note that BCS transition temperatures can be extremely small due to
the exponential dependence upon 1/kF a. For example, at a typical interaction
strength in alkali fermionic gases (a = 100 a0 ) and a typical kF (1/kF = 2000
a0 ) the BCS transition temperature would be 1014 TF , which is a completely
inaccessible temperature in contemporary atomic systems. Still, at 1/kF a = 1
where BCS theory nearly applies, the transition temperature is on the order of
0.1 TF , which is accessible in current atomic systems.
In the BCS limit pairing and the phase transition to a superfluid state occur
at the same temperature. However, in the BEC limit this is not the case; because
the constituent fermions are very tightly bound, pairs can form far above Tc . It is
natural to expect that there would be a crossover between these two behaviors in
the BCS-BEC crossover, i.e., at the cusp of the crossover the pairing temperature,
T , would be distinct from Tc , yet not far from it. In a simple picture T should
be related to a pair dissociation temperature, which in the case of molecules is
Eb /kb [8]. It is important to differentiate between the superconducting order
parameter, which exists below Tc , and the pairing gap, which exists below T
[18]. The pairing gap is associated with so-called preformed pairs, which are pairs

CHAPTER 2. BCS-BEC CROSSOVER PHYSICS

19

that are not yet phase coherent. Aspects of the pseudogap observed in high-Tc
superconductors may be a manifestation of pre-formed pairs [18].

2.5

Modern challenges

The discussions and calculations above, while providing an introduction to basic

crossover theory, are far from the state-of-the-art for theory in the field. There are
noticeable problems with the NSR theory presented in Sec. 2.3. For example, the
result for the chemical potential at unitarity (1/kF a = 0) is significantly different
from the result of more accurate calculations using full Monte Carlo simulations
[102, 103]. The chemical potential is often written as = (1 + )EF , and the
NSR theory produced = 0.41 at unitarity, while the Monte Carlo simulation
of Ref. [103] finds = 0.58 0.01 at unitarity. As another example, extension
of the NSR theory predicts that amm = 2aaa , while a full 4-body calculation
in the BEC limit yields amm = 0.6aaa [95]. (Experiments at ENS have shown
that amm = 0.6+0.3
0.2 aaa [71], in agreement with the full calculation but not the
NSR theory.) Both of these problems point to the fact that the NSR ground
state, which only includes two-particle correlations, is not sufficient to accurately
describe the system. Thus, it is clear that adding higher-order correlations to
BCS-BEC crossover theory is necessary, yet not a simple task [104].
An even greater challenge is to extend accurate theories to nonzero temperature where predictions can be made about the critical temperatures and the role
of pre-formed pairs. Furthermore, all of the calculations considered thus far are
carried out for a homogeneous Fermi system. However, the experiments with
ultracold gases take place in traps (most often harmonic traps). The use of an
inhomogeneous density gas can lead to qualitative changes in the system that
must be accounted for in theory: Strong interactions can modify the density of
the trapped gas [105], and signals can become blurred as the density, and hence
the gap, varies across the sample [80].

Chapter 3
Feshbach resonances
In the previous chapter we determined that varying the fermion-fermion interaction is the key to accessing BCS-BEC crossover physics. We also observed that to
arbitrarily vary the interactions we could imagine using a variable attractive potential with a bound state near threshold. Amazingly atomic systems can achieve
exactly such a physical situation. The attractive potential is provided by the van
der Waals interaction between two atoms, and the knob to tune this potential
is a homogeneous magnetic field. The sensitive magnetic-field dependence of the
potential can be provided through a scattering resonance known as a Feshbach
resonance [106, 107, 108, 109]. The goal of this chapter will be to understand
exactly how a Feshbach resonance allows us to arbitrarily tune the interaction
using a magnetic field. In this chapter I will be discussing only two-body physics,
namely the problem of two fermions scattering in vacuum. We will find that the
Feshbach resonance used for the experiments in this thesis approximates well the
classic two-body system required for study of the BCS-BEC crossover problem of
Ch. 2.

3.1

Description

Calculating the interaction between two ground state alkali atoms is a nontrivial
problem that has been studied extensively in atomic physics [110]. Study of this
problem shows that for S-state atoms the interatomic potential is repulsive for
very small r and has a weak attractive tail that goes as C6 /r6 as r .
20

21

CHAPTER 3. FESHBACH RESONANCES

This weak attractive tail is a result of the interaction between mutually induced
dipole moments of the atoms, which is known as the van der Waals interaction.
The interatomic potentials are deep enough to contain a large number of bound
vibrational states. A Feshbach resonance occurs when one of these bound states
(often called the bare molecule state) coincides with the collision energy of two
free atoms in a different scattering channel. Such a situation is depicted in Fig.
3.1(a). The interatomic potential of the two free atoms is often referred to as the
open channel, while the potential containing the bare molecule state is referred
to as the closed channel. When the closed and open channels describe atoms
in different magnetic sublevels, they can be shifted with respect to each other
through the Zeeman effect using an external magnetic field (Fig. 3.1(b)).
(a)

(b)
closed channel
open channel

Energy

Energy

C6

r6
internuclear separation r

magnetic field B

Figure 3.1: (a) Feshbach resonances are the result of coupling between a molecular
state in one interatomic potential with the threshold of another. (b) The bare
molecule state of the closed channel tunes differently with magnetic field than the
open-channel threshold. This can lead to a crossing of the two levels.
Typically the effect of the coupling between the closed and open channels is
small, but at a Feshbach resonance when the open-channel dissociation threshold
is nearly degenerate with the bare molecule state, the effect of the coupling can be
significantly enhanced. This coupling changes the effective interatomic potential,
which we will refer to as the multichannel potential [111]. A bound state will
be added to this multichannel potential at a magnetic field value near (but not
exactly at) the magnetic field position of the crossing of the bare molecule and
open-channel threshold. When I use the term molecule in this thesis I am
always referring to this additional bound state of the multichannel potential, a
so-called dressed molecule. The wavefunction of these molecules is generally a

22

CHAPTER 3. FESHBACH RESONANCES

linear superposition of open-channel and closed-channel wavefunctions. As we will

see in the next section, the open-channel component dominates for the molecules
studied in this thesis.
As the magnetic field is tuned this multichannel bound state moves through
threshold, and the scattering length between atoms in the open channel diverges.
The scattering length near a Feshbach resonance varies with the magnetic field,
B, according to the following equation [110].

a(B) = abg 1

w
B B0

(3.1)

Here abg is the triplet background (nonresonant) scattering length for atoms scattering in the open channel, B0 is the magnetic field position at which the molecular
bound state of the coupled system goes through threshold, and w is the magneticfield width of the Feshbach resonance, defined as the distance in magnetic field
between B0 and the magnetic field at which a = 0. Figure 3.2 shows how a diverges according to Eqn. 3.1 for the 40 K resonance described in the next section.

3000
2000

a (a0)

1000
0
-1000
-2000
-3000
190

195

200

205

210

215

B (gauss)
Figure 3.2: Behavior of the scattering length at a Feshbach resonance in
between the f =9/2 and mf =7/2, 9/2 states.

40

23

CHAPTER 3. FESHBACH RESONANCES

3.2

A specific example

To further illustrate Feshbach resonance physics it is useful to consider a specific

atomic example. Here I will discuss the case of the 40 K resonance that is used
for most of the BCS-BEC crossover studies in this thesis. The open channel in
this case describes the scattering of |f, mf i = |9/2, 9/2i and |9/2, 7/2i atoms,
which are the two lowest energy states in the 40 K system (see Fig. 4.1). f describes
the total atomic angular momentum and mf is the magnetic quantum number.
For this particular problem the open channel couples to only one closed channel,
|f, mf i = |9/2, 9/2i and |7/2, 7/2i atoms. We will compare the results of this
example to the requirements for studying the classic BCS-BEC crossover problem
of Ch. 2. In the classic problem instead of two coupled channels a single variable
depth potential with r0 a is considered (where r0 is the range of the potential).
To calculate exactly the properties of our 40 K resonance we would need to
carry out a full coupled channels calculation using realistic potassium potentials.
The description of such a calculation is beyond the scope of this thesis but is
described nicely in, for example, Ref. [110].1 Instead, for the demonstrative
purposes of this chapter, we will examine the results of a simpler technique that
is derived from K. Goral et al. in Ref. [112] and applied to the case of 40 K by M.
Szyma
nska et al. in Ref. [111]. We will solve the coupled Schrodinger equations
that describe our two-channel system with a few simplifying assumptions. This
technique reproduces the important physics of our 40 K resonance using a small
set of experimentally measurable parameters. The main approximation is the
so-called pole approximation described in Ref. [112] that holds when the open
channel is strongly coupled to only one bare molecule state.
Our main goal will be to use the simplified two-channel calculation to determine the binding energy of the molecular state in the multichannel potential as a
function of the magnetic field. The Hamiltonian for our two-channel system is

h 2 + Voc
W

H= m

h2 2
W
m + Vcc

(3.2)

We have compared some of the data presented in this thesis to full coupled channels calculations carried out by Chris Ticknor in John Bohns group at JILA. See Ch. 5 and Ch.
6.

24

CHAPTER 3. FESHBACH RESONANCES

where Voc is the uncoupled open-channel potential, Vcc is the uncoupled closedchannel potential, the potential W describes the coupling between the open and
closed channels, and m is the mass of 40 K. It can be shown that the solution
to this problem depends on only a few accessible parameters of the 40 K system
[112]. For the 40 K resonance we are considering, these parameters include the
background scattering length abg = 174 a0 [54], the van der Waals coefficient
C6 = 3897 a.u. [113], the resonance width w = 7.8 G [57], and the resonance
position B0 = 202.1 G [73]. Also useful is the binding energy of the first bound
state in Voc , E1 = 8.75 MHz (which can be attained from abg and C6 ) [112,
111]. Finally, we need the difference in magnetic moment (change in energy with
magnetic field) of the open channel threshold with respect to the closed channel
bare molecule. In our simplified calculation we will assume this to be the linear
value that best approximates the result of a full coupled channels calculation,
co = 1.679 B = h 2.35 MHz/G [111, 114].2 These parameters, which in the
end come from experimentally measured values, of course all have uncertainties,
which we will ignore for now.

- Eb/h (MHz)

-10
mco

-20
-30
-40
0

100

200
B (gauss)

300

Figure 3.3: Multichannel bound state of a Feshbach resonance in 40 K determined

through the simplified calculation described in the text [112, 111].
2

For 40 K co from the result of the full calculation comes quite close to the difference in
magnetic moments of the closed and open channel thresholds, which is co = h 2.49 MHz/G at
200 G according to the Breit-Rabi formula [115].

25

CHAPTER 3. FESHBACH RESONANCES

The solution to the coupled Schrodinger equations based on the Hamiltonian

above using the pole approximation is outlined in Ref. [112]. The calculation is
not computationally intensive and, after inserting the parameters above, provides
the multichannel binding energy Eb , which I plot as the solid line in Fig. 3.3.
Here Eb is plotted with respect to the energy of the open channel threshold, i.e.,
the open channel threshold is zero for all values of B. The dotted line shows the
movement of the bare molecule energy with magnetic field, and the flat dashed
line is the value of E1 . The multichannel bound state is the dressed state of the
avoided crossing of these two levels. The bare molecule state crosses threshold
about 9 G higher than the position where the multichannel bound state comes
through threshold, B0 . The difference between these two crossings is proportional
to the resonance width parameter w and related to the interchannel coupling
parameter in the Hamiltonian, W [112]. Note that for 40 K the multichannel
bound state adiabatically connects to the the highest-lying vibrational state of
the open channel at low field, rather than the bare molecule of the closed channel.

- Eb/h (MHz)

-0.5
-1.0
calculation of Fig. 3.3
-h2/m(a-r0)2

-1.5

-h2/ma 2

-2.0
190

195
B (gauss)

200

Figure 3.4: Binding energy near the Feshbach resonance peak using three calculations of varying degrees of approximation. These results agree with those in Ref.
[111].

The physics we study in this thesis actually occurs over a very small region near
threshold in Fig. 3.3. The blue line in Fig. 3.4 shows a closeup of the multichannel
bound state near threshold. Within a few gauss range of the Feshbach resonance

26

CHAPTER 3. FESHBACH RESONANCES

h
position the result obeys the expectation for a >> r0 , Eb = ma
2 (where a is
determined through Eqn. 3.1), which is shown by the green line in Fig. 3.4.
Farther away there is a clear deviation from quadratic behavior in 1/a. This
behavior can be estimated by subtracting the range of the van der Waals potential

h2
r0 from a in the calculation of the binding energy, Eb = m(ar
2 (red line in Fig.
0)
3.4) [112, 111, 116]. The range of the van der Waals potential is given by [117]

1 (3/4) mC6
r0 =
2
8 (5/4) h

The value of r0 is 60 a0 for

40

1/4

(3.3)

K (and 30 a0 for 6 Li).

0
-20
0.04

closed channel
fraction

- Eb/h (kHz)

EF

-40
-60

201.0

0.02

0.00
201.0

201.5
B (gauss)

201.5

202.0

202.0

Figure 3.5: Binding energy in the magnetic field regime used for the BCS-BEC
crossover studies of this thesis.
In Fig. 3.5 we examine the binding energy even closer to threshold. Here

h2
the result of the simplified two-channel calculation and Eb = m(ar
2 (blue) are
0)

h2
indistinguishable, while the Eb = ma2 prediction (green) only deviates slightly.
We also plot the closed channel contribution to the molecule wavefunction, which
b
is given by dE
/co . The Feshbach molecules that we will be interested in for
dB
the BCS-BEC crossover studies will have binding energies on order of or smaller
than EF . A typical value of EF in our experiments (h 15 kHz) is shown by the
dotted line in Fig. 3.5. At this point the closed channel fraction according to our
two-channel calculation is only 2%; this result is in agreement with full coupled

CHAPTER 3. FESHBACH RESONANCES

27

channels calculations [118, 114]. This open-channel dominance means that, for
our crossover studies, the two coupled channels problem of the Feshbach resonance
can be approximated by small effective changes to the open channel potential. A
resonance for which this is true is often referred to as a broad resonance, while
the limit in which the closed channel dominates for Eb EF is referred to as a

h2
narrow resonance [119, 120, 121]. In the case in which Eb m(ar
2 we can
0)
derive a simple equation that describes this criterion for a broad resonance in a
Fermi system,
w2 2co
1.
(3.4)

h2
4EF ma
2
bg

For the 40 K resonance we have been considering the left side of this expression
has a numerical value of 43, and it has a value of 10,000 for the 6 Li resonance at
B0 = 834 G [116, 122], indicating both Feshbach resonances used for BCS-BEC
crossover studies thus far can be considered broad.
In conclusion, we have seen how in atomic systems a Feshbach resonance can
be used to add an additional bound state to an interatomic potential, leading
to a divergence in the zero-energy scattering cross section for atoms colliding
through the open channel. The Feshbach resonances that have so far been used
for BCS-BEC crossover studies are broad resonances. These resonances can be
approximated by a single-channel problem and display universal properties, i.e.,
they have no dependence on the details of the atomic structure, but rather only on
the parameters a and kF . For the 40 K resonance there are slight deviations from
the a r0 limit that must be taken into account for precise measurements (Fig.
3.5). Still the physics of a 40 K gas at this resonance should basically reproduce
the classic BCS-BEC crossover scenario envisioned by Leggett [47] and described
in Sec. 2.2.

Chapter 4
Cooling a Fermi gas and
measuring its temperature
To access the superfluid state of a Fermi gas in the crossover the gas must be cooled
below the critical temperature, Tc . To achieve such a temperature with a trapped
gas of fermions was one of our largest challenges. In this chapter I describe how we
cool 40 K and assess the success of this cooling through temperature measurements.
More technical details related to the contents of this chapter can be found in Ch.
10.

4.1

Cooling

40

The apparatus used to cool 40 K for these BCS-BEC crossover experiments employs
the strategy used for some of the first experiments with 87 Rb BEC [22, 123, 124].
We perform the usual combination of trapping and cooling in a magneto-optical
trap (MOT) followed by evaporative cooling [19, 20, 21]. The laser cooling uses
light from semiconductor diode lasers on the 40 K D2 line (4S1/2 to 4P3/2 transition
at 766.7 nm), and a two chamber apparatus allows for an ultra high vacuum region
for evaporative cooling [83]. The major difference compared to the 87 Rb experiments stems from the fact that elastic collisions between identical fermions are
suppressed at ultracold temperatures. This is because quantum statistics require
antisymmetry of the total wavefunction for two colliding fermions, which forbids

28

29

CHAPTER 4. COOLING A FERMI GAS

40

f=7/2
7/2 = mf

1.3 GHz
+7/2
-9/2 -7/2

-5/2

+9/2=mf

f=9/2

Figure 4.1: 40 K ground state level diagram, with exaggerated Zeeman splittings.
The two levels represent the hyperfine structure, which originates from the coupling of the nuclear spin (I = 4) with the electron spin (S = 1/2). Note the
hyperfine structure of 40 K is inverted.

s-wave collisions for identical fermions. While odd partial wave collisions, such
as p-wave, are allowed, these collisions are suppressed below T 100 K because
of the angular momentum barrier [125]. Thus, since evaporative cooling requires
collisions to rethermalize the gas, a mixture of two distinguishable particles is
required to cool fermions.
40
K provides an elegant solution to this problem. Figure 4.1 shows the ground
state energy levels of 40 K. The large angular momentum of the lowest ground
state hyperfine level (f =9/2) provides 10 distinct spin states. The two highest
energy states, mf =+9/2 and mf =+7/2, can both be held with reasonable spatial
overlap in a magnetic trap, which is the type of trap most proven for evaporative
cooling when starting from a MOT. In this way an apparatus designed for only
one atomic species could provide two distinguishable states for cooling. To remove
the highest energy atoms for evaporative cooling, microwaves at 1.3 GHz were
used to transfer atoms to untrapped spin states in the upper hyperfine state
[32, 83]. With this technique, quantum degeneracy was reached in 1999 and by
2001 two-component 40 K Fermi gases at temperatures of 0.25 TF could be created.
A detailed description of the work to achieve these temperatures is recorded in
Brian DeMarcos Ph.D. thesis (Ref. [83]).
One of the first steps in accessing the BCS-BEC crossover was to create a

30

CHAPTER 4. COOLING A FERMI GAS

^
z

Figure 4.2: Optical trap at the focus of a gaussian laser beam.

degenerate Fermi gas in an equal mixture of the f =9/2 Zeeman states between
which a Feshbach resonance was predicted, mf =9/2 and 7/2 (Fig. 4.1). To
accomplish this we needed to trap the high-field-seeking spin states of the f =9/2
manifold; these states cannot be trapped in a magnetic trap. Thus, we developed
a far-off-resonance optical dipole trap (FORT) to confine the required spin states
for the crossover experiments. Such a trap can be formed at the focus of a gaussian
laser beam whose optical frequency is far detuned from the 40 K transitions (Fig.
4.2).
We found that the best way to realize a cold gas in an optical dipole trap was
to load a relatively hot gas of fermions after some evaporation in the magnetic
trap. Typically we load the optical trap when the sample in the magnetic trap
reaches T /TF 3. After transfer to the mf =9/2, 7/2 spin states the gas
is evaporated in the optical trap simply by lowering the depth of the trap and
allowing the hottest atoms to spill out. This evaporation and the subsequent
temperature measurements are typically performed at a magnetic field of 235
G. At this field the scattering length is the background (nonresonant) value of
174 a0 , and from this field Feshbach resonances can be easily accessed.1
Evaporative cooling in the optical trap actually allowed us to achieve colder
temperatures than previous records. We could cool to T /TF 0.1 (and possibly
colder) with up to 106 atoms in each spin state. At these temperatures another
impediment to cooling fermions appears - Pauli blocking of collisions [126]. In
a degenerate Fermi sea the Pauli exclusion principle forbids collisions for which
the final state would place fermions in already occupied levels. This results in a
1

The background scattering length is reasonably large compared to many other alkali atoms,
resulting in nice evaporative cooling. Yet the interaction it provides still allows us to approximate
the gas as an ideal Fermi gas for the purpose of temperature measurement.

CHAPTER 4. COOLING A FERMI GAS

31

suppression of the collision rate compared to the classical expectation and may
make evaporative cooling more difficult in an all-fermion system. Many recent
experiments have used a bosonic atom as their second particle as a possible
approach to avoiding this problem [33, 34, 35, 37, 38].
An important question is why colder temperatures could be achieved in the
optical trap than had been previously achieved with the same apparatus in the
magnetic trap. There are two main possibilities for the answer to this question.
First, evaporation in a magnetic trap relies on the ability to drive spin-changing
transitions with rf or microwaves. At cold temperatures the microwave frequencies
required for the mf =+9/2 and the mf =+7/2 states differ, requiring the use of
a more complicated two-frequency evaporation scheme [83]. The optical trap on
the other hand evaporates both states equally, maintaining the optimum equal
mixture automatically. Second, the history of the cooling attempts with the
apparatus have shown that, not surprisingly, a low heating rate is essential to
achieving a degenerate Fermi gas. We have found that the coldest achievable
temperatures have been proportional to the heating rate in the trap. The heating
rate in our optical trap is typically 5 nK/sec in weak traps (after solving the
technical issues discussed in Ch. 10). For a similar atom number in the magnetic
trap the heating rate is 20 nK/sec [83].

4.2

Measuring the temperature of a Fermi gas

In atomic gas experiments the standard technique for measuring the temperature
of a gas is to suddenly turn off the trapping potential confining the atoms and
then wait for an expansion time t. A two-dimensional image of the atomic
distribution then reveals the velocity, or momentum, distribution of the gas [20].
This technique is often referred to as time-of-flight imaging. For a classical gas
the velocity spread is directly related to the temperature of the gas through the
Maxwell-Boltzmann distribution. In the quantum regime measuring temperature
from such distributions becomes trickier. In this section I will describe techniques
for measuring the temperature of an ideal Fermi gas using images of gas distributions.

CHAPTER 4. COOLING A FERMI GAS

4.2.1

32

An optically trapped Fermi gas

Before presenting temperature measurements, it is useful to describe the expected

distribution of an optically trapped Fermi gas. First, we analyze the optical dipole
potential created from a single gaussian laser beam using a harmonic approximation; this is the potential used for most of the BCS-BEC crossover experiments
presented in this thesis. Then we present the most relevant formulas for describing
the spatial and momentum distributions of a Fermi gas in such a harmonic trap.
A more thorough description of a harmonically trapped Fermi gas can be found
in Ch. 5.2 of Ref. [83].
Optical dipole trap
An optical dipole trap results from the interaction between a light field and the
oscillating atomic electric dipole moment that is induced by the light field. This
effect, known as the ac Stark shift, forms the conservative part of the interaction of atoms with light. The shift is proportional to the light intensity, and the
sign depends upon the sign of the detuning of the light compared to the atomic
resonance. When the light is detuned red of the atomic transition and the light
intensity varies in space, an attractive potential well can be formed. While the
photon scattering rate decreases quadratically with the frequency detuning between the light and the atomic transition, the ac Stark shift varies linearly with
this detuning. Thus, a trap with a large detuning results in a significant dipole
potential at a low photon scattering rate. This allows use of the dipole potential
for atom traps with long storage times, as was first realized in Ref. [127].
The simplest optical dipole trap is formed by the focus of a single red-detuned
gaussian laser beam (Fig. 4.2). The trap potential from the beam is proportional
to the laser intensity, which is given by
I(r, z) =

2r 2 /w2
Ipk
1+(z/zr )2
e
1 + (z/zR )2

(4.1)

where w is the beam waist (1/e2 radius), zR = w2 / is the Rayleigh range,

2p
Ipk = w
2 is the peak intensity, and p is the total optical power.
Calculating the strength of the potential from atomic properties is straightfor-

33

CHAPTER 4. COOLING A FERMI GAS

ward [128]. Let be the laser wavelength and 0 be the wavelength of the atomic
transition; correspondingly = 2c/ and 0 = 2c/0 . The dipole potential is
then given by
3c2
1
1
U (r, z) =
+
I(r, z)
3
20
0 0 +

(4.2)

where is the linewidth of the atomic transition.

For our potassium trap = 1064 nm, = 2 6.09 MHz, and typically w=15
to 20 m. Equation 4.2 is in principle only valid for a two-level system, while
potassium is of course multi-level, with the most relevant structure being the
fine structure giving the D1 and D2 lines at 770.1 nm and 766.7 nm, respectively.
However, in this case where the fine structure splitting is small compared to 0
we can approximate the system as two-level with 0 given by the center of the
two lines, 768.4 nm [128]. Note though that we do not make the rotating wave
approximation (RWA) in the Eqn. 4.2; for our detuning the 01+ term actually
contributes a 14% effect.
Figure 4.3 shows cross-sections of the potential for a typical trap that would
hold a degenerate Fermi gas in our system. In Eqn. 4.2 we have ignored the effect
of gravity on the trap. Yet when the trap becomes shallow, gravity can play a
large role. The y-direction in Fig. 4.3 shows the gravitational potential added
to Eqn. 4.2. Gravity effectively lowers the trap depth in the y-direction, making
evaporation in shallow traps one-dimensional.
At the bottom of the trap we can approximate the gaussian potential as a
parabolic potential (dotted line in Fig. 4.3). In this case we have a harmonic trap
with oscillator frequencies given by
r = 2r =

4U0
mw2

z = 2z =

4U0
.
mzR2

(4.3)

where U0 = |U (0, 0)|. In shallow traps the harmonic approximation breaks down
since atoms sample the potential near the top of the well where the gaussian
rises less steeply than the parabola. We account for this effect in experiments
by modelling the real trap as a harmonic trap with a lower effective frequency.
We determine the effective frequency experimentally through observation of the

34

CHAPTER 4. COOLING A FERMI GAS

2
1
U/kb (mK)

0
-1
-2
-3
-4
-30 -15 0 15 30
x (mm)

-30 -15 0 15 30 -3.0 -1.5 0 1.5 3.0

z (mm)
y (mm)

Figure 4.3: Potential for potassium atoms created by a single 1064 nm gaussian
laser beam with w=15 m and a power of 10 mW. The dotted line in the leftmost graph shows the parabola that approximates the potential for small x. The
y-direction includes the effect of gravity. Note that the horizontal scale in the
third graph (z-direction) is a factor of 100 larger than the scale in the first two
graphs.

Fermi gas excitation frequencies in the weakly interacting regime.

Trapped, ideal Fermi gas distribution
We now consider a Fermi gas in a harmonic potential. We start with the Hamiltonian for a particle in a harmonic well
H=

1
p2
+ mr2 2
2m 2

(4.4)

where p2 = p2x + p2y + p2z and 2 = x2 + y 2 + 2 z 2 , ( = z /r ). The density of

2
= (r2 z )1/3 [101]. The statistics of the Fermi gas
states is g() = 2(h )3 where
are described by the Fermi-Dirac distribution function
f () =

1
e

kb T

/ + 1

(4.5)

where = e/kb T is the fugacity.

We can calculate the Fermi energy, which is defined as the energy of the highest
occupied level of the potential at T = 0. We simply equate the integral over all

35

CHAPTER 4. COOLING A FERMI GAS

states up to EF to the number of particles in one fermion spin state, N :

N=
The result is
TF =

EF

g()d.

(4.6)

EF
=
(6N )1/3 .
kb
kb

(4.7)

The temperature compared to TF describes the degeneracy of the Fermi gas and
in the classical limit is related to peak phase space density through P SDpk =
(T /TF )3 /6.

E / EF

3
2
1
0
0

0.2

0.4 0.6
T / TF

0.8

1.0

Figure 4.4: Energy of an ideal Fermi gas.

Also using the distribution function and the density of states we can obtain
thermodynamic quantities such as the energy per particle, which will become
important in Ch. 9.

Li4 ()
g()f ()d
U
= 3kb T
= R0
E=
N
Li3 ()
0 g()f ()d

The function Lin (x) =

k=1

(4.8)

xk /k n appears often in the analysis of a harmonically

trapped Fermi gas. Lin is the Poly-Logarithmic function of order n, sometimes

written gn . Figure 4.4 plots the result of Eqn. 4.8. In the classical regime the
energy is proportional to the temperature, while in the Fermi gas limit the energy
asymptotes to 43 EF ( 38 EF kinetic energy and 83 EF potential energy).

36

CHAPTER 4. COOLING A FERMI GAS

Table 4.1: Distribution functions for a harmonically trapped Fermi gas.

Validity

Spatial distribution where r2 =

all T /TF

n() =

kb T
mr2

and rF2 =

2EF
mr2

2
2
N
Li3/2 (e /2r )/Li3 ()
(2)3/2 r3

T /TF 1 nc () =

2
2
N
e /2r
(2)3/2 r3

T =0

n0 () =

N 8
3 2
rF

Validity

Momentum distribution where p2 = mkb T and p2F = 2mEF

all T /TF

(p) =

2
2
rF

3/2

for < rF , 0 otherwise

2
2
N
Li3/2 (ep /2p )/Li3 ()
(2)3/2 p3

T /TF 1 c (p) =

2
2
N
ep /2p
(2)3/2 p3

T =0

N 8
p3F 2

0 (p) =

p2
p2F

3/2

for p < pF , 0 otherwise

Distribution functions in position and momentum can be determined through

standard statistical mechanics techniques and the Thomas-Fermi approximation,
which holds when many oscillator states are occupied (see Refs. [129, 83]). Table
4.1 shows the resulting Fermi-Dirac distribution functions in position and momentum space. The full expression is shown first, but it is also useful to understand
the classical and T = 0 limits. In the classical limit the distribution is gaussian,
and at T = 0 the distributions are only nonzero for values less than the Fermi radius rF or Fermi momentum pF = h
kF . It is useful to explicitly state the meaning
of the Fermi wavevector, kF , as it is a quantity we normalize to in the BCS-BEC

h2 k2
crossover. kF is defined through EF = 2mF and is related to the inverse distance
between fermions at the center of the trap; at T = 0 the number density of one
k3
fermion spin state at the center of the trap is written in terms of kF as npk = 6F2 .

37

CHAPTER 4. COOLING A FERMI GAS

4.2.2

Measuring temperature from the momentum distribution

In the experiment we access the distribution of the Fermi gas through absorption
images of an expanded gas. Absorption images are aquired by illuminating the
atoms with a resonant laser beam and imaging the shadow cast by the atoms onto
a CCD camera. These images effectively integrate through one dimension to give
a two-dimensional image (for example Fig. 4.6(a)). The appropriate function for
this distribution is n() or (p) (see Table 4.1) integrated over one dimension
[83]. Integrating through x and writing the result in terms of the experimentally
observed optical depth (OD), we find for the Fermi-Dirac distribution

ODF D (y, z) = ODpk Li2 (e

y2
2
2y

z2
2
2z

)/Li2 ().

(4.9)

In the classical limit this equation becomes a two-dimensional gaussian function:

ODgauss (y, z) = ODpk e

y2
2
2y

z2
2
2z

(4.10)

These forms are applicable for both the spatial and momentum profiles, and for
kb T
2
2
arbitrary expansion times through the relations y2 = m
2 [1 + (r t) ] and z =
r
kb T
[1 + (z t)2 ], where t is the expansion time.
mz2
Examples of theoretical cross sections of ODF D for various degeneracies, and
constant TF , are shown in Fig. 4.5. As the gas becomes colder the distribution
becomes narrower and the shape of the distribution becomes flatter. Note that as
the temperature is lowered far below TF the changes in the distribution become
small compared to TF . Still, down to T /TF 0.1 the temperature can be determined from least-squared fits to such distributions. Figure 4.6(a) is a sample
absorption image of an expanded Fermi gas. The black points in Fig. 4.6(b) are
the result of an azimuthal average of the image. The red line shows the result of
a surface fit of the two-dimensional image to the Fermi-Dirac distribution (Eqn.
4.9), which reveals that the gas is at a temperature of 0.1 TF . For comparison the
blue line is the result of a fit appropriate for a classical distribution (Eqn. 4.10).
Clearly the experimental distribution is flatter than a classical distribution.

38

CHAPTER 4. COOLING A FERMI GAS

OD (arb)

1.0

T/TF
0
0.1
0.2
0.5
1.0

0.5

0
0

0.5

1.0

1.5

k / kF
Figure 4.5: Theoretical cross sections of a harmonically trapped Fermi gas distribution integrated along one dimension. TF is held fixed as T /TF varies.

Optical depth

(a)

1.4
1.2
1.0
0.8
0.6
0.4
0.2
0
0

data
Fermi gas fit
Gauss fit

T/TF = 0.1

(b)
10

20

30

40

50

radius (arb)
Figure 4.6: Nonclassical momentum distribution of Fermi gas. (a) Sample false
color absorption image of the momentum distribution of a degenerate Fermi gas.
Here the integration is along the z-direction. (b) Azimuthally averaged profile
of the absorption image. The error bars represent the standard deviation of the
mean of averaged points.

39

CHAPTER 4. COOLING A FERMI GAS

To evaluate this thermometry, we can examine the results of least-squared

surface fits for gases at a variety of expected temperatures. In the fits ODpk ,
y , z , and are independent fit parameters. y and z tell us the temperature;
can be viewed as a shape parameter that is directly related to T /TF through
Li3 () = (T /TF )3 /6. As a check on the fits we compare the result for to
T /TF calculated through
y2 mr2
T
=
.
TF
h

(6N )1/3 (1 + (r t)2 )

(4.11)

We use the measured trap frequencies for r and z and the number of atoms in
each spin state N calculated from the total absorption in the image. For most
purposes we can extract N from a gaussian fit (Eqn. 4.10), in which case the total
2ODpk r z
number of atoms in the image is
, where is the photon absorption

320
cross-section, 2 . This result is exact for a classical gas and is only 7% off at
T /TF = 0.1.

105

1.0

(a)
z

T/TF from z

103

0.8

101
10-1
0 0.2 0.4 0.6 0.8 1.0

T/TF

(b)

0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1.0

T/TF

Figure 4.7: Analysis of fits to Eqn. 4.9 for expansion images of an optically
trapped gas with an equal mixture of mf =9/2, 7/2 atoms [59]. For these data
the integration was in the x-direction, and T was extracted from y .

Figure 4.7(a) shows a comparison between T /TF from Eqn. 4.11 and . The
black line shows the expected relationship for an ideal Fermi gas. In Fig. 4.7(b)
is converted to T /TF for a more direct comparison. In general the two values

40

CHAPTER 4. COOLING A FERMI GAS

agree, indicating the fits are extracting the correct information. Note that the
noise in the points becomes large at temperatures > 0.5 TF . This is expected
because the changes in the shape of the distribution become small in this limit. A
similar effect occurs in the low temperature limit where the distribution changes
little as the T = 0 Fermi gas limit is approached. However, the success of this
thermometer in the 0.1 < T /TF < 0.5 range has made this method the workhorse
of temperature measurements in the Jin lab since Brian DeMarcos work.

4.2.3

Measuring temperature using an impurity spin state

A second technique for measuring temperature that we have explored is impurity

spin-state thermometry. Eric Cornell proposed this idea in the JILA hallway as
a method to check the Fermi-Dirac surface-fit technique outlined in the previous
section. A check is especially necessary for the coldest temperature gases at 0.1 TF
and below, due to the decrease in the sensitivity of the Fermi-Dirac fits at these
temperatures. We have not done extensive work using impurity thermometer.
However, as we will see here this thermometer works quite well and has great unexplored potential, in particular as a technique that could measure temperatures
less than 0.1 TF .

mf
non-degenerate
"impurity" state

two-component
degenerate Fermi gas

5/2
7/2
9/2

Figure 4.8: Measurement of T through an embedded impurity spin state. All

three components are overlapped in the optical trap.

The idea of the impurity spin-state technique is to embed a small number of

atoms in a third state within the usual two-component gas (Fig. 4.8). In the
limit where the number of atoms in the impurity spin, Nim , is small compared to

CHAPTER 4. COOLING A FERMI GAS

41

the particle number in the original states, the Fermi energy of the impurity state
will be low enough that the impurity gas will be nondegenerate. Provided all of
the spin states in the system are in thermal equilibrium, the temperature of the
m 2
system will be kb tim
2 , where im is simply determined from a fit of the impurity
gas momentum distribution to a gaussian distribution (Eqn. 4.10).
A difficulty with this method is that EF scales weakly with particle number.
Suppose we originally have a gas with a particle number of 105 at T /TF = 0.1.
For T /TF of the impurity to be 1, Nim would need to be 100, and detecting the
distribution of 100 atoms with good signal to noise is not trivial. However, the
fully classical limit does not need to be reached to gain information about the
temperature from the impurity state. It is only required that the T /TF be large
enough that the energy of the impurity gas changes significantly with temperature.
Figure 4.4 illustrates for the range of T /TF for which this is the case.
To see if an impurity spin-state thermometer was feasible we designed an
experiment to test this thermometer against the surface-fit technique described
previously. We started with a (not necessarily equal) mixture of atoms in the
mf =+7/2, +9/2 spin states. Part way through the evaporative cooling process
a small fraction of the mf =+7/2 atoms were transferred to the mf =+5/2 state,
which served as our impurity (Fig. 4.8). For this experiment the gas was prepared
at a low magnetic field of a few gauss where the three-state mixture is fully stable.
Here the scattering length between any pair of the three spin states is 174 a0 . The
spin states were selectively imaged by applying a large magnetic field gradient
of 80 G/cm during the expansion to spatially separate the spin states (SternGerlach imaging) [83].
For analysis of the impurity spin-state data in the most general case where
the impurity is not fully classical, it is useful to introduce the variable Tgauss . For
m 2
, where gauss is the
a momentum space distribution Tgauss is defined as kgauss
2
bt
result from a least-squared fit to the gaussian distribution of Eqn. 4.10. As we
noted in Fig. 4.6 the Fermi distributions is not well fit by a gaussian, but the
result is a well-defined quantity. Figure 4.9 displays Tgauss /TF as a function of
degeneracy, as determined through least-squared fits to theoretical distributions.
Tgauss provides much the same information as the Fermi gas energy E (Fig. 4.4).
However, it is more useful for our current purposes because it can be extracted

42

CHAPTER 4. COOLING A FERMI GAS

1.0

Tgauss/TF

0.8
0.6
0.4
0.2
0
0

0.2

0.4

0.6

0.8

1.0

T/TF
Figure 4.9: Dimensionless plot of the variable Tgauss (defined in the text) versus
temperature.

from real images with better signal to noise than E, and it is convenient because
in the classical limit it becomes precisely the real temperature of the system.
In Fig. 4.10 is Tgauss as a function of N for a cold gas of atoms distributed
among the three states mf =9/2, 7/2, and 5/2. N is the measured number of
atoms in the spin state from which Tgauss is extracted. The seven sets of points
come from multiple iterations of the experiment for which the temperature was
expected to be constant. Since the trap strength was held constant N uniquely
defines EF , and given that the temperature T is constant for all points on the plot,
we can translate the dimensionless theoretical result of Fig. 4.9 to the situation
of Fig. 4.10. The black line in Fig. 4.10 shows the best-fit curve to the data, in
which the only free parameter is the real temperature T . Note T is the value of
Tgauss as N goes to zero. If the gas were fully classical, Tgauss would be constant as
a function of N ; it is a manifestation of Pauli blocking that different components
of an equilibrium gas can have the same temperature but different energies [126].
Figure 4.11 shows the results of four experiments like the one shown in Fig.
4.10. The temperature result from the impurity measurement is compared to
the result from the surface fits described in Sec. 9.1 (applied to the mf =9/2
cloud). We see that both methods agree to within the uncertainty for clouds in
the T /TF = 0.1 0.2 range.

43

CHAPTER 4. COOLING A FERMI GAS

0.20
T = 674 nK

Tgauss (mK)

0.15
0.10
mf=9/2
mf=7/2
mf=5/2

0.05
0

104

105
N

Figure 4.10: Impurity spin state thermometry. Tgauss is plotted for all of the
states in the gas, for multiple experiment iterations in which T was held fixed.
The mf =9/2 points have an average degeneracy of T /TF = 0.13.

4.2.4

Temperature in the BCS-BEC crossover

Our focus thus far has been on measuring the temperature of an ideal, normal
Fermi gas. For the BCS-BEC crossover experiments we would like to know the
temperature of our gas at any interaction strength. To this point there have been
no experiments that have directly measured the temperature of a gas throughout
the entire crossover. The main technique that has been used thus far is to instead
measure the entropy of the system, which, through theory, can be translated to
a temperature at any point in the crossover. The entropy, S, can be determined
through temperature measurements in the weakly interacting regime. For example, in the Fermi gas limit the entropy is given by S = ( 43 E )/kb T [130]. Then if
experiments are performed using adiabatic magnetic-field ramps to the crossover
regime, the entropy will be held constant at the weakly-interacting value. As
we have seen in 40 K it is convenient to measure the temperature of the weakly
interacting gas above the Feshbach resonance with a Fermi gas, while in most of
the 6 Li experiments the temperature is measured through the condensate fraction
close to the BEC limit. In either case, the limitation of this technique is that it

CHAPTER 4. COOLING A FERMI GAS

44

T/TF comparison: mf=9/2

T/TF from z

0.3
0.2
0.1
0
0

0.1
0.2
0.3
T/TF from impurity technique

Figure 4.11: Comparison of thermometers. The y-axis shows the result of the
Fermi-Dirac fits to the mf =9/2 distribution, and the x-axis shows the result of
the impurity spin-state technique.

relies upon theory to convert between entropy and temperature in the crossover.
Some initial work on this theoretical problem can be found in Ref. [131].
An alternative temperature measurement in the crossover was applied in the
group of J. Thomas where they measured the temperature at unitarity by using
fits to the momentum distribution in the hydrodynamic limit [79]; here the distribution is simply a rescaled Fermi distribution and thus amenable to the same
fitting procedure as the weakly interacting regime [132]. The impurity spin-state
thermometer described in Sec. 4.2.3 is another possible, but as of yet unexplored,
direct thermometer in the crossover regime. Since the mf =5/2 state is weakly
interacting with mf =9/2, 7/2 states even at the mf =9/2, 7/2 Feshbach
resonance, the mf =5/2 distribution could be used to measure T on resonance.
However, at high magnetic field values the mf = 5/2, 9/2 mixture is only
metastable; experimentally we have found this mixture has an exponential decay
time of 1 sec at 200 G for a typical density of 1013 cm3 . This short lifetime
makes a possible measurement more difficult, but not impossible.

Chapter 5
Elastic scattering near Feshbach
resonances between fermionic
atoms
Some of the first signatures of the presence of fermionic Feshbach resonances
were the observation of magnetic-field dependent changes in the elastic-scattering
properties of a normal Fermi gas [54, 58, 55, 57, 59, 60, 61, 62]. Here I present
three different techniques that we used to probe changes in scattering properties
at a fermionic Feshbach resonance. Two techniques probed the collision cross
section, which reveals the magnitude of a. In the third measurement we observed
the sign of a change and saw evidence for unitarity-limited interactions.

5.1

Measuring the elastic collision cross section

The first Feshbach resonance we searched for experimentally is the resonance

described in Ch. 3 that occurs between the mf =9/2 and mf =7/2 spin states.
The original theoretical prediction for the location of this resonance was B0 =
196+9
21 G, based on available potassium potentials [53]. We first experimentally
measured the position of this resonance using the technique of cross-dimensional
rethermalization, which measures the collision cross section [133]. This was a
technique that had provided much information about a Feshbach resonance in

45

CHAPTER 5. ELASTIC SCATTERING AT FESHBACH RESONANCES

46

bosonic 85 Rb gas [30, 134].

For this measurement we started with a gas of fermions in the mf =7/2, 9/2
spin states at T 2 TF . The gas was taken out of thermal equilibrium by
modulating the optical trap intensity at 2 y , which caused selective heating in
the y-direction. (We could selectively modulate one radial direction because for
this measurement our optical trap was not cylindrically symmetric, x = 1.7
y .) The exponential time constant for energy transfer between the two radial
directions, , was measured as a function of magnetic field.q is related to the swave collision cross section through 1/ = 2hniv/. v = 4 kb T /m is the mean
R
relative speed between colliding fermions and hni = N1tot n7 (r) n9 (r) d3 r is the
density weighted density. is the calculated number of binary s-wave collisions
required for rethermalization [125].

10-9

s (cm2)

10-10
10-11
10-12
10-13
160

180

200 220
B (gauss)

240

260

Figure 5.1: Collision cross section measured near an s-wave Feshbach resonance
between 40 K atoms in the mf = 7/2, 9/2 spin states at T = 4.4 K [57]. In
between the peak and dip in the interaction is attractive; everywhere else it is
repulsive.

Figure 5.1 plots the result of this measurement as a function of the magnetic
field, B. The magnetic field was calibrated through radio-frequency (rf) transitions between mf levels in the 40 K system. An advantage of the cross-dimensional

CHAPTER 5. ELASTIC SCATTERING AT FESHBACH RESONANCES

47

rethermalization technique is that it allows measurements of over a large range.

Through cross section measurements that extend over four orders of magnitude,
both the position of the divergence of the scattering length, B0 , and the position
of the zero crossing could be measured (Fig. 5.1). This told us the magneticfield width of the resonance w, which as we saw in Ch. 3 describes the coupling
strength.
The line in Fig. 5.1 is the result of a full coupled channels calculation of
carried out by C. Ticknor and J. Bohn, in which the parameters of the potassium potential were adjusted to achieve a best fit to our data from two different
40
K resonances [57, 135]. This calculation took into account the distribution of
collision energies in the gas by thermally averaging over a gaussian distribution
defined by a temperature of 4.4 K. The fit result placed the Feshbach resonance
at B0 = 201.6 0.6 G and the zero crossing at 209.9 0.6 G.

5.2

Anisotropic expansion

A disadvantage of the cross-dimensional rethermalization method is that it only

provides a valid measurement of in the so-called collisionless regime. A trapped
gas is considered collisionless if the trap oscillator period 1/ is much shorter than
the mean time between collisions in the gas, 1/. In the opposite limit where
the gas is collisionally hydrodynamic, obeying the classic hydrodynamic
equations. If a gas is hydrodynamic the cross-dimensional rethermalization time
will be determined by the oscillator period 1/ instead of the mean time between
collisions.
At the peak of the Feshbach resonance the fermion-fermion interactions can
easily become strong enough to make the gas collisionally hydrodynamic. In this
regime a technique more suited to measuring changes in the elastic cross section is
anisotropic expansion. In the hydrodynamic limit collisions during the expansion
transfer kinetic energy from the elongated axial (z) gas dimension into the radial
(r) direction. This changes the aspect ratio of the expanded gas (z /r ) compared
to the collisionless expectation (see Ch. 4). This effect was first observed in a 6 Li
Fermi gas in Ref. [58].
Figure 5.2 presents a measurement of anisotropic expansion in a 40 K gas at

CHAPTER 5. ELASTIC SCATTERING AT FESHBACH RESONANCES

48

T /TF = 0.34. Here we enhanced interactions between the mf =9/2, 5/2 spin
states using a Feshbach resonance between these states at 224 G. At the position
of the Feshbach resonance where is large, the aspect ratio z /y decreases. As
the gas becomes collisionless away from the resonance the aspect ratio smoothly
evolves to the collisionless value. The key to observing anisotropic expansion
is to hold the magnetic field at the value near the Feshbach resonance during
the beginning of expansion. In these experiments the magnetic field remained
high for 5 ms of expansion and a resonant absorption image was taken after a
total expansion time of 20 ms. In this measurement the mf =9/2, 5/2 gas was
created at the field B by starting with a mf =9/2, 7/2 gas and applying a
pulse between the -5/2 and -7/2 states 0.3 ms before expansion. This technique
avoids complications due to atom loss but creates a nonequilibrium gas.

1.2

s z / sy

1.0
0.8
0.6
0.4

220

224
B (gauss)

228

Figure 5.2: Anisotropic expansion of a strongly interacting Fermi gas [59].

In general the expected aspect ratio in the regime between collisionless and
hydrodynamic behavior is difficult to calculate. We can however check to see if
some degree of hydrodynamic expansion of the normal gas is expected. We can
calculate the elastic collision rate = 2hniv in the gas, using an elastic collision
cross section given by = 4a259 and |a59 | = 2000 a0 (as was measured near the
resonance peak [59]) to find = 46 kHz. Comparing this rate to the trapping

CHAPTER 5. ELASTIC SCATTERING AT FESHBACH RESONANCES

49

frequencies we find /r = 37 and /z = 2400. Hence, it is not surprising that

we observe anisotropic expansion. For a gas that was fully hydrodynamic, with
r , z , we would expect our measured aspect ratio to reach 0.4 [136, 137].
In ultracold gas experiments anisotropic expansion is often associated with
superfluidity of a BEC. This is because a typical ultracold Bose gas is collisionless, while below the superfluid transition the gas obeys superfluid hydrodynamic
equations. Near a Feshbach resonance however where the gas can be collisionally hydrodynamic, anisotropic expansion can be observed both above and below
the superfluid transition temperature. Still anisotropic expansion has been put
forth as a possible signature of superfluidity in the BCS-BEC crossover regime
[136, 137]. In pursuing this signature it is important to carefully distinguish between collisional and superfluid hydrodynamics and take into account that changes
during expansion will affect the many-body state [138]. Such analysis is possible
and has been considered in Refs. [58, 71].

5.3

Measuring the mean-field interaction

Measurements of the collision cross section are useful for detecting the strength
of the interaction but are not sensitive to whether the interactions are attractive
or repulsive. The mean-field energy, on the other hand, is a quantum mechanical,
many-body effect that is proportional to na. For Bose-Einstein condensates with
repulsive interactions the mean-field energy (and therefore a) can be determined
from the size of the trapped condensate [29, 139], while attractive interactions
cause condensates with large atom number to become mechanically unstable [140,
141]. For an atomic Fermi gas the mean-field interaction energy has a smaller
impact on the thermodynamics. Here I discuss a novel spectroscopic technique
that measures the mean-field energy of a two-component Fermi gas directly [59,
60].
In this measurement we again used the Feshbach resonance between the mf =5/2
and 9/2 spin states. At magnetic fields near the resonance peak, we measured the mean-field energy in the Fermi gas using rf spectroscopy (Fig. 5.3(a)).
First, optically trapped atoms were evaporatively cooled in a 72/28 mixture of
the mf =9/2 and mf =7/2 spin states. After the evaporation the optical trap

CHAPTER 5. ELASTIC SCATTERING AT FESHBACH RESONANCES

50

(a)
energy

mf=-5/2

hnrf
mf=-7/2
mf=-9/2

mf=-7/2 peak OD

was recompressed to achieve a larger density, and the magnetic field was ramped
to the desired value near the resonance. We then quickly turned on the resonant
interaction by transferring atoms from the mf =7/2 state to the mf =5/2 state
with a 73 s rf -pulse. The fraction of mf =7/2 atoms remaining after the pulse
was measured as a function of the rf frequency.

1.0
0.8
0.6
0.4

(b)

0.2
0

-40

-20

20

40

Dn (kHz)
Figure 5.3: Radio-frequency spectra. (a) Transition of interest. (b) Rf lineshapes
with (solid line) and without (dotted line) interactions [59].

The relative number of mf =7/2 atoms was obtained from a resonant absorption image of the gas taken after 1 ms of expansion from the optical trap.
Atoms in the mf =7/2 state were probed selectively by leaving the magnetic
field high and taking advantage of nonlinear Zeeman shifts. Sample rf absorption
spectra are shown in Fig. 5.3(b). At magnetic fields well away from the Feshbach
resonance we could transfer all of the mf =7/2 atoms to the mf =5/2 state
and the rf lineshape had a Fourier width defined by the rf pulse duration. At the
Feshbach resonance there were two changes to the rf spectra. First, the frequency
for maximum transfer was shifted relative to the expected value from a magnetic
field calibration. Second, the maximum transfer was reduced and the measured
lineshape is wider.
Both of these effects arise from the mean-field energy due to strong interactions
between mf =9/2, 5/2 atoms at the Feshbach resonance. The mean-field energy

CHAPTER 5. ELASTIC SCATTERING AT FESHBACH RESONANCES

51

produces a density-dependent frequency shift given by

=

2
h
n9 (a59 a79 ),
m

(5.1)

where n9 is the number density of atoms in the mf =9/2 state, and a59 (a79 ) is
the scattering length for collisions between atoms in the mf =9/2 and mf =5/2
(mf =7/2) states [142]. Here the nonresonant interaction term proportional to
the population difference between the mf =7/2 and mf =5/2 states is ignored;
this term equals 0 for a perfect -pulse. For our spatially inhomogeneous trapped
gas, the density dependence broadens the lineshape and lowers the maximum
transfer. This effect occurs on both sides of the Feshbach resonance peak. In
contrast, the frequency shift for maximum transfer reflects the scattering length
and changes sign across the resonance.
We measured the mean-field shift as a function of B near the Feshbach resonance peak. The rf frequency for maximum transfer was obtained from Lorentzian
fits to spectra like those shown in Fig. 5.3(b). The expected resonance frequency
was then subtracted to yield . The scattering length a59 was obtained using
Eqn. 5.1 with n9 = 0.5 npk and a79 = 174 a0 [54]. The peak density of the trapped
mf =9/2 gas npk was obtained from gaussian fits to absorption images. The numerical factor 0.5 multiplying npk was determined by modelling the transfer with
a pulse-width limited Lorentzian integrated over a gaussian density profile.
The measured scattering length as a function of B is shown in Fig. 5.4. This
plot, which combines data taken for two different gas densities, shows that we
were able to realize both large positive and large negative values of a59 near the
Feshbach resonance. The solid line in Fig. 5.4 shows a fit to the expected form for
a Feshbach resonance (Eqn. 3.1). Data within 0.5 G of the peak were excluded
from the fit. With abg = 174 a0 we found that the Feshbach resonance peak occurs
at 224.21 0.05 G and the resonance has a width w of 9.7 0.6 G.
When B is tuned very close to the Feshbach resonance peak, we expect the
measured a59 to have a maximum value on the order of 1/kF due to the unitarity
limit. This saturation can be seen in the data shown in Fig. 5.4. Two points that
were taken within 0.5 gauss of the Feshbach resonance peak, one on either side
of the resonance, clearly lie below the fit curve. The unitarity-limited point on the

CHAPTER 5. ELASTIC SCATTERING AT FESHBACH RESONANCES

52

attractive interaction side of the resonance (higher B) has an effective scattering

length of 2/kF . (Here h
kF is the Fermi momentum for the mf =9/2 gas.)

scattering length (a0)

3000
2000
1000
0
-1000
-2000
-3000

215

220
225
B (gauss)

230

Figure 5.4: Scattering length as measured through the mean-field interaction [59].
These data were taken for a normal Fermi gas at T /TF = 0.4 and at two different
densities: npk = 1.8 1014 cm3 (circles) and npk = 0.58 1014 cm3 (squares).

Table 5.1: Observed Feshbach resonances in

open channel |f, mf i
|9/2, 9/2i + |9/2, 7/2i
|9/2, 9/2i + |9/2, 5/2i
|9/2, 7/2i + |9/2, 5/2i
|9/2, 7/2i

5.4

40

40

l B0 (G)
w (G)
s 202.10 0.07 7.8 0.6
s 224.21 0.05 9.7 0.6
s 174
7
p 198.8

K.
reference
[57, 73]
[59]
unpublished
[57, 143]

K Feshbach resonance summary

Table 5.1 lists the Feshbach resonances we have studied experimentally in 40 K.

All of these resonances were originally located by measuring scattering properties
using the techniques described above. I include the states between which the

CHAPTER 5. ELASTIC SCATTERING AT FESHBACH RESONANCES

53

resonance occurs, the partial wave of the resonant collision l, our most precise
measurement of the resonance position B0 , and the resonance width w.

Chapter 6
Creating molecules from a Fermi
gas of atoms
After locating Feshbach resonances in our 40 K system, we wanted to observe evidence of a molecular bound state near threshold on the low-field side of the
Feshbach resonances. Creating molecules in this bound state, referred to as Feshbach molecules, would be the first step towards achieving the BEC limit of
the crossover problem. We were motivated to believe that it would be possible
to create Feshbach molecules by experiments carried out in the Wieman group
at JILA [144, 145]. By pulsing a magnetic field quickly towards a Feshbach resonance they were able to observe coherent oscillations between atoms and Feshbach
molecules in a 85 Rb BEC. We hoped to employ a slightly different approach to
creating molecules in which we would ramp the magnetic field fully across the
Feshbach resonance. In this chapter I will present how, using this technique, we
were able to efficiently and reversibly create Feshbach molecules from a Fermi gas
of atoms. Another of our contributions to the study of Feshbach molecules was a
spectroscopic detection technique that firmly established that Feshbach molecules
had been created. I also present our current understanding of the physics of
conversion of atoms to molecules using adiabatic magnetic-field ramps; this understanding was gained through a study of the conversion dependences led by the
Wieman group [146]. Lastly, I describe how the dissociation of molecules in a
sample at low density provided our most precise measurement of the Feshbach
resonance position.
54

55

CHAPTER 6. CREATING MOLECULES

B(t)
B0

atoms

Feshbach
molecule
Figure 6.1: Creating molecules using magnetic-field ramps across a Feshbach resonance.

6.1

Magnetic-field association

Figure 6.1 shows the behavior of the bound molecular state at a Feshbach resonance presented in Ch. 3. Given this picture, one would expect that atoms could
be converted to molecules simply by ramping the magnetic field in time across
the Feshbach resonance position B0 [147, 148, 149]. The only requirement to creating molecules in this way is that the magnetic-field ramp must be slow enough
to be adiabatic with respect to the two-body physics of the Feshbach resonance
(two-body adiabatic). To a very good approximation the Feshbach molecules have
twice the polarizability of the atoms [150] and therefore would be confined in the
optical dipole trap along with the atoms.1
We performed such an experiment using a magnetic-field ramp across the
mf = 5/2, 9/2 resonance introduced in the previous chapter. We started with
a nearly equal mixture of the two spin states mf = 5/2, 9/2 at a magnetic field
of 227.81 G. The field was ramped at a rate of (40 s/G)1 across the resonance
to various final values. The number of atoms remaining following the ramp was
determined from an absorption image of the gas at 4 G after expansion from the
optical trap. Since the light used for these images was resonant with the atomic
transition, but not with any molecular transitions, we selectively detected only the
atoms. Figure 6.2 shows the observed total atom number in the mf = 5/2, 9/2
states as a function of the final magnetic-field value of the ramp. We found that
1
In fact the atoms and molecules have the same trapping frequency, but the molecule trap
depth is twice as large as the atom trap depth.

56

CHAPTER 6. CREATING MOLECULES

B0

atom number (106)

1.5
1.0
0.5
0

215

220

225

230

B (gauss)
B(G)
Figure 6.2: Creation of molecules as seen through atom loss [63]. A fit to an error
function provides a guide to the eye. Atom loss occurs at precisely the expected
position of the Feshbach resonance given a previous measurement of the scattering
length divergence [59].

CHAPTER 6. CREATING MOLECULES

57

the atoms disappeared abruptly at the Feshbach resonance peak (dashed line).
We also found in similar experiments that we could recover the lost atoms with
an immediate magnetic-field ramp back above the Feshbach resonance. This ruled
out many atom loss processes and strongly suggested that all of the lost atoms
were converted to Feshbach molecules. We were surprised at the efficiency of the
conversion of our Fermi gas of atoms to a Bose gas of molecules; we could easily
create hundreds of thousands of Feshbach molecules.

6.2
6.2.1

Detecting weakly bound molecules

Rf spectroscopy

While suggestive of molecule creation, the measured atom loss was not conclusive
proof for the existence of Feshbach molecules. We thus employed a spectroscopic
technique to probe the molecules. First, we created the molecules with a magneticfield ramp across the Feshbach resonance and stopped at a magnetic field Bhold .
At Bhold a 13 s radio frequency (rf) pulse was applied to the gas; the rf frequency
was chosen so that the photon energy was near the energy splitting between the
mf =5/2 and mf =7/2 atom states (see Fig. 5.3(a)). The resulting population
in the mf =7/2 state, which was initially unoccupied, was then probed selectively
either by separating the spin states spatially using a strong magnetic-field gradient
during free expansion (Stern-Gerlach imaging) or by leaving the magnetic field
high (215 G) and taking advantage of nonlinear Zeeman shifts.
Figure 6.3(a) shows a sample rf spectrum at Bhold = 222.49 G; the resulting
number of atoms in the mf =7/2 state is plotted as a function of the frequency
of the rf pulse. We observed two distinct features in the spectrum. The sharp,
symmetric peak was very near the expected mf =5/2 to mf =7/2 transition
frequency for free atoms. With the Stern-Gerlach imaging we saw that the total
number of mf =5/2 and mf =7/2 atoms was constant, consistent with transfer
between these two atom states. The width of this peak was defined by the Fourier
width of the applied rf pulse. Nearby was a broader, asymmetric peak shifted
lower in frequency. Here we found that after the rf pulse the total number of
observed atoms (mf = 5/2 + 7/2) actually increased. Also, the resulting

58

CHAPTER 6. CREATING MOLECULES

transfer (arb)

1.0
0.8

(a)

0.6
0.4

kinetic energy (MHz)

0.2
0
0.4

(b)

0.3
0.2
0.1
0
49.8

50.1
50.4
50.7
rf frequency (MHz)

Figure 6.3: Rf spectrum for an atom/Feshbach molecule mixture [63]. (a) Transfer
to the mf =7/2 states as a function of rf frequency. The left feature is the
molecule dissociation spectrum and the right feature represents the transfer of
atoms between mf =5/2 and mf =7/2. (b) Corresponding kinetic energy of
the mf =7/2 state.

CHAPTER 6. CREATING MOLECULES

59

mf =7/2 gas in this region had a significantly increased kinetic energy, which
grew linearly for larger frequency shifts from the atom peak (Fig. 6.3(b)).
The asymmetric peak corresponds to the dissociation of molecules into free
mf =7/2 and mf =9/2 atoms. Since the applied rf pulse stimulates a transition
to a lower energy Zeeman state, we expected hrf = h0 Eb E, where Eb is
the binding energy of the molecule, 0 is the atom-atom transition frequency for
noninteracting atoms, and we have ignored mean-field interaction energy shifts.
The remaining energy, E, must be imparted to the dissociating atom pair as
kinetic energy. Two separate linear fits were applied to the kinetic energy data in
Fig. 6.3(b) to determine the threshold position. The slope beyond threshold for
the data is 0.49 0.03; this indicates that the atom pair (mf =7/2 + mf =9/2)
does indeed receive the additional energy, E, beyond the binding energy when
the molecule is dissociated.
The observed lineshape of the asymmetric peak in Fig. 6.3(a) should depend upon the Franck-Condon factor, which gives the overlap of the molecular
wavefunction with the atomic wavefunction. C. Ticknor and J. Bohn calculated
this multichannel Franck-Condon overlap as a function of energy. The resulting
transition rate, convolved with the frequency width of the applied rf and scaled
vertically, is shown as the solid line in Fig. 6.3(a). The agreement between theory
and experiment for the dissociation spectrum is quite good. This well-resolved
spectrum provides much information about the molecular wavefunction. A useful
discussion of the theoretical aspects of these dissociation spectra and their relation
to the wavefunction of the initial and final states can be found in Ref. [151].
In Fig. 6.4 is the magnetic-field dependence of the frequency shift = rf
0 , which to first order should correspond to the molecular binding energy. While
could in principle be obtained directly from the transfer spectrum (Fig. 6.3(a)),
we used the appearance of the threshold in the energy of the mf =7/2 gas, as it
is more clear (Fig. 6.3(b)). We compared the position of this energy threshold to
the expected atom-atom transition frequency 0 based upon a calibration of the
magnetic-field strength. The data are consistent with a theoretical calculation of
the binding energy (solid line) based upon a full coupled channels calculation with
no free parameters carried out by C. Ticknor and J. Bohn. This measurement of
Eb accentuates the fact that these Feshbach molecules are not typical molecules.

60

CHAPTER 6. CREATING MOLECULES

Dn (kHz)

-100
-200
-300
-400
-500
220

221

222

223

224

B (gauss)
Figure 6.4: The frequency shift () from the expected mf = 5/2 7/2
transition plotted versus magnetic field for the mf =7/2 atoms (squares) and the
molecules (circles). The line corresponds to a calculation of the binding energy of
the molecules as a function of detuning from the Feshbach resonance [63].

CHAPTER 6. CREATING MOLECULES

61

With binding energies on the order of h 100 kHz (41010 eV) they are extremely
weakly bound compared to molecules chemists are accustomed to studying.
The excellent agreement with theory in Fig. 6.4 left no doubt that efficient
creation of Feshbach molecules is possible. In addition, our rf spectroscopy technique was extended for a variety of other measurements in paired systems. It
was proposed that rf spectroscopy could be used to measure the excitation gap
in a superfluid Fermi gas [152, 153]; such a measurement is published in Ref.
[80]. Our rf spectroscopy technique was also extended to detect confinement induced molecules in a one-dimensional Fermi gas [154]. Molecule dissociation via
rf spectroscopy has proven useful for giving atoms in a molecule a large relative
momentum, as in Ref. [155]. Dissociating the molecules far above threshold produces a fun absorption image. The dissociated atoms fly out in a spherical shell,
and the resulting absorption image is a ring structure (Fig. 6.5).

(b)

(a)
+k
-k

Figure 6.5: Dissociation of molecules with radio frequencies [155]. (a) The atoms
that result from the dissociation have equal and opposite momenta. (b) False
color absorption image of a dissociated molecular gas.

6.2.2

Dissociation through magnetic-field modulation

A very similar technique to rf spectroscopy that we later employed to study

fermionic atom pairs such as molecules was magnetic-field modulation [81]. After
creating Feshbach molecules at a magnetic field Bhold , we modulated the homogeneous magnetic field at a frequency near Eb /h, as shown in Fig. 6.6(a). To

62

CHAPTER 6. CREATING MOLECULES

quantify the response of the gas, we determined the increase in the gas temperature after the perturbation was applied. We found that a sensitive way to detect
this increase in temperature was to measure the number of atoms that escaped
from a shallow optical dipole trap due to evaporation [81]. By varying the frequency of the modulation we could map out a molecule dissociation spectrum
and measure the dissociation threshold . The result of this measurement for a
variety of magnetic fields Bhold is shown in Fig. 6.6(b).

0
Dn (kHz)

(a)
B
Bhold

(b)

-20
-40

-h2/m(a-r0)2

-60
201.5 202.0 202.5
Bhold (gauss)
Figure 6.6: (a) Magnetic field modulation to dissociate molecules. (b) Resulting
binding energy.

Note that this measurement is considerably more precise than the rf spectroscopy measurement. One reason for this is that the B-modulation technique
does not change the magnetic moment of the atoms, meaning the measurement
is not sensitive to magnetic-field fluctuations. The lines in Fig. 6.6(b) show the
result of the calculation of Ch. 3, with the dashed lines indicating the uncertainty
on the curve due to the uncertainty in the measured Feshbach resonance parameters B0 and w.2 The data agree with the calculation to within the uncertainty,
but note in this measurement many-body effects may result in systematic shifts.
This magnetic-field modulation method of dissociating molecules was extended to
associate free 85 Rb atoms into Feshbach molecules in Ref. [156].
2

Most of this uncertainty originates in B0 , and the value and uncertainty in B0 comes from
the measurement in Sec. 6.4. Fig. 6.6(b) suggest that the resonance position is actually at the
upper limit of the measurement of Sec. 6.4.

CHAPTER 6. CREATING MOLECULES

6.3

63

Understanding molecule conversion efficiency

While using magnetic-field ramps to create molecules was very successful, there
were many outstanding questions about the physics of the process. For example,
what parameters define the conversion efficiency from atoms to molecules? The
first important parameter turns out to be the rate of the magnetic-field ramp
across the resonance. If the ramp is too fast no molecules will be created because
the ramp will be diabatic with respect to the atom-molecule coupling. As the
ramp is made slower, however, atoms will pair to form molecules. This effect is
shown in Fig. 6.7 where molecule creation through atom loss is shown. Theoretical
predictions find that this effect can be well modelled by the Landau-Zener formula
for the transition probability at a two-level crossing
f = fm (1 eLZ )

(6.1)

where f is the fraction of atoms converted to molecules, fm is the maximum

fraction of atoms that can be converted to molecules, and LZ is the LandauZener parameter [149, 112]. Let LZ be (dB/dt)1 where (dB/dt)1 is the inverse
magnetic-field ramp rate across the resonance. We can fit Fig. 6.7 with as the
fitting parameter to find in this case 20 s/G [63, 146]. Reference [112]
predicts that = n w abg where n is the atomic gas density, w is the width
of the Feshbach resonance, abg is the background scattering length, and is a
proportionality constant. A study by Hodby et al. verified the linear dependence
of upon the density n, but the proportionality constant for this expression is
still under investigation [146].
Notice in Fig. 6.7 that even at rates a few times slower than not all the atoms
are converted to molecules. One would expect that if the atom-molecule system
were in chemical equilibrium and the temperature of the molecular sample were
much less than Eb /kb then 100% of the atoms should be converted to molecules.
An important point to recognize for all of the experiments in this chapter is that
we are not operating in chemical equilibrium. At the final magnetic field values
in these experiments the time scale for chemical equilibrium is significantly longer
than time scales in Fig. 6.7, and we routinely work on time scales intermediate
between the time scale of and the chemical equilibrium time scale. Thus, the

64

CHAPTER 6. CREATING MOLECULES

molecule fraction

observed saturation in molecule conversion is important to understand. Reference

[146] studies this phenomenon both for a bosonic gas of 85 Rb and our fermionic
gas.

1.0
0.8
0.6
0.4
0.2
0
0

20

40

60

80

(dB/dt)-1 (ms/G)
Figure 6.7: Time scale for two-body adiabaticity [63].

Eric Cornell suggested that the saturation in molecule conversion was likely
related to phase space density (or T /TF in our case) based on intuitive arguments.
An adiabatic process smoothly alters the wavefunction describing atom pairs but
does not change the occupation of states in phase space. Thus to form a molecule
a pair of atoms must initially be sufficiently close in phase space that their combined wavefunction can evolve smoothly into the Feshbach molecule state as the
resonance is crossed. In other words one would expect a molecule to form if a pair
of atoms has a relative position rrel and relative velocity vrel such that
|rrel mvrel | < h

(6.2)

where is an experimentally determined constant.

Figure 6.8 shows the result of a measurement of the saturation in molecule
conversion (fm ) for a 40 K gas as a function of T /TF , which is monotonically
related to the phase space density of the gas. We see that indeed the conversion
fraction does increase as T /TF decreases (phase space density increases) with a
maximum conversion at our lowest temperatures of about 90%. 85 Rb data show

65

CHAPTER 6. CREATING MOLECULES

molecule fraction

1.0
0.8
0.6
0.4
0.2
0
0

0.2

0.4

0.6 0.8
T/TF

1.0

1.2

Figure 6.8: Dependence of molecule conversion on the initial T /TF of a twocomponent Fermi gas [146].

a similar dependence. To quantify the conversion expected for the many-body

problem given the two-body criterion above (Eqn. 6.2) an algorithm described in
Ref. [146] was developed. The line in Fig. 6.8 is the result of this algorithm for
the best fit value of . We found that for the fermion data = 0.38 0.04 and
for the boson data = 0.44 0.03, indicating that a similar process is at work in
both the Fermi and Bose cases.

6.4

A precise measurement of B0

A precise determination of the magnetic-field location of the Feshbach resonance

B0 is an essential ingredient for exploring the BCS-BEC crossover regime. Knowledge of the position and the width of the resonance allows a precise calculation
of the interaction strength at a particular magnetic field (see Ch. 3). As we saw
in Ch. 5, B0 can be measured via scattering properties of the resonance, but
our most precise measurement of B0 actually comes from the study of Feshbach
molecules. In particular we looked for dissociation of Feshbach molecules in a low
density gas as a function of magnetic field. To determine if the molecules had
been dissociated or not we probed the gas at low magnetic field; here atoms not

66

CHAPTER 6. CREATING MOLECULES

3
2

203
B (gauss)

Atom number (105)

bound in molecules can be selectively detected.

Figure 6.9 shows the result of such a measurement. Molecules created by
a slow magnetic-field ramp across the resonance were dissociated by raising the
magnetic field to a value Bprobe near the Feshbach resonance (inset to Fig. 6.9).
Note that to avoid many-body effects, we dissociated the molecules after allowing
the gas to expand from the trap to much lower density. This plan also allowed
us to be certain we would not create molecules in the ramp of the magnetic field
to near zero field for imaging. The measured number of atoms increased sharply
at B0 = 202.10 0.07 G. This measurement of the resonance position agreed well
with previous less precise results [57, 67].3

202
201
-5

1
0

5 10 15
t (ms)

201.8

202.0

202.2

202.4

Bprobe (gauss)
Figure 6.9: Determination of the position of the Feshbach resonance via molecule
dissociation in a low-density Fermi gas [73]. A fit of the data to an error function
reveals B0 = 202.10 0.07 G, where this uncertainty is given by the 10%-90%
width.

3
Note that this measurement of the resonance position is quite similar to the molecule association result of Fig. 6.2. However, the disadvantages of the association method are: (1) The
molecule creation must take place in a high-density sample; hence many-body effects may play
a role. (2) The details of the magnetic-field ramp to low field for imaging, in particular its initial
rate, are crucial.

Chapter 7
Inelastic collisions near a
fermionic Feshbach resonance
Elastic collisions between atoms (discussed in Ch. 5) are often referred to as good
collisions. These collisions allow rethermalization in the gas but do not change
the internal state of the atoms or molecules. In atomic gas experiments a constant
concern is inelastic collisions, often referred to as bad collisions. In these collisions
the products often are particles in lower energy internal states. The difference in
energy between the incoming particles and the products of the collision must be
carried away in the form of kinetic energy. When this energy difference is large
compared to energy scales such as the trap depth particles can be lost from the
trap and significant heating of the sample can occur.
Near a Feshbach resonance inelastic collisions can be enhanced along with
elastic collisions. To accomplish the work in this thesis we spent a large fraction
of our time understanding the inelastic processes near 40 K Feshbach resonances
and designing experiments that minimize the effect of inelastic collisions. In this
chapter I will discuss the inelastic collisions near a 40 K Feshbach resonance and
present measurements of relevant inelastic collision rates. We observed clear evidence of inelastic processes near the fermionic Feshbach resonance, but found
that despite these inelastic processes the lifetime of the sample was long enough
to study BCS-BEC crossover physics.

67

CHAPTER 7. INELASTIC COLLISIONS

7.1

68

Expected inelastic decay processes

Let us first consider the stability of free fermionic atoms on the BCS side of
a fermionic Feshbach resonance. In particular consider the Feshbach resonance
between the |f, mf i = |9/2, 9/2i and |9/2, 7/2i states that is used for many
of the experiments in 40 K. Since these are the two lowest energy states of 40 K
the only inelastic collision involving two of these fermions that is energetically
favorable is
|9/2, 9/2i + |9/2, 7/2i |9/2, 9/2i + |9/2, 9/2i.

(7.1)

This process however is forbidden due to the fermionic nature of the particles [53].
Thus, any inelastic collision with these states must involve at least three fermions.
A three-body inelastic collision in a two-component Fermi gas with components
X and Y would take the form
X + X + Y X + (XY ) .

(7.2)

Here the subscript represents a lower-energy molecular state. Such lower-energy

molecular states are always present in these atomic gas systems as there are many
vibrational levels of the interatomic potential. To conserve energy and momentum in the collision the products X and (XY ) carry away the binding energy
of the (XY ) molecule in the form of relative kinetic energy. Theory predicts
that this three-body collision process will be suppressed for s-wave interactions
between fermions because it requires that two identical fermions approach each
other [90, 91, 95, 157]. However, while the rate of this inelastic collision process is
suppressed, it is not forbidden, making it an important process near a Feshbach
resonance [57, 158].
As we cross the Feshbach resonance to the BEC side with a cold 40 K Fermi
gas we must consider the stability of a mixture of fermionic atoms and Feshbach
molecules. An isolated Feshbach molecule for the mf = 7/2, 9/2 40 K resonance
will be stable for the same reason as the two fermion mixture is stable. Note that
the case in which the two atoms in the molecule are not in the lowest energy
internal states, such as molecules created using 85 Rb, is quite different [144, 146].

69

CHAPTER 7. INELASTIC COLLISIONS

These 85 Rb Feshbach molecules will spontaneously dissociate as observed in Ref.

[159]. For our 40 K molecules however we again expect that any decay processes
will require more than two fermions, for example [95, 160]
X + XY X + (XY )

(7.3)

XY + XY XY + (XY ) .

(7.4)

The first process is reminiscent of Eqn. 7.2 above. These processes are often
referred to as collisional quenching of vibrations [161, 162, 163]. Again we expect
some suppression of these decay channels due to Fermi statistics since two identical
fermions must approach each other, as shown schematically in Fig. 7.1.

(a)

(b)

Figure 7.1: Particles involved in inelastic collisions in a Feshbach molecule/atom

mixture. (a) Illustration of Eqn. 7.3 (b) Illustration of Eqn. 7.4.

7.2

Lifetime of Feshbach molecules

In this section I present experimental data on the stability of a mixture of atoms

and Feshbach molecules. To obtain these data we created a molecule sample at
the mf = 7/2, 9/2 Feshbach resonance in which typically 50% of original atom
gas was converted to molecules. We then measured the molecule number as a
function of time while holding the molecule/atom mixture in a relatively shallow
optical dipole trap [67]. Figure 7.2 shows the result of this measurement at a
variety of magnetic fields on the BEC side of the Feshbach resonance. The plot
shows N /N versus the atom-atom scattering length a. Here N is the number of
molecules and N is the initial linear decay rate.

70

CHAPTER 7. INELASTIC COLLISIONS

N/N (ms-1)

We found that far from resonance the molecules decay quickly, but the decay
rate changes by orders of magnitude as the Feshbach resonance is approached.
Physically this effect is partially related to the overlap between the wavefunctions
of the XY molecule and the (XY ) molecule. As the Feshbach resonance is
approached the XY molecules become extremely weakly bound and quite large,
and hence they have less overlap with the small (XY ) molecules.

0.1
0.01
0.001

1000

10000
a (a0)

Figure 7.2: Feshbach molecule loss rate as a function of the atom-atom scattering
length near a Feshbach resonance in 40 K [67]. N here is the number of molecules.
The line is a fit of the closed circles () to a power law. The open circles () are
data for which the pair size is expected from two-body theory to be larger than
the interparticle spacing.
A scaling law for the dependence of the molecule decay rate upon the atomatom scattering length a was found in [95] and later in [157] for the processes
described by Eqns. 7.3 and 7.4 (Fig. 7.1). The scaling law was found by solving
the full few-body problem in the limit where the molecules are smaller than the
interparticle spacing, yet a r0 . Physical effects important to the result are
the Fermi statistics and the wavefunction overlap. The prediction for Eqn. 7.3
(molecule-atom collisions) is that the decay rate should scale with a3.33 and for
Eqn. 7.4 (molecule-molecule collisions) with a2.55 .
Since our measurement was carried out with thermal molecules the density of
the molecule gas remains approximately constant over the a = 1000 a0 to 3000 a0
range. (The total peak atom density in one spin state in the weakly interacting
regime was n0pk = 7.5 1012 cm3 .) Thus, we could measure the power law

CHAPTER 7. INELASTIC COLLISIONS

71

by fitting the data in Fig. 7.2 to the functional form Cap , where C and p are
constants. We included only points for which the interatomic spacing in the center
of the sample was larger than the expected size of an isolated molecule, a/2. We
found p = 2.30.4, consistent with the predicted power law for molecule-molecule
collisions. A similar power law was observed in a gas of Li2 molecules at the 834
G Feshbach resonance [164].
In general we find that the lifetime of the molecules is surprisingly long near the
Feshbach resonance. The molecule lifetime for magnetic fields at which a > 3000
a0 is greater than 100 ms. This is much longer than lifetimes observed in bosonic
systems for similar densities and internal states [165, 166]. 100 ms is actually
a long time compared to many other time scales in our Fermi gas such as the
time scale for two-body adiabaticity, the average time between elastic collisions,
and the radial trap period. This comparison suggested that it would indeed be
possible to study BCS-BEC crossover physics using atomic 40 K gases.

7.3

Three-body recombination

We also observe inelastic decay of fermionic atoms on the BCS side of the Feshbach
resonance, where the decay is due the process described by Eqn. 7.2 [57]. These
collisions cause both particle loss and heating. The heating can result from a
combination of two processes: First, the density dependence of the three-body
process results in preferential loss in high density regions of the gas [167]. Second,
the products of the inelastic collision, which have kinetic energies on order the
binding energy of the XY molecules, can collide with other particles on the way
out of the sample causing transfer of energy to the gas.
We found that one relevant measure of the effect of inelastic decay processes
on our ability to study BCS-BEC crossover physics is the heating of our Fermi gas
during magnetic-field ramps that are sufficiently slow to be adiabatic compared
to many-body time scales (see Ch. 8). We performed an experiment in which
we approached the Feshbach resonance at rate of (6 ms/G)1 , waited 1 ms, and
then ramped back at the same slow rate to the weakly interacting regime. The
result of this experiment is shown in Fig. 7.3. If we started with a gas initially
at T /TF = 0.10, T /TF upon return increased by less than 10% for a ramp to

72

CHAPTER 7. INELASTIC COLLISIONS

T/TF upon return

1/kF0 a = 0 (yet by 80% for a ramp to 1/kF0 a = 0.5). kF0 is the Fermi wavevector
measured in the weakly interacting regime. For this measurement the initial peak
density in one spin state was n0pk = 1.2 1013 cm3 .

0.3
0.2
0.1
0.0

-1

-2

-3

1/kF0a
Figure 7.3: T /TF measured in the weakly interacting regime as a function of the
final magnetic field in an adiabatic ramp towards the Feshbach resonance and
back [78]. The magnetic field is represented through the dimensionless parameter
1/kF0 a.

7.4

Comparison of

40

K and 6Li

In both 40 K and 6 Li gases the scaling law of the decay rate as a function of
molecule size was found to be the same [67, 164]. However, comparison of the
absolute decay rates for similar densities and a 5000 a0 shows that the inelastic
decay of 40 K occurs more than two orders of magnitude faster than at the broad
6
Li Feshbach resonance [39, 164, 122]. This difference has so far not been fully
explained theoretically. However, it must be related in some way to the difference in the full three-body potentials for 40 K versus 6 Li atoms. This difference in
lifetime between the two atomic species affects how experiments in the BCS-BEC
crossover are approached in 40 K versus 6 Li. For example, in our experiments we
obtain the coldest gases in 40 K by evaporating a Fermi gas in the weakly interacting regime where inelastic decay is negligible. Experimenters using 6 Li, however,

CHAPTER 7. INELASTIC COLLISIONS

73

obtain their coldest gases through evaporation near the Feshbach resonance or
slightly to the molecular side of the resonance [168, 69].
In our discussions thus far we have only considered inelastic collisions in which
the final state is a more deeply bound molecule. In the 6 Li systems experimenters
make use of inelastic collisions that interconvert atoms on the repulsive side of the
Feshbach resonance and Feshbach molecules [168, 69]. The rate of these inelastic
collisions in the 6 Li system is faster than inelastic decay to more deeply bound

h2
molecules, and the binding energy of the Feshbach molecules Eb ma
2 is small
enough to not cause significant problems in the cooling process. Hence, they
typically start with a hot gas of atoms on the BEC side of the resonance and after
evaporative cooling observe a pure sample of Feshbach molecules [66, 69]. Thus,
in contrast to the 40 K experiments in this thesis, 6 Li experiments can operate with
chemical equilibrium between atoms and Feshbach molecules.

Chapter 8
Making condensates from a Fermi
gas of atoms
So far we have considered the normal state of the Feshbach resonance/Fermi
gas system and found that it has all the elements necessary to study BCS-BEC
crossover physics. However, the true test of whether we could access BCS-BEC
crossover physics with our atomic gas would be to observe a phase transition. The
phase transition could be distinguished through observation of the onset of either
superfluid behavior or condensation. Due to the linked nature of these phenomena
one would necessarily imply the other. Just as with alkali BEC in 87 Rb and 23 Na
(Fig. 1.3), the observable of choice for the first experiments to observe this phase
transition in the BCS-BEC crossover was condensation.
In this chapter I will discuss how we were able to show condensation of
fermionic atom pairs in the BCS-BEC crossover regime. This demonstration relied heavily upon our previous knowledge of the normal state of a Fermi gas at
a Feshbach resonance. First, I present our observation of condensation of Feshbach molecules to create one of the first molecular BECs [68]. This work led
the way to observation of condensation of fermionic atom pairs in the crossover
regime. Then I present a phase diagram of the BCS-BEC crossover regime attained through measurements of condensate fraction.

74

CHAPTER 8. CONDENSATES IN THE CROSSOVER

8.1

75

Emergence of a molecular condensate from

a Fermi gas of atoms

We have seen that a Feshbach resonance can be used to create a large number of
ultracold molecules starting with a Fermi gas of atoms. After observing that these
molecules can be long lived, the creation of a BEC from these bosonic Feshbach
molecules was an obvious goal. Previously we had created molecules by applying a
magnetic-field ramp just slow enough to be two-body adiabatic; in the experiments
here the idea will be to apply a magnetic-field ramp that is not only two-body
adiabatic, but also slow with respect to the many-body physics timescale in our
gas (many-body adiabatic). With such a magnetic-field ramp across the Feshbach
resonance the entropy of the original quantum Fermi gas should be conserved
[65, 169]. For an initial atom gas with a sufficiently low T /TF the result should
be a low entropy sample of bosonic molecules, which for a low enough entropy is
a BEC.
To pursue this idea experimentally we again used the Feshbach resonance
between the mf =9/2 and mf =7/2 spin states starting with a Fermi gas at
temperatures below quantum degeneracy. We applied a time-dependent ramp of
the magnetic field starting above the Feshbach resonance and ending below the
resonance. The magnetic field was typically ramped in 7 ms from B = 202.78 G
to either B = 201.54 G or B = 201.67 G, where a sample of 78% to 88% Feshbach
molecules was observed. A critical element of this experiment is that the lifetime
of the Feshbach molecules can be much longer than the typical collision time in
the gas and longer than the radial trapping period (see Ch. 7). The relatively long
molecule lifetime near the Feshbach resonance allows the atom/molecule mixture
to achieve thermal equilibrium during the magnetic-field ramp. Note however that
since the optical trap used for these experiments is strongly anisotropic (r /z
80) we may attain only local equilibrium in the axial direction.
To study the resulting atom-molecule gas mixture after the magnetic-field
ramp, we measured the momentum distribution of both the molecules and the
residual atoms using time-of-flight absorption imaging. After typically 10 to 20 ms
of expansion we applied a radio frequency (rf) pulse that dissociated the molecules
into free atoms in the mf =5/2 and mf =9/2 spin states [63]. Immediately af-

CHAPTER 8. CONDENSATES IN THE CROSSOVER

76

ter this rf dissociation pulse we took a spin-selective absorption image. The rf

pulse had a duration of 140 s and was detuned 50 kHz beyond the molecule
dissociation threshold where it did not affect the residual unpaired atoms in the
mf =7/2 state. We selectively detected the expanded molecule gas by imaging
atoms transferred into the previously unoccupied mf =5/2 state by the rf dissociation pulse. Alternatively we could image only the expanded atom gas by
detecting atoms in the mf =7/2 spin state.
Close to the Feshbach resonance, the atoms and molecules are strongly interacting with effectively repulsive interactions. The scattering length for atommolecule and molecule-molecule collisions was calculated by Petrov et al. to be
1.2 a and 0.6 a respectively, for values of a much larger than r0 and smaller than
the interparticle spacing [95]. During the initial stage of expansion the positive
interaction energy is converted into additional kinetic energy of the expanding
gas. Therefore, the measured momentum distribution is very different from the
original momentum distribution of the trapped gas. In order to reduce the effect
of these interactions on the molecule time-of-flight images we used the magneticfield Feshbach resonance to control the interparticle interaction strength during
expansion. We could significantly reduce momentum kick due to the interaction
energy by rapidly changing the magnetic field before we switched off the optical
trap for expansion. The field was typically lowered by 4 G in 10 s. At this
magnetic field farther away from the resonance the atom-atom scattering length
a was reduced to 500 a0 . We found that this magnetic-field jump resulted in a
loss of typically 50% of the molecules, which is a result of the reduced molecule
lifetime away from the Feshbach resonance.
To attempt to observe condensation of molecules we monitored the molecule
momentum distribution while varying the temperature of the initial weakly interacting Fermi gas, (T /TF )0 . Below (T /TF )0 of 0.17 we observed the sudden onset of
a pronounced bimodal momentum distribution. Figure 8.1 shows such a bimodal
distribution for an experiment starting with an initial temperature of 0.1 TF ; for
comparison the figure shows the resulting molecule momentum distribution for an
experiment starting at 0.19 TF . The bimodal momentum distribution is a striking
indication that the gas of weakly bound molecules has undergone a phase transition to a BEC [22, 23]. To obtain thermodynamic information about the molecule

77

optical density

CHAPTER 8. CONDENSATES IN THE CROSSOVER

2.0

2.0

1.5

1.5

1.0

1.0

0.5

0.5

0
-200 -100

100

position (mm)

200

-200 -100

100

200

position (mm)

Figure 8.1: Momentum distribution of a molecule sample created by applying a

magnetic-field ramp to an atomic Fermi gas with an initial temperature of 0.19 TF
(0.1 TF ) for the left (right) picture [68]. In the right sample the molecules form a
Bose-Einstein condensate. The lines illustrate the result of bimodal surface fits.

Figure 8.2: Molecular condensate fraction N0 /N versus the scaled temperature

T /Tc [68]. The temperature of the molecules is varied by changing the initial
temperature of the fermionic atoms prior to the formation of the molecules, yet
measured through the momentum distribution of the molecular thermal gas.

78

CHAPTER 8. CONDENSATES IN THE CROSSOVER

gas the momentum distribution is fit with a two-component distribution. The fit
function is the sum of an inverted parabola describing the Thomas-Fermi momentum distribution of a bosonic condensate and a gaussian momentum distribution
describing the noncondensed component of the molecule gas. In Fig. 8.2 the
measured condensate fraction is plotted as a function of the fitted temperature of
the molecular thermal component in units of the critical temperature for an ideal
Bose gas Tc = 0.94(N r2 z )1/3 h/kB . In this calculated Tc , N is the total number of
molecules measured without changing the magnetic-field for the expansion. The
critical temperature for the strongly interacting molecules measured from Fig. 8.2
is 0.8 0.1 Tc . Such a decrease of the critical temperature relative to the ideal
gas prediction is expected for a strongly interacting gas [170].

N0 / N

0.15
0.10
0.05

4
6
8
ramp time (ms)

10

Figure 8.3: Time scale for many-body adiabaticity [68]. The fraction of condensed
molecules is plotted versus the time in which the magnetic field is ramped across
the Feshbach resonance from 202.78 G to 201.54 G.
As expected we found that the creation of a BEC of molecules requires that the
Feshbach resonance be traversed sufficiently slowly to be many-body adiabatic.
This many-body time scale should be determined by the time it takes atoms
to collide and move in the trap. In Fig. 8.3 the measured condensate fraction
is plotted versus the ramp time across the Feshbach resonance starting with a
Fermi gas at a temperature 0.1 TF . Our fastest ramps resulted in a much
smaller condensate fraction while the largest condensate fraction appeared for a

CHAPTER 8. CONDENSATES IN THE CROSSOVER

79

magnetic-field ramps slower than 3 ms/G.

energy (nK)

15
10
5
0
0

500

1000

1500

a (a0)
Figure 8.4: Expansion energy of the molecular condensate versus the interaction
strength during expansion. In this measurement the BEC is created in a regime
where the atom-atom scattering length is 3300 a0 . For time-of-flight expansion
the magnetic field is rapidly changed to different final values, characterized by the
atom-atom scattering length a.

Rapidly changing the interaction strength for time-of-flight expansion of the

condensate allowed us to measure the interaction energy in the molecular sample.
Figure 8.4 shows a plot of the expansion energy of the molecule BEC for various
interaction strengths during time-of-flight expansion. Here the condensate was
created at a fixed interaction strength, and thus the initial peak density npk was
constant. The linear dependence of the energy upon a suggests that the moleculemolecule scattering length is proportional to the atom-atom scattering length as
predicted in Ref. [95]. In addition the expansion energy extrapolates to near zero
energy for a=0. This is consistent with a Bose-Einstein condensate of molecules.
Assuming the molecule-molecule interaction strength calculated in Ref. [95], the
peak density of the strongly interacting condensate was npk = 7 1012 cm3 .
In conclusion, I have discussed the creation of a BEC of weakly bound molecules
starting with a gas of ultracold fermionic atoms. With a relatively slow ramp of
an applied magnetic field that converts most of the fermionic atoms into bosonic
molecules and an initial atomic gas below T /TF = 0.17, we observed a molecular
condensate in time-of-flight absorption images taken immediately following the

CHAPTER 8. CONDENSATES IN THE CROSSOVER

80

magnetic-field ramp. Our experiment approaches the BEC limit, in which superfluidity occurs due to BEC of essentially local pairs whose binding energy is larger
than the Fermi energy. Strikingly, our 40 K molecular condensate is not formed by
any active cooling of the molecules, but rather merely by traversing the BCS-BEC
crossover regime. At the same time as these experiments in 40 K, experiments using 6 Li created a similar BEC of Feshbach molecules. Their approach however
was direct evaporative cooling of the Feshbach molecules [69, 70].

8.2

Observing condensates in the crossover

To create a molecular condensate we started with a quantum Fermi gas, slowly

traversed the BCS-BEC crossover regime, and ended up with a BEC of molecules.
An obvious question was whether condensation also had occurred in the crossover
regime that we had passed through. To answer this question we needed to overcome a number of challenges. First, we required a probe of the momentum distribution of pairs in the crossover. In the BEC limit the momentum distribution of
the molecules could be measured using standard time-of-flight absorption imaging.
However, this method is problematic in the crossover because the pairs depend on
many-body effects and are not bound throughout expansion of the gas. Second,
to prove observation of condensation in the BCS-BEC crossover regime we had to
show that we were not simply seeing condensation of pairs in the two-body bound
state (two-body pairs), but rather condensation of pairs requiring many-body effects to form (many-body pairs). A clear example of condensation of many-body
pairs would be condensation on the a < 0, or BCS, side of the Feshbach resonance.
Here the two-body physics of the resonance no longer supports the weakly bound
molecular state; hence, only many-body effects can give rise to a condensation of
fermion pairs.
To solve the problem of measuring the momentum distribution of pairs in
the crossover we introduced a technique that took advantage of the Feshbach
resonance to pairwise project the fermionic atoms onto Feshbach molecules. We
were able to probe the system by rapidly ramping the magnetic field to the BEC
side of the resonance, where time-of-flight imaging could be used to measure the
momentum distribution of the weakly bound molecules. The projecting magnetic-

CHAPTER 8. CONDENSATES IN THE CROSSOVER

81

field ramp was completed on a timescale that allowed molecule formation but was
still too brief for particles to collide or move significantly in the trap. This is
possible due to the clear separation of the two-body and many-body time scales.
The timescale for many-body adiabaticity in Fig. 8.3 is two orders of magnitude
longer than the timescale for two-body adiabaticity shown in Fig. 6.7.
The key to the second problem, verifying condensation of many-body pairs,
came from careful understanding of the two-body physics. As discussed in Ch. 6
we were able to precisely measure the magnetic-field position above which a twobody bound state no longer exists, B0 (Fig. 6.9). If we observed condensation of
fermionic atom pairs at B > B0 we could be assured that these were pairs that
were the result of many-body effects.
To perform experiments making use of these ideas, we continued with the
same experimental setup as the last section where we discussed the creation of
molecular condensates. We initially prepared the ultracold two-component atom
gas at a magnetic field of 235.6 G, far above the Feshbach resonance. Here the
gas is not strongly interacting, and we measured (T /TF )0 through surface fits to
time-of-flight images of the Fermi gas (Ch. 4). The field was then slowly lowered
at the many-body adiabatic rate of 10 ms/G to a value of Bhold near the resonance.
Whereas before we had considered only values of Bhold below B0 on the BEC side
now we explore the behavior of the sample when ramping slowly to values of Bhold
on either side of the Feshbach resonance.
To probe the system we projected the fermionic atoms pairwise onto molecules
and measured the momentum distribution of the resulting molecular gas. This
projection was accomplished by rapidly lowering the magnetic field by 10 G
at a rate of (40 s/G)1 while simultaneously releasing the gas from the trap
(Fig. 8.5). This put the gas far on the BEC side of the resonance, where it was
weakly interacting. After a total of typically 17 ms of expansion the molecules
were selectively detected using rf photodissociation immediately followed by spinselective absorption imaging. To look for condensation, these absorption images
were again surface fit to a two-component function that was the sum of a ThomasFermi profile for a condensate and a gaussian function for noncondensed molecules.
Figures 8.6 and 8.7 present the main result of this section. In Fig. 8.6(a) is the
measured condensate fraction N0 /N as a function of the magnetic-field detuning

CHAPTER 8. CONDENSATES IN THE CROSSOVER

82

B (gauss)

204
202
200
198
-9

-6

-3

3 16 18

t (ms)
Figure 8.5: A typical magnetic-field ramp used to measure the fraction of condensed fermionic atom pairs [73]. The system is first prepared by a slow magneticfield ramp towards the resonance to a variable position Bhold (two-sided arrow).
After a time thold the optical trap is turned off at t = 0 and the magnetic field
is quickly lowered by 10 G. After expansion, the molecules are imaged on the
BEC side of the resonance ().

CHAPTER 8. CONDENSATES IN THE CROSSOVER

83

from the resonance, B = Bhold B0 . The data in Fig. 8.6(a) were taken for
a Fermi gas initially at T /TF = 0.08 and for two different wait times at Bhold .
Condensation was observed on both the BCS (B > 0) and BEC (B < 0)
sides of the resonance. We further found that the condensation that occurs on
the BCS side of the Feshbach resonance was distinguished by its longer lifetime.
The lifetime was probed by increasing thold to 30 ms (triangles in Fig. 8.6(a));
this time was much longer than the previously measured lifetime of the molecular
condensate in the BEC limit [68]. Not surprisingly then we found that for the
BEC side of the resonance no condensate was observed for thold = 30 ms except
very near the resonance. However, for all data on the BCS side of the resonance
the observed condensate fraction was still > 70% of that measured for thold = 2
ms.
An essential aspect of these measurements is the fast magnetic-field ramp that
projects the fermionic atoms pairwise onto molecules. It is a potential concern
that the condensation might occur during this ramp rather than at Bhold . To
verify that condensation did not occur during the ramp we studied the measured
condensate fraction for different magnetic-field ramp rates. Figure 8.6 compares
the condensate fraction measured using the 40 s/G (circles) rate to that using
a ramp that was 7 times faster (open diamonds). We found that the measured
condensate fraction was identical for these two very different rates, indicating
that this measurement constitutes a projection with respect to the many-body
physics. The validity of the magnetic-field projection technique was also explored
in studies of a 6 Li gas at MIT. Researchers there first reproduced the observation of
condensation using the pairwise projection technique with a 6 Li gas [74]. They also
monitored the delayed response of the many-body system after modulating the
interaction strength [171]. They found that the response time of the many-body
system was slow compared to the rate of the rapid projection magnetic-field ramp.
There have been a number of theoretical papers on the subject of the pairwise
projection technique for measuring condensate fraction in the crossover [172, 173,
174]. Work thus far has established that observation of condensation of molecules
following a rapid projection ramp indicates the pre-existence of condensation of
fermionic atom pairs before the projection ramp.
To summarize, in this section we have introduced a method for probing the

84

CHAPTER 8. CONDENSATES IN THE CROSSOVER

N0 / N

0.15

(a)

0.10
0.05

N0 / N

0
0.15

(b)

0.10
0.05
0
-0.6 -0.4 -0.2

0.2 0.4 0.6 0.8

DB (gauss)
Figure 8.6: Measured condensate fraction as a function of detuning from the
Feshbach resonance B = Bhold B0 [73]. (a) Data for thold = 2 ms () and
thold = 30 ms () with an initial gas at T /TF = 0.08. (b) Data for two different
projection magnetic-field ramp rates: 40 s/G () and 6 s/G (). The dashed
lines B = 0 reflect the uncertainty in the Feshbach resonance position.

CHAPTER 8. CONDENSATES IN THE CROSSOVER

85

momentum distribution of fermionic atom pairs and seen how this technique could
be employed to observe condensation near a Feshbach resonance. By projecting
the system onto a molecule gas, we observed condensation of fermionic pairs as
a function of the magnetic-field detuning from the resonance as shown in Fig.
8.7. While Fig. 8.7 is reminiscent of Fig. 1.3 where condensation was observed a
function of T /Tc , note that the condensate here actually appears as a function of
interaction strength, not temperature.

Figure 8.7: Time-of-flight images showing condensation of fermionic atom pairs.

The images, taken after the projection of the fermionic system onto a molecule
gas, are shown for B = 0.12, 0.25, and 0.55 G (right to left) on the BCS side of
the resonance. The original atom gas starts at (T /TF )0 = 0.07. The 3D artistry
is courtesy of Markus Greiner.

8.3

Measurement of a phase diagram

In addition to varying B and measuring the condensate fraction, we can also

vary the initial temperature of the Fermi gas. Figure 8.8 is a phase diagram

86

CHAPTER 8. CONDENSATES IN THE CROSSOVER

created from data varying both B and (T /TF )0 . B is converted to the dimensionless parameter 1/kF0 a, where a is calculated directly from B through
Eqn. 3.1 and kF0 is extracted from the weakly interacting Fermi gas. The colors represent the measured condensate fraction using the projection technique.
The boundary between the light and dark blue regions shows where the phase
transition occurs in the BCS-BEC crossover. On the BCS side of the resonance
the condensate forms for higher initial T /TF as B decreases, as expected based
upon BCS-BEC crossover theories (Ch. 2).

0.20

N0/N
-0.020
0.010
0.025
0.050
0.075
0.100
0.125
0.150
0.175

(T/TF)0

0.15
0.10
0.05
0
1.0

0.5

-0.5 -1.0 -1.5

1/(kF0a)
Figure 8.8: Transition to condensation as a function of both B and (T /TF )0 [73].
The data for this phase diagram were collected using the same procedure as in
Fig. 8.5 with thold 2 ms. The false color surface and contour plot were obtained
using a Renka-Cline interpolation of approximately 200 distinct data points.

These data lie precisely in the regime that is neither described by BCS nor
by BEC physics, 1 < 1/kF a < 1 (see Fig. 2.3). The condensed pairs in these
experiments are pairs with some qualities of diatomic molecules and some qualities
of Cooper pairs. Thus, these experiments realized a phase transition in the BCSBEC crossover regime and initiated experimental study of this physics.
Finally, as in our previous measurements performed in the BEC limit, the mea-

CHAPTER 8. CONDENSATES IN THE CROSSOVER

87

sured condensate fraction in Fig. 8.8 always remains well below one [68]. This is
not observed in the case of 6 Li experiments [74], suggesting that technical issues
particular to 40 K may play a role. As part of our probing procedure the magnetic
field is set well below the Feshbach resonance where the molecule lifetime is only
on the order of milliseconds [67, 95]. This results in a measured loss of 50% of
the molecules and may also reduce the measured condensate fraction. Furthermore, temperature measurements using the surface-fitting procedure become less
accurate below T /TF = 0.1 (see Sec. 4). It also is expected that the isentropic
ramp from a weakly interacting Fermi gas to the crossover results in a larger value
of T /TF than (T /TF )0 . The extent of this adiabatic heating is currently under
theoretical investigation [131]. Density-dependent loss processes could also play
a role in heating of the sample, especially as the BEC limit is approached (see
Ch. 7). Whether any or all of these processes play a role in the small condensate
fraction is a subject of current study in the Jin group.

Chapter 9
The momentum distribution of a
Fermi gas in the crossover
The measurements of condensation in the crossover described in the last chapter probed the phase coherence between fermionic atom pairs. As discussed in
Ch. 2, it is only in the BCS limit that the pairs are always coherent, and when
the interaction becomes large so-called pre-formed pairs are predicted to exist
above the phase transition temperature Tc [8, 18]. To verify such theories it is
important that techniques for detecting pairing be developed alongside studies
probing the phase transition. There have been a number of probing techniques
used to detect pairing including rf spectroscopy [80], magnetic-field modulation
[81], spectroscopic probes [72], and measurements of the atomic momentum distribution [62, 78]. This chapter focuses on measurements of the atomic momentum
distribution in the BCS-BEC crossover using 40 K [78].
The classic characteristic change of the momentum distribution of a superfluid
Fermi system is a broadening of the Fermi surface (see for example [92]). Figure
9.1 (inset) shows the expected momentum distribution of a homogeneous, zerotemperature Fermi system. In the BCS limit (1/kF a ) the amount of
broadening is small and associated with . As the interaction increases this
effect grows until at unitarity (1/kF a = 0) the effect is on order of EF , and in
the BEC limit (1/kF a ) the momentum distribution becomes the square of
the Fourier transform of the molecule wavefunction (see for example [48]). This
kinetic energy increase can be interpreted as a cost of pairing. Amazingly, the
88

CHAPTER 9. MOMENTUM DISTRIBUTIONS IN THE CROSSOVER

1.0

0.8

n(k)

n(k/kF0) kF0 2 / N

1.0

0.5

0.6
0.4

0
0

0.2
0
0

89

0.5

1.0

1.5

k/kF

2.0

2.5

k/k 0
F

Figure 9.1: Theoretical column integrated momentum distributions of a trapped

Fermi gas n(k)
calculated for T = 0 using NSR theory [175]. The normalization is
R
given by 2 n(k)kdk = N . The lines correspond to a = 0 (blue), 1/kF0 a = 0.66
(green), 1/kF0 a = 0 (red), and 1/kF0 a = 0.59 (black). (inset) Corresponding
distributions for a homogeneous system.

total energy of the system is lowered despite this large kinetic energy increase.
Here I will discuss the atom momentum distribution of a trapped Fermi gas.
The case of an inhomogeneous trapped gas is more complicated than the homogeneous case. Nonetheless, in the strongly interacting regime the momentum distribution probes the pair wavefunction and consequently the nature of the pairs:
Small, tightly bound pairs will broaden the momentum distribution more than
large, weakly bound pairs. Similarly a fully paired gas will broaden the distribution more than a gas with a small fraction of paired atoms. It is expected that
this measurement will probe the pairs independent of whether they have formed
a condensate. Throughout this chapter I will compare our data to theory of the
momentum distribution of a trapped gas. This theoretical work is the result of a
collaboration between our group, Murray Holland of JILA, and Stefano Giorgini,
a JILA visiting fellow from Trento, Italy.

CHAPTER 9. MOMENTUM DISTRIBUTIONS IN THE CROSSOVER

9.1

90

Measuring the momentum distribution

We found that the momentum distribution of a Fermi gas in the crossover could
actually be measured using the standard technique of time-of-flight expansion followed by absorption imaging [22]. The key to measuring the atom momentum
distribution is that the gas must expand freely without any interatomic interactions; to achieve this we quickly changed the scattering length to zero for the
expansion. This was particularly convenient using 40 K because the zero crossing
of the scattering length occurs only 7.8 G above B0 . Bourdel et al. pioneered
this type of measurement using a gas of 6 Li atoms at T /TF 0.6, where TF is
the Fermi temperature [62]. In this work we carried out measurements down to
T /TF 0.1, where pairing becomes a significant effect and condensates have been
observed [73, 74].
To understand what we expect for our trapped atomic system, we can predict
the atomic momentum distribution using a local density approximation and the
results for the homogeneous case. In the trapped gas case, in addition to the local
broadening of the momentum distribution due to pairing, attractive interactions
compress the density profile and thereby enlarge the overall momentum distribution. Figure 9.1 shows a calculation of an integrated column density from the
result of a mean-field calculation at T = 0 as described in Ref. [175].
First, I will discuss the atomic momentum distribution measured with a low
temperature Fermi gas. We started with a weakly interacting mf = 7/2, 9/2
gas at T = 0.12 TF in a trap with a radial frequency of r = 280 Hz and
z /r = 0.071.1 We then adiabatically increased the interaction strength by ramping the magnetic field at a rate of (6.5 ms/G)1 to near the mf = 7/2, 9/2
Feshbach resonance. After a delay of 1 ms, the optical trap was switched off and
simultaneously a magnetic-field ramp to a 0 (B = 209.6 G) at a rate of (2
s/G)1 was initiated. The rate of this magnetic-field ramp was designed to be
fast compared to typical many-body timescales as determined by EhF = 90 s.
The gas was allowed to freely expand for 12.2 ms, and then an absorption image is taken. The imaging beam propagated along z and selectively probed the
1

For this measurement we introduced the use of a crossed dipole trap configuration. In
addition to the usual beam with a 15 m waist, we focused a w=200 m beam oriented parallel
to the force of gravity (
y ) on the atoms.

CHAPTER 9. MOMENTUM DISTRIBUTIONS IN THE CROSSOVER

91

0.8

OD k3 (arb)

normalized OD

1.0

0.6
0.4
0.2
0
0

0.5

4
2
0
-2
4
2
0
-2

1.0

1.5

2.0

2.5

k/k 0
F

Figure 9.2: Experimental, azimuthally averaged, momentum distributions of a

trapped Fermi gas at (T /TF )0 = 0.12 normalized such that the area under the
curves is the same as in Fig. 9.1 [78]. The curves correspond to 1/kF0 a = 71
(blue), -0.66 (green), 0 (red), and 0.59 (black). Error bars represent the standard
deviation of the mean of averaged pixels. (inset) Curves for 1/kF0 a = 71 (top)
and 0 (bottom) weighted by k 3 . The lines are the results of a fit to Eqn. 9.1.

mf =9/2 state.
Samples of these absorption images, azimuthally averaged, are shown in Fig.
9.2 for various values of 1/kF0 a. There is a dramatic change in the distribution
that is qualitatively very similar to the prediction in Fig. 9.1. Some precautions
need to be taken in the quantitative comparison of Figs. 9.1 and 9.2. First,
the magnetic-field ramp to the Feshbach resonance, while adiabatic with respect
to most time scales, is not fully adiabatic with respect to the axial trap period.
Second, in the experiment an adiabatic field ramp keeps the entropy of the gas, not
T /TF , constant. However, the resulting change in T /TF should have a minimal
effect on the distribution for 1/kF0 a < 0 [131]. Third, the theory assumes T = 0
and does not include the Hartree term, thus underestimating the broadening on
the BCS side compared to a full theory [103].

CHAPTER 9. MOMENTUM DISTRIBUTIONS IN THE CROSSOVER

9.2

92

Extracting the kinetic energy

It is natural now to consider extracting the kinetic energy from the momentum
distribution. While the momentum distribution should be universal for small momenta, for large momenta it is influenced by details of the interatomic scattering
potential. In the extreme case of a delta potential, which we used for the calculation in Fig. 9.1, the momentum distribution has a tail with a 1/k 4 dependence,
giving rise to a divergence of the kinetic energy. In the experiment we avoid a
dependence of the measured kinetic energy on details of the interatomic potential because the magnetic-field ramp is never fast enough to access features on
the order of the interaction length of the Van der Waals potential, r0 60 a0
for 40 K [117]. Thus, the results presented here represent a universal quantity,
independent of the details of the interatomic potential. Although universal in
this sense, the measured kinetic energy is intrinsically dependent on the dynamics
of the magnetic-field ramp, with faster ramps corresponding to higher measured
energies.
To obtain the kinetic energy from the experimental data exactly we would
need to take the second moment of the distribution, which is proportional to
P 3
P
k OD/ kOD. As illustrated in Fig. 9.2 (inset) this is difficult due to the
decreased signal-to-noise ratio for large k. Thus, our approach was to apply a 2D
surface fit to the image and extract an energy from the fitted function. In the limit
of weak interactions the appropriate function is that for an ideal, harmonically
trapped Fermi gas (see Ch. 4)

OD(x, y) = ODpk Li2 (e

x2
2
2x

y2
2
2y

)/Li2 ().

(9.1)

The kinetic energy per particle is then given by

Ekin =

3 mx y Li4 ()
2 t2 Li3 ()

(9.2)

where t is the expansion time.2 Although Eqn. 9.1 is not an accurate theoretical
2

This calculation gives the total Ekin in all three dimensions. Since the momentum distribution was only measured in two directions, Eqn. 9.2 requires the assumption that the third
dimension reveals an identical distribution.

Ekin / Ekin0

CHAPTER 9. MOMENTUM DISTRIBUTIONS IN THE CROSSOVER

6
5
4
3
2
1
0

93

1.4
1.2
1.0

1.0 0.5

-2

-4

-6

0 -0.5 -1.0 -1.5

1/kF0 a

Figure 9.3: The measured energy Ekin of a Fermi gas at at (T /TF )0 = 0.12
0
normalized to Ekin
= 0.25 kb K. The red line is the expected energy ratio from
a calculation only valid in the weakly interacting regime (1/kF0 a < 1). The
0
green bar represents the expected value of Ekin /Ekin
at unitarity just due to
0
density profile rescaling. In the molecule limit (1/kF a > 1) we calculate the
expected energy for an isolated molecule (blue line). (inset) A focus on the weakly
interacting regime with the same axes definitions as the main graph.

description of the Fermi gas in the crossover, empirically we found it to be a

fitting function that describes the data reasonably well throughout the crossover,
as illustrated in Fig. 9.2 (inset).
Figure 9.3 shows the result of extracting Ekin as a function of 1/kF0 a; we see
that Ekin more than doubles between the noninteracting regime and unitarity.
Using the fitting function of Eqn. 9.1 we could also extract information about the
shape of the distribution through the parameter . Since can range from -1 to
it is convenient to plot the quantity ln(1 + ) (Fig. 9.4). The shape evolves
smoothly from that of an ideal Fermi gas at T /TF 0.1 in the weakly interacting
regime, to a gaussian near unitarity, and to a shape more peaked than a gaussian
in the BEC regime. These qualitative features are predicted by the mean-field
calculation of the distributions in Fig. 9.1.

CHAPTER 9. MOMENTUM DISTRIBUTIONS IN THE CROSSOVER

94

ln (1+z)

4
0
-4
-8

0.5

-0.5

-1.0

-1.5

-2.0

1/kF0 a
Figure 9.4: Shape of the momentum distribution as described through the parameter (Eqn. 9.1). ln(1 + ) = 0 corresponds to a gaussian distribution, and for
an ideal Fermi gas ln(1 + )1 T /TF in the limit of low T /TF . The dashed lines
show the uncertainty in the Feshbach resonance position [73].

9.3

Comparing the kinetic energy to theory

As mentioned earlier Ekin of a trapped gas is affected both by the broadening due
to pairing (Fig. 9.1 (inset)) and by changes in the trapped gas density profile.
0
In the BCS limit, the broadening due to pairing scales with e/2kF |a| and is thus
exponentially small compared to density profile changes, which scale linearly with
0
kF0 |a|. In this limit we could calculate Ekin /Ekin
using a mean-field calculation in
2048 0
0
0
the normal state [105]; to lowest order in kF |a| we found Ekin /Ekin
= 945
2 kF |a|+1.
This result is plotted in Fig. 9.3 (inset) and shows good agreement for the weakly
interacting regime (1/kF0 a < 1). However, caution must be taken with the apparent agreement for 0.5 < 1/kF0 a < 1. The agreement could be explained by
the lack of pairing in this theory being compensated by the theorys overestimation
of the density profile change when |a| becomes larger than 1/kF .
In the crossover regime where the pairs are more tightly bound, pairing provides a significant contribution to the change in the momentum distribution. At
unitarity a full Monte Carlo calculation predicts the radius of the Fermi gas density profile will become (1 + )1/4 R0 = 0.81R0 , where R0 is the Thomas-Fermi
radius of a noninteracting Fermi gas [103]. Just this rescaling would result in

CHAPTER 9. MOMENTUM DISTRIBUTIONS IN THE CROSSOVER

95

0
Ekin /Ekin
= 1.54 (green bar in Fig. 9.3). Thus, at unitarity, pairing effects on the
momentum distribution must account for a large fraction of the measured value
0
of Ekin /Ekin
= 2.3 0.3 (Fig. 9.3) and all of the observed change in distribution
shape (Fig. 9.4).
In the BEC limit and at T = 0 the measured energy should be that of an
isolated diatomic molecule after dissociation by the magnetic-field ramp. Provided the scattering length associated with the initial molecular state, a(t = 0),
is much larger than r0 60 a0 , the wavefunction for the molecule is given by
= Aer/a(t=0) /r where r is the internuclear separation and A is a normalization
constant. M. Holland and S. Giorgini calculated the measured energy from the
solution of the Schrodinger equation with
a time-dependent boundary condition

d log(r)
1
on the two-particle wavefunction
, where a(t) is the scattering

= a(t)
dr
r=0
length fixed by the magnetic field at time t. In Fig. 9.3 is the result of this
calculation for a pure gas of molecules with zero center of mass momentum and a
(2 s/G)1 ramp rate. We found reasonable agreement considering that there is
a large systematic uncertainty in the theory prediction resulting from the experimental uncertainty in the magnetic-field ramp rate and also that this two-body
theory should match the data only in the BEC limit (1/kF0 a 1).
A greater theoretical challenge is to calculate the expected kinetic energy for
all values of 1/kF0 a in the crossover. This is a difficult problem because it requires
an accurate many-body wavefunction at all points in the crossover and the ability
to time-evolve this wavefunction. Recent work in Ref. [176] has carried out this
calculation using the NSR ground state (see Ch. 2 and [48]). Qualitatively the
result is sensible, but in the strongly interacting regime does not accurately reproduce the measured kinetic energy; this suggests that more sophisticated crossover
theories are necessary.

9.4

Temperature dependence

We also studied the dependence of the momentum distribution on (T /TF )0 . To

vary the temperature of our gas, we recompressed the optical dipole trap after
evaporation and heated the gas through modulation of the optical trap power. The
experimental sequence for measuring the momentum distribution was the same

DEkin / EF0

CHAPTER 9. MOMENTUM DISTRIBUTIONS IN THE CROSSOVER

1.0
0.8
0.6
0.4
0.2
0

96

(T/TF )0 = 0.11
0.13
0.20
0.3
0.5
0.7
1.0

0.5

-0.5

-1.0

-1.5

1/kF0 a
0
Figure 9.5: Temperature dependence of Ekin = Ekin Ekin
normalized to the
Fermi energy at a = 0. (T /TF )0 is the temperature of the noninteracting gas.

as in Sec. 9.1 except the ramp rate to a = 0 for expansion was (8 s/G)1 .
0
Figure 9.5 shows the measured kinetic energy change Ekin = Ekin Ekin
. Ideally
for this measurement we would have liked to have varied the temperature while
maintaining a fixed density. However, for these data the density does vary some.
For example, for the coldest dataset (black) the peak density, for atoms in one spin
state, in the weakly interacting regime is n0pk = 1.41013 cm3 and EF0 = 0.56 K.
For the hottest dataset (magenta) n0pk decreases to 6 1012 cm3 and EF0 = 0.79
K.
On the BEC side of the resonance, Ekin decreases dramatically with (T /TF )0 .
Because Ekin should be proportional to the molecule fraction, this result is
closely related to the recent observation that the molecule conversion efficiency
scales with T /TF [146]. In the strongly interacting regime we also observed a
decrease in Ekin with increasing (T /TF )0 . However, the decrease is not as steep
as would be expected if the phase transition, which occurs at (T /TF )0 0.15 as
measured through N0 /N (Fig. 8.8), were responsible for the change in the momentum distribution. Instead the temperature dependence of Ekin is consistent
with the expectation that the changes in the kinetic energy are caused by pairing
and not condensation [73, 80, 18].

Chapter 10
The apparatus
The apparatus used for the experiments described in this thesis originated during
the graduate career of Brian DeMarco; the elements constructed prior to 2001
are described in his thesis Quantum behavior of an atomic Fermi gas [83].
The technical work I completed involved, through the years, fixing a great deal
of the apparatus components and inserting additional capabilities required to
access BCS-BEC crossover physics. I will not document every change made to the
apparatus. Instead, I will first provide an overview of the current apparatus (as of
December 2005), namely a description of the components required for each step of
the cooling and trapping sequence and a report on the final cold gas. (A portion of
this information was already discussed in Ch. 4.) Then I will focus on six specific
topics that were particularly important to the BCS-BEC crossover experiments or
could be useful pieces of information for building future 40 K experiments. These
topics include our diode laser system, our optical dipole trap, the setup we use
to create a large magnetic field, our circuit for fast magnetic-field control, our rf
delivery system, and our absorption imaging configurations.

10.1

Procedure for making an ultracold

40

K gas

The cycle to create an ultracold potassium gas with our apparatus begins with
a vapor cell magneto-optical-trap (MOT). Since 40 K is a very rare isotope of
potassium (0.012% natural abundance), creating a large 40 K MOT required development of enriched potassium sources [178]. These sources, which we refer to as
97

98

CHAPTER 10. THE APPARATUS

Table 10.1: Hyperfine constants for the

[177].
I
gI
A 4S1/2
A 4P3/2
B 4P3/2

40

K levels of interest in our experiments

4
-1.298099/4
-285.731 MHz
-7.59 MHz
-3.5 MHz

f ' = 5/2 (54.5 MHz)

f ' = 7/2 (30.6 MHz)
f ' = 9/2 (-2.3 MHz)

4P3/2

f ' = 11/2 (-45.7 MHz)

766.7 nm

f = 7/2 (714.3 MHz)

4S1/2
f = 9/2 (-571.4 MHz)

Figure 10.1: Hyperfine structure of relevant ground and excited state levels of 40 K
[177].

CHAPTER 10. THE APPARATUS

99

getters, are based on the design of SAES Getters Inc. but are made in the JILA
shop. The JILA-made getters contain commercially available enriched KCl (5%
40
K) and a reducing agent, Ca. Enriched potassium is released through ohmic
heating of the getter inside an evacuated glass cell; this forms a vapor from which
40
K can be captured in a MOT. Because of the relatively low room-temperature
vapor pressure of potassium, we heat our vapor cell to 50 C to prevent potassium from sticking to the walls of the cell. Originally four potassium getters were
placed in the vacuum system. We currently use the third of these getters. This
particular getter has provided a constant 40 K vapor pressure per amount of ohmic
heating for over three years.
The 40 K MOT uses light red detuned of the f =9/2 f =11/2 transition (trap
light) and of the f =7/2 f =9/2 transition (repump light). Due to the small
excited-state hyperfine splitting of 40 K (Table 10.1 and Fig. 10.1), a relatively
large repump intensity is required. Our vapor cell MOT is aligned in a retroreflected configuration using laser beams with a 1.4 inch diameter. The total trap
power is 90 mW and the repump power 50 mW. Setting the potassium vapor
pressure such that the exponential MOT loading time is 1 sec, the result is a
40
K MOT containing over 109 atoms.
The vapor cell where we collect our potassium atoms we refer to as the collection cell (Fig. 10.2). The background pressure in this cell is too high to
evaporatively cool the atoms. Thus, for the next stage of the experiment we
transfer the atoms from the collection cell to another glass cell, the science cell,
which is separated from the collection cell by a long narrow tube (the transfer
tube) [123]. The limited conductance of the transfer tube isolates the collection
and science cells (Fig. 10.2), such that the vacuum pressure in the science cell is
characterized by an exponential atom lifetime of 150 sec. The UHV components
required to achieve this pressure are shown in Fig. 10.2; the main components
of the original system were two ion pumps and a titanium sublimation pump
(TSP). The ion pump along the transfer tube (P2) was turned off when it began
to adversely affect the vacuum. Now the system is only pumped by one 40 l/s
ion pump (P1). It has not been necessary to fire the titanium sublimation pump
since 2001.
To transfer the atoms from the collection cell to the science cell we use a

100

CHAPTER 10. THE APPARATUS

P1

TSP

P2 off

Collection
Cell

transfer tube

Science Cell

Figure 10.2: The vacuum system for our experiments as viewed from above; the
system has remained essentially unchanged since 1998. Figure adapted with permission from Ref. [83].

resonant laser beam to push the atoms down the transfer tube [83]. The atoms
are guided by hexapole magnets that surround the tube. In the science cell the
atoms are caught by a second MOT, the science MOT. The push beam, the
collection MOT, and the science MOT are currently all left on continuously as
the atoms are transferred. With a well-optimized alignment of the push beam,
this technique results in an exponential science MOT loading time of 10 sec. In
the science cell the atoms are held in what is known as a dark spot MOT; here
the repump beam intensity is significantly decreased at the location of the atom
gas, resulting in fewer collisional losses [179].
The next step is to load atoms from the science MOT into a magnetic trap
where they can be cooled using microwave evaporation. The magnetic trap we
use is a cloverleaf design Ioffe-Pritchard style trap (Fig. 10.3). Before loading
the atoms into the magnetic trap the MOT is moved to a position optimized for
loading into the center of the magnetic trap, the trap light detuning is moved closer
to resonance to provide as much Doppler cooling as possible (CMOT stage),1 and
1

Note this procedure is opposite of the requirement for sub-Doppler cooling. While subDoppler cooling has been observed in other potassium MOTs [180], we have never found it
beneficial in our apparatus.

101

CHAPTER 10. THE APPARATUS

the atoms are optically pumped towards the f =9/2, mf =9/2 state (Fig. 4.1).
We rely upon imperfect optical pumping to create the mixture of atoms in the
mf =9/2 and mf =7/2 spin states required for evaporative cooling. We succeed
in loading 55% of the atoms from the science MOT into a weak magnetic trap.
Following the load we compress the magnetic trap for evaporation to a trap with
a radial frequency of 250 Hz and an axial frequency of 20 Hz.

center
coil

bias coil

petal coil

Figure 10.3: Coils of the cloverleaf Ioffe-Pritchard magnetic trap. Figure adapted
with permission from Ref. [83]. The coils are wound with hollow copper tubing
through which cooling water flows. These coils are reused to create a homogeneous
magnetic field for accessing Feshbach resonances.

To remove the highest energy atoms for evaporative cooling in the magnetic
trap, microwaves between 1200 and 1300 MHz are used to transfer atoms to
untrapped spin states in the f =7/2 hyperfine state (Fig. 4.1) [32, 83]. The
microwaves are delivered to the atoms using a stub-tuned microwave coil designed
to deliver radiation in the 1200 - 1300 MHz range [83]. In the current procedure
only one microwave frequency is required at a given time for evaporation; this is in
contrast to the two-frequency evaporation scheme used to achieve two-component
quantum degenerate gases in Ref. [83]. Typically we only cool the 40 K sample in
the magnetic trap to T /TF 3 (T 5 K with a few 107 atoms) before loading
the atoms into an optical dipole trap.
The procedure for obtaining a cold gas of atoms in our optical trap in the
Feshbach resonance spin states includes the following. After loading the optical
trap we immediately begin forced evaporation by lowering the optical trap power.

CHAPTER 10. THE APPARATUS

102

(This process occurs continuously throughout the next steps.) We then transfer
the atoms from the mf =+9/2, +7/2 spin states to the mf =9/2, 7/2 spin states
using a rf frequency sweep across all the Zeeman transitions at B 20 G. Next
we move to a magnetic field above a Feshbach resonance, typically 235 G. This
large magnetic field is supplied mainly by the bias coils of Fig. 10.3. Here we
apply a /2 rf pulse on the mf =9/2 mf =7/2 transition to achieve an equal
mixture of the two spin states. We follow this spin transfer by further cooling at
high magnetic field. We then hold the atoms for typically 100 ms at a final optical
trap power that defines both the final evaporative cut and the parameters of the
final trap.
Experiments studying BCS-BEC crossover physics happen in a relatively short
amount of time at the end of the experiment cycle. The crossover experiments
mostly involve changing the magnetic field, either slowly with the large magnetic
field coils of Fig. 10.3 or very quickly with a low-inductance auxiliary coil pair
(Sec. 10.5). The final step in an experiment cycle is to acquire an absorption image
of the gas. It is important for the success of our experiments to have the ability
to take absorption images of the gas with high resolution and in many different
circumstances, i.e., at high and low magnetic fields, from multiple directions, and
with the ability to distinguish the different spin states.
Many of the tasks during the experiment cycle that I have now described
require precise timing. The heart of the system used for this timing is described
in Brian DeMarcos thesis [83]. The software for the system is quite old, dating
back to the JILA experiment of Ref. [124]. A QuickBASIC program is used
to control TTL, DAC, and GPIB boards, operating with a clock now provided
by a function generator operating at 3 kHz. The TTL and 12-bit DAC boards
were made by Keithley Instruments. We have added additional precise timing
that is controlled by the main timing system via TTL triggers. Precise timing
instruments we have added include function generators (both the Agilent 33220A
and the SRS DS345), two Quantum Composers pulse generators (Model 9614),
and a National Instruments 6733 analog board.
Analysis of the absorption images shows that our apparatus can cool a Fermi
gas to T /TF 0.1 or possibly colder (see Ch. 4). One figure of merit of the
apparatus, besides the final T /TF it can reach, is the number of atoms achieved at

CHAPTER 10. THE APPARATUS

103

a phase space density near quantum degeneracy. A large atom number (and hence
a large density) is not necessarily always a desirable feature; however, producing a
large enough atom number for ample signal has contributed to the success of many
of our experiments. Under optimum conditions we can achieve a two-component
gas at T /TF = 0.52 with 2 106 atoms per spin state. Thus, if our atoms were
bosons, we would have two Bose gases at Tc with 2 106 atoms in each gas. This
result is comparable to or better than many BEC experiments.

10.2

Diode laser system

The laser light for the trapping, cooling, and probing of 40 K in our apparatus is
provided by a semiconductor-diode laser system. The laser design is outlined in
Fig. 10.4; this schematic diagram does not represent the actual geometric layout
on our optics table and omits, for example, all mirrors and polarization rotation
optics. The two main components are external cavity diode lasers (ECDLs), which
provide narrow linewidth light (1 MHz) near the repump and trap transitions.
The first of these lasers, the repump laser, is stabilized to the f =2 transition of
39
K by locking to the peak of the corresponding saturated absorption spectroscopy
peak (Fig. 10.5). The locking light is shifted using an acousto-optic modulator
(AOM) 448 MHz blue of the main beam, which is near the repump transition.
The trap laser is stabilized using a dichroic-atomic-vapor laser lock (DAVLL)
[181]; this lock provides a wide dispersive signal that allows the laser to be locked
anywhere near the potassium transitions [83]. Beating the trap laser against locking light from the repump laser using a 1 GHz photodetector provides a frequency
reference for setting the DAVLL lock point. Using the DAVLL we can typically
tune the trap laser by 220 MHz in an experiment cycle.
Constructing external cavity diode lasers for the potassium transition (767 nm)
is more difficult than, for example, for the rubidium transition (780 nm). This is
because the readily available diodes in this wavelength range are centered at 780
to 785 nm. Since laser diodes typically decrease their wavelength with cooling
by 1 nm per 5 C, historically an approach to making lasers for our apparatus
was to cool laser diodes. A diode centered at 780 nm at room temperature will
reach 767 nm when cooled to around -40 C; selecting nominally 780 nm diodes

104

CHAPTER 10. THE APPARATUS

repump
ECDL

+ 448 MHz

repump
amplifier

to beat

to coll. MOT

Dn
to sci. MOT
(dark spot)

lock to 39K f=2 line

to sci. MOT
(no dark spot)

trap
ECDL

trap
amplifier

MOT: Dn=556 MHz

CMOT: Dn=570 MHz
+ 242 MHz
off resonance: Dn=487 MHz
imaging: Dn=600 to 700 MHz

to sci. MOT

to coll. MOT
+ 150 to 290 MHz
to optical pumping

to push beam
to low-B imaging
- 2X(65 to 90) MHz

to high-B imaging

Figure 10.4: Laser frequencies used in the experiment for trapping, cooling, and
probing of 40 K . The green boxes are AOMs; the blue lines represent light near
the 40 K f =9/2 f =11/2 transition (trap light); the red lines represent light
near the 40 K f =7/2 f =9/2 transition (repump light). A + sign indicates a
blue shift of the frequency, and a - sign indicates a red shift of the frequency. The
loops depict optical fiber.

105

CHAPTER 10. THE APPARATUS

blue

red

signal (arb)

39

K f=2

-0.2
-0.4
-0.6
-0.8

40

K f = 9/2

40

-1.0
-1.0

K f = 7/2

-0.5

0.5

1.0

D frequency (GHz)
Figure 10.5: Saturated absorption spectroscopy of an enriched source of potassium
atoms. Our lasers are locked to the 39 K f = 2 line, which lies between the two
40
K lines. Data from Ref. [182]. See, for instance, Ref. [183] for a discussion of
diode laser stabilization using saturated absorption spectroscopy.

with anomalously low wavelengths can help make the cooling requirement less
severe [83]. Nonetheless, building and maintaining a stable cold laser requires
considerable time and effort; thus, we have moved towards alternatives. These
alternatives include, for example, buying ECDLs made by New Focus Inc. (Vortex
lasers) designed for 767 nm. The Vortex laser option has the disadvantage that
the frequency noise on these lasers is often larger than JILA-built lasers. Another
alternative is to use diodes that are anti-reflection (AR) coated. AR coating tends
to pull the gain profile of the diode blue as well as allow a much wider tuning
range within an external-cavity configuration. A particularly good source of AR
coated diodes that has emerged recently is Eagleyard Photonics.
The ECDLs currently used in our apparatus each produce 10 mW of laser
power. Thus, for our experiments we amplify the power from the ECDLs to attain
the power we require for the MOTs. For this purpose we use SDL Inc. power
amplifiers (Fig. 10.4). In contrast to nearly everything else in the laser system,
these amplifiers have not been modified or replaced at all in over 5 years. It

CHAPTER 10. THE APPARATUS

106

is astounding that both of these sensitive devices have had such a long lifetime.
Unfortunately, SDL no longer produces power amplifiers, but Eagleyard photonics
now makes similar devices, yet with a somewhat lower gain and perhaps a shorter
lifetime.
The laser setup of Fig. 10.4 has proven to be very robust in its ability to
reliably operate day-to-day. Key features that contribute to this ability include:
The design keeps the number of components to a minimum. There are only two
lasers to maintain and only four AOMs to align and control. Further, for many
of the AOMs and incidental beams in the bottom section of Fig. 10.4, there is
always ample power available. This is a result of a design in which light for various
purposes is siphoned off a main beam by AOMs.

10.3

A stable optical dipole trap

In Ch. 4 we discussed how an optical trap can be created using a gaussian laser
beam far red-detuned of atomic transitions. In this section I outline our current system for creating a stable optical trap and provide tips for building and
stabilizing such a trap.
The most important design parameters of an optical trap are the waist and
the power. The power is an easily tunable parameter, while the waist is most
often fixed during an experiment cycle. Together the waist and power determine
the trap depth and frequencies (Eqns. 4.2 and 4.3). The depth of the trap is
very important for evaporative cooling as it defines the temperature of the gas
during evaporation (evaporative cut); the trap frequencies determine the collision
rate for evaporation as well as the density of the final sample used for BCS-BEC
crossover studies. A smaller waist will result in a higher mean trap frequency
for a given trap depth (Eqn. 4.3); on the other hand, a larger waist will make a
weaker trap for an equivalent depth. The waist size also affects the aspect ratio:
A small waist results in a less extreme aspect ratio (a smaller r /z ). If the waist
is made too large the axial direction will become too weak at deep cuts to hold
atoms against realistic magnetic-field gradients.
In Ch. 4 we discussed only the case of a trap formed by a single gaussian
laser beam, and this is indeed the trap that was used for most of the experiments

CHAPTER 10. THE APPARATUS

107

(a)

(b)

Figure 10.6: Useful options for collimating a 1064 nm beam out of a fiber. (a)
A self-constructed multielement lens (design by M. Greiner). This set contains a
piece of silica (n=1.45), a Melles Griot meniscus lens (LAM111/077), and a Melles
Griot achromat (LAO111/077); it has an effective focal length of 48.2 mm and a
diameter of 18 mm. (b) A pre-assembled multielement lens, the Melles Griot 06
GLC 001, which has an effective focal length of 6.56 mm and a 10 mm diameter.

described in this thesis. However, a single beam trap has the disadvantage that
the aspect ratio is severe and cannot be controlled independently of the trap
strength. Thus, our current system contains a crossed dipole beam configuration
in which a second red-detuned beam perpendicular to the first is overlapped with
the atoms. In our apparatus this beam points along y and thus is referred to as
the vertical beam. To maintain a relatively low density atom gas the waist of
this beam is made significantly larger than the first beam. For example, for the
measurements in Ch. 9 the main beam had a 15 m waist and the vertical beam
a 200 m waist. The technical challenges related to assembling a second beam
are equivalent or simpler than those for the main beam. Thus, in the rest of this
section I will focus solely on constructing our main optical trap beam.
To create our optical dipole trap we start with light from a CW laser at
1064 nm. We purchased a diode-pumped YAG laser made by Spectra Physics
(Spectra Physics part number T20-BL-106C). This laser runs multimode (but
single transverse mode) and can produce greater than 4 W of power. The light

CHAPTER 10. THE APPARATUS

108

from the laser is coupled into a polarization maintaining fiber, which must be
angle cut or AR coated to avoid intensity fluctuations due to cavity formation in
the fiber. The light coming out of the fiber is then collimated to the appropriate
size for creating the desired gaussian waist. For example, to create a 15 m waist
at the atoms using a 200 mm focusing lens the the beam must be collimated to
w=4.5 mm.
Usually fiber output couplers are designed to collimate a beam to a much
smaller size than 4.5 mm. Thus, if a typical fiber output coupler is used a telescope
must be employed to expand the small beam. The optical design of Fig. 10.6(a)
avoids these multiple steps and creates a much more compact setup by directly
collimating the beam out of the fiber to a large size. Care must be taken in
designing such a lens system to maintain low spherical aberration. It is important
to optimize the thickness of the plate of glass that forms the first element of
Fig. 10.6(a). This optimization can be done with a ray-tracing program that can
calculate the geometrical spot size that results from propagation through a set
of optics. A program we use for this purpose is a product of Lambda Research
Corporation, OSLO Light. (OSLO stands for Optics Software for Layout and
Optimization.) This program, in addition to carrying out computations based
upon geometric optics, will perform propagation of a gaussian beam through a
lens system using ABCD matrix analysis. Lens specification files can be imported
directly into the program, and the software also creates drawings such as those in
Fig. 10.6.
Fig. 10.6(b) shows a suggested coupler in the case where the beam must be
collimated to a small size and later expanded with a telescope. Such a design
might be useful if a small optic needs to be placed in the beam after the fiber,
such as an AOM or a polarizer. Using the collimator also creates a slightly more
flexible system because the size of the beam, and hence the waist, can be changed
by altering just one lens in the telescope, instead of replacing and reoptimizing
the collimating lens.
Following assembly of the optical components it is important to check for
aberrations and astigmatism (different focal positions in x and y) in the final
focus. A useful tool for this is a beam-profiling camera; we have found that the
WinCamD beam-profiling camera manufactured by DataRay Inc. works well. Fig.

109

CHAPTER 10. THE APPARATUS

2w (mm)

150
wy=18.8(2) mm
cy=3.73(2) mm

x
y

100

wx=19.3(2) mm
cx=3.73(3) mm

50
0

4
z (mm)

Figure 10.7: Measured optical trap axial profile. The beam sizes in the x and y
planes are measured as a function of z using a beam profiling camera. wx and wy
are the fitted gaussian
waist sizes and cx and cy the center positions. The fitting
q
(zc) 2
function is 2w 1 + ( w2 ) .
10.7 shows the result of measuring the beam size on this camera as a function
of the position along the axis of the beam. Good signs are that the profile is
symmetric about the minimum of w, the points fit well to the expected form for
a gaussian focus for 1064 nm light, and the two directions (x and y) focus at the
same position.
After creating a suitable gaussian focus, the next concern is the stability of
the optical trap, as fluctuation in optical trap parameters can cause heating of
the atom gas. Noise on the laser intensity leads to fluctuations in the trap spring
constant, while noise in the pointing of the gaussian beam leads to fluctuations
in the trap center. Thus, intensity noise leads to heating when the fluctuations
are at twice the trap frequency, while pointing noise leads to heating when the
fluctuations are at the trap frequency. A theoretical analysis of the effect of
intensity and pointing noise on optically trapped atoms is presented in Ref. [184].
We minimize intensity fluctuations by servoing the optical trap power. Our
servo uses feedback from a pickoff of the trapping light power after the optical
fiber, and an AOM before the fiber controls the power. To achieve good pointing
stability of the optical trap we do not actively stabilize the beam position, but

CHAPTER 10. THE APPARATUS

110

rather rely on mounting the optical trap optics in a way that minimizes vibrations.
When we first started building optical traps, we did not pay particular attention
to the stability of the mounted optics. In this case we found that the heating rate
of the atoms in the trap was severely limited by the stability of the optics. Thus,
in subsequent configurations the optics have been mounted on thick posts, which
we sometimes fill with lead shot to damp vibrations. Moving parts, such as mirror
mounts or lens translators, are minimized. Following continual improvement to
the stability of the optics the heating rate of our trap is now 5 nK/sec.
The last step is to align the optical trap focus on the atoms. In our apparatus
we must send the beam through the science cell along an optical path already
occupied by a MOT beam. Useful methods of combining the MOT light (767
nm) and the optical trapping light (1064 nm) include using motorized mirrors
that can be triggered with a TTL signal during the course of the experiment and
dichroic plates that reflect 767 nm light, yet transmit 1064 nm light. (CVI Laser
manufactures dichroic plates in a variety of sizes with better than 99% reflectance
at 767 nm and 95% transmission at 1064 nm.) The optical trap beam should be
aligned at a few degrees with respect to the cell normal; this avoids interference
patterns caused by back-reflections from the cell wall. Precisely aligning the
optical trap beam onto the atoms is not a trivial task since the beam must be
placed with 10 m precision to trap the atoms. The best way to roughly align
the beam is to image both the atoms and the optical trap beam onto the same
position on a CCD camera. (One must be careful to account for any chromatic
shifts, however.) This brings the beam close enough to trap a fraction of the
atoms. We determined the final beam alignment by optimizing the number of
atoms in the optical trap after evaporation.
To evaporate the atoms in the optical trap the power at the atoms is controlled
with the AOM intensity servo described above. The power during a typical cycle
of the experiment is shown in Fig. 10.8. The trap is ramped on in 100 ms while
the magnetic trap is still on. Then the magnetic trap is switched off suddenly and
the optical trap evaporation is immediately initiated.2 At the end of the sequence
the gas is expanded for time-of-flight imaging by turning off the optical trap power
2

Switching off the magnetic trap suddenly will not result in a perfectly adiabatic transfer.
However, for the hot gases we load into the trap the effect is negligible.

111

CHAPTER 10. THE APPARATUS

500

Power (mW)

400
300
200
final evaporative cut

100
magnetic trap off

0
0

time (sec)
Figure 10.8: A typical optical trap evaporation trajectory to a modest final optical
trap cut. The functional form of the decrease in intensity is a series of exponentials
with changing time constants.

CHAPTER 10. THE APPARATUS

112

Figure 10.9: False color absorption image of two molecular condensates in a

double-well potential created with a spatially-modulated optical trap. The gases
are different sizes because there is an imbalance in the molecule number in the
two wells.

very quickly (< 1 s) using an AOM as a switch.

An additional feature of the current version of our optical trap is the placement
of two crossed AOMs in the beam after the optical fiber. These AOMs can deflect
the optical trap beam in x and y. The position of the optical trap can thus be
arbitrarily modulated at rates up to 1 MHz. Since the trap can be modulated on
time scales much faster than the trap oscillator frequencies (at most a few kHz), a
time-averaged potential can be created. This idea is similar to the time-orbitingpotential (TOP) magnetic trap [124] and was suggested by M. Greiner. With
this design we succeeded in creating a double-well potential; Fig. 10.9 shows two
molecular condensates that were creating in such a double-well potential. They
are imaged by kicking the two condensates velocity in opposite directions just
before time-of-flight expansion. A spatially-modulated optical trap could perhaps
also be used to create a quartic potential or a rotating elongated trap for vortex
creation.

10.4

Large magnetic-field control

For accessing Feshbach resonances we needed to a create a stable, homogeneous

magnetic field at the atoms of up to 250 G. To do this we modified the control
circuitry for the cloverleaf Ioffe-Pritchard magnetic trap (Fig. 10.3) to allow the

113

CHAPTER 10. THE APPARATUS

F1

(+)

H5

F5
center and
petal coils

power supply
(-)

H2

H4

F4

F3

bias
coils

F2

Figure 10.10: Schematic diagram of the circuit used to control the current through
the coils of the Ioffe-Pritchard trap (Fig. 10.3). The coils can either be used to
form a magnetic trap [83] or to provide a flat bias field for Feshbach resonance
studies. The brown lines indicate long cables that contribute some resistance.

CHAPTER 10. THE APPARATUS

114

coils to be used not only as a trap, but also as a bias magnetic field. A schematic
of our control over the coils is shown in Fig. 10.10. We can manipulate the current
in the bias coils independently of the current through the petal and center coils.
We start with a HP 6628A power supply operating in constant voltage mode.
The current is monitored with F. W. Bell Hall effect current sensors (H in Fig.
10.10) placed in various locations. A low noise servo of these currents is built
using high power FETs to control the current (F in Fig. 10.10). Each of the
five FETs shown in Fig. 10.10 consists of a pair of water cooled, Advanced Power
Technology MOSFETs in parallel. F1 and F3 are used only as switches for fast
turn-on or turn-off of the current. F2, F4, and F5 are controlled via servos.
For operation of the magnetic trap F5 is switched off, F4 servos the full current
(measured with H4), and F2 maintains the required current through the bias coil
(measured with H2). For creating a bias field, F4 is switched on, F5 is allowed
to pass current, and F2 again maintains the current through the bias coils. To
make the magnetic field as homogeneous as possible across the axial direction of
our trapped gas (see Fig. 10.15), we route a small amount of current through
the center and petal coils to cancel the axial magnetic-field curvature created by
the bias coils; this current is servoed with F5. Lastly, a small current through
the science MOT quadrupole coils cancels out the residual axial magnetic field
gradient (due to differences in the precise centers of the bias and center coil fields).
A key aspect of the magnetic-field control in our apparatus is its low noise
and drift. To attain this we use PI servos with low-noise control voltages (see
Ref. [83]). There is a designated ground for these magnetic-field servos to avoid
60 Hz noise from ground loops. The two most important current sensors, H2 and
H4, are temperature stabilized and never used to monitor the performance of the
servo (as monitoring the hall probe output on an oscilloscope can lead to ground
loops and small shifts). At the position of H2 and H4 are additional monitor
sensors, solely for monitoring the resulting current in the coils.
To test the magnetic-field stability that we achieve we probe the magnetic
field with the atoms by driving rf transitions. Fig. 10.11 shows the result of
such a measurement using optically trapped atoms at a bias field of 240 G.
Atoms were transferred from the mf =7/2 spin state to the mf =5/2 spin state
using a square rf pulse with a duration of tpulse =240 s. A long rf pulse (and

115

fraction mf = -5/2

CHAPTER 10. THE APPARATUS

1.0
0.8
0.6
0.4
0.2
0
53.345

2/tpulse

53.350

53.355

53.360

n (MHz)
Figure 10.11: Transfer of atoms at 240 G between the mf =7/2 and mf =5/2
spin states as a function of rf frequency. The data are fit to the function for fourier
transform of square pulse, A [Sinc(tpulse ( 0 ))]2 . tpulse is fixed at 240 s, and
A and 0 are allowed to vary.

hence a narrow Fourier width) allowed us to sensitively measure the magneticfield stability. We took multiple points at a frequency on the side of the line and
measured the standard deviation. The points in Fig. 10.11 were taken over an
hour and each error bar represents the standard deviation of over 10 points. This
standard deviation corresponds to a 1 kHz fluctuation in the resonance frequency,
which corresponds to a magnetic-field stability of 7 mG.

10.5

Fast magnetic-field control

For many of the BCS-BEC crossover experiments we performed it was essential

to be able to change the magnetic field very quickly. The large coils we use for
the 200 G bias field are not well suited to this purpose since they have a large
inductance and the servos that are used to control them were designed with a
bandwidth of 3 kHz. Thus, for magnetic-field ramps with inverse ramp rates of
1 s/G to 100 s/G we used low-inductance auxiliary coils and a high-bandwidth
servo to ramp the magnetic field. A circuit diagram for this system is shown in
Fig. 10.12. A Hall effect current sensor monitors the current and a Powerex IBGT
controls the current through the circuit. The current is provided by discharging a

116

CHAPTER 10. THE APPARATUS

5 kW

current sensor

0.5 W

100 V
+

coil 1

100 kW

coil 2

6.8 mF

LED
G

IGBT
E

Figure 10.12: Circuit used to created fast magnetic-field pulses. The two coils
shown are placed on opposite sides of the 1 inch wide cell holding the atoms, are
0.9 inches in diameter, have 5 turns each, and are wound from 24 AWG wire.

capacitor that is charged to high voltage by a HP 6207B power supply [185]. Two
coils, each mounted along the inner diameter of one of the center coils, produce
the magnetic field at the atoms. The direction of this field can be quickly switched
using the H-bridge configuration shown in Fig. 10.12. This allows us to jump the
magnetic field either above or below the Feshbach resonance.
Fig. 10.13 shows the magnetic field produced with this circuit as determined
by observing the current through the auxiliary coils as a function of time for
various control voltage ramps. In the blue, red, and black curves the control
voltage ramp occurs in 7 s, 20 s, and 40 s, respectively. According to this
plot, we find that we can ramp the magnetic field controllably with inverse rates
less than 1 s/G. Current induced in other coils in the system may reduce the
true rate of the magnetic-field ramp.

10.6

Rf delivery

For the spin transfer to achieve a mixture of atoms in the Feshbach resonance
states, we required a rf delivery system that could produce high power rf in two

117

CHAPTER 10. THE APPARATUS

20

DB (gauss)

15
10
5
0
-40

-20

0
time (ms)

20

40

Figure 10.13: Magnetic field as determined through measuring the current through
the current sensor in Fig. 10.12 as a function of time. For the three different curves
the servo control voltage is changed at three different rates.

distinct frequency ranges. The transfer from the mf =+9/2, +7/2 spin states to
the mf =9/2, 7/2 at 20 G requires rf at 6 MHz, and the transfer amongst the
mf =5/2, 7/2 and 9/2 states at magnetic fields near the Feshbach resonances
requires rf in the 40 to 50 MHz range. The rf system we use consists of a 0-80
MHz function generator (Agilent 33250A), a high-power rf amplifier, and a 0.75
inch diameter coil consisting of 8 turns of wire located about 1.8 inches from the
atoms.
The key to delivering rf at the required frequencies was to properly impedance
match the rf coil to the 50 impedance of the transmission line. To deliver the
most power to the atoms at a specific frequency it is best to purely reactively
impedance match, i.e., not to add any resistance to the circuit. A system suitable
for reactively impedance matching for our requirements is shown in Fig. 10.14
[186]. For the 40 to 50 MHz frequency range and our coil inductance, the required
value of L is a few H and the value of C is a few pF. The advantage of this circuit
is that since the capacitor is in parallel with the rf coil the dc response of the coil
is maintained. This allows efficient delivery of rf at 6 MHz.
While the coil can deliver enough power throughout the 40 to 50 MHz range it
is of course not perfectly impedance matched throughout this range. (Even at the
optimum frequency only about 75% of the power is delivered to the coil.) Thus,

118

CHAPTER 10. THE APPARATUS

line Z0

coil

Figure 10.14: The circuit for the rf coil used to drive spin-changing transitions in
the optical trap.

invariably some power will be reflected back from the coil; this has the capability
of damaging the amplifier. To avoid this problem we now use a class A amplifier
made by ENI, the ENI525LA. The final result of our rf system is that we are able
to apply a pulse between any of the mf =5/2, 7/2, 9/2 states in 15 s.

10.7

Absorption imaging

Nearly all of the information that we extract from our experiment comes from absorption imaging on the f =9/2 f =11/2 cycling transition in 40 K. The concept
of absorption imaging was discussed in Ch. 4, and detailed derivations related
to absorption imaging can be found in, for example, Refs. [124, 187]. In this
section I will concentrate on the specific configurations that we use for absorption
imaging 40 K gases to extract the information needed for our BCS-BEC crossover
experiments. Especially important for our crossover studies were the abilities to
image spin selectively and at high magnetic field.
Fig. 10.15 shows the basic geometry for absorption imaging in our apparatus.
The black oval represents the atom gas within the science cell and the two red
arrows show the two possible resonant imaging beams. Each imaging configuration
has a designated Princeton Instruments back-illuminated CCD camera, the CCD512TKB along x and the CCD-512EBFT along z.
In our first imaging option the resonant beam propagates along x and thus
provides information about the axial and radial directions of the trapped gas. This
imaging beam is perpendicular to the large bias magnetic field. Thus, imaging

119

CHAPTER 10. THE APPARATUS

large
magnetic
field

to z^ camera
^

to x camera

Figure 10.15: Imaging configurations. Gravity points into the page, and the large
bias field and the axis of the optical trap both point along z. The 40 K gas (black
oval) can be imaged along either z where there is a 1 inch window or x where
there is a 1/2 inch window. Diagram not to scale.

at high magnetic field with this beam is not optimum since the quantization is
90 off. Most imaging along this direction takes place at low magnetic field where
a small (4 G) magnetic field can be applied along x for the quantization. At
low magnetic field the imaging light does not distinguish between different spin
states. Thus, if we need to image the spin states separately a large magnetic field
gradient (80 G/cm [83]) is applied along y to separate the spin states spatially
(Stern-Gerlach imaging). Since we have the capability to switch the direction of
the small quantization field along this axis, this setup can image atoms in either
the + or - spin states of the f =9/2 manifold.
In the second imaging option the resonant beam propagates along z where
it is combined with the optical trap beam using dichroic plates. This setup
images both radial directions of the trapped gas and is optimized for imaging
the mf =5/2, 7/2, 9/2 spin states at high magnetic field near Feshbach resonances. The quantization is provided by the large bias magnetic field (Sec. 10.4),
and the atoms can easily be imaged spin selectively since the transitions for the
different spin states are separated by 30 MHz near 200 G due to the nonlinear
Zeeman shift. We can collect an image for each of two spin states per experiment

120

CHAPTER 10. THE APPARATUS

cycle using the kinetics mode and frame transfer capabilities of our CCD camera.
It is important that the optics used for imaging the atom gas are designed
correctly to provide sufficient imaging resolution. The optics used to image the
gas along the z direction are shown in Fig. 10.16. The blue lines roughly represent
the image of the atom gas, which is collimated by the first set of lenses and focused
onto the camera with the second. We expect the diffraction limit to be defined by
an Airy disc pattern (see Fig. 10.17 red line) the diameter of which is 1.22/NA,
where NA= d/2f , d is the lens diameter, and f the lens focal length. Diffraction
limited performance of the first lens set in Fig. 10.16 at 767 nm should result in
an Airy disc diameter (spot size) of 4.5 m.

30 mm

f = 71.5 mm

f = 300.6 mm

Figure 10.16: Optics used to image the atom gas along the z direction. The atoms
are to the left and the CCD camera to the right. The meniscus and achromat lens
pair provide diffraction limited performance.

To test whether this imaging system could provide diffraction limited performance, we imaged a 1 m pinhole onto the WinCamD beam profiling camera
using this set of lenses. Using a gaussian fit of the resulting spot on the camera,
and accounting for the magnification and the difference in test wavelength and 767
nm, we found that the full-width-half-max (fwhm) of the spot was 2.1 0.3 m.
Converting this result to an Airy disc diameter (Fig. 10.17), we found 5.1 0.6
m, which is consistent with diffraction limited performance. While we cannot
create a gas small enough to fully test the resolution limit of this system in the
actual experiment, measurements in Ref. [155] are consistent with a spot size of
5 m.

121

CHAPTER 10. THE APPARATUS

The imaging optics for the path along x are similar to those in Fig. 10.16,
with the differences being the first lens is an achromat (instead of an achromat/meniscus lens pair) with a 140 mm focal length and the second lens has a
500 mm focal length. The expected diffraction limited performance of this imaging system is a spot size diameter of 8.7 m (fwhm of 3.7 m). It is also possible
to place two more lenses between the f=500 mm lens and the camera to make
a low-magnification imaging system for probing the MOT or hot gases in the
magnetic trap [83].

1.0
0.8

amp

0.6

-2r2/w2

0.4

hwhm

0.2

0
0

(2J1(r)/r)2
0.50

1.00

1.50

r / Airy disc radius

Figure 10.17: Gaussian distribution compared to an Airy disc pattern; a gaussian
was fit to the Bessel function that describes the Airy disc pattern. Numbers used
to characterize the radius of gaussian distributions include the rms size , the halfwidth-half-max (hwhm), and the waist (w). These are related through w=2=1.7
hwhm. The hwhm of the Airy disc and the hwhm of the gaussian occur at nearly
the same position, 0.42 and 0.41 times the Airy disc radius, respectively.

Chapter 11
Conclusions and future directions
11.1

Conclusion

In this thesis I have presented the story of the realization of BCS-BEC crossover
physics with a gas of 40 K atoms. Experiments with 40 K and 6 Li have shown
that a Fermi gas at a broad Feshbach resonance crosses a phase transition to a
superfluid state and displays properties of the classic BCS-BEC crossover problem.
With this system we can study the evolution from fermionic superfluidity of pairs
nearly described by BCS theory to BEC of diatomic molecules. This new system
provides a physical link between two descriptions of superfluid systems, BCS and
BEC, that were historically thought to be distinct.
The fermionic superfluids created in these gases have extremely high transition temperatures, Tc 0.1 TF . In these model systems the absolute value
of Tc 100 nK is very cold. However, for typical values of TF in metals the
corresponding transition temperature would be above room temperature. Fully
understanding this model system will perhaps contribute to the efforts to create
higher transition temperature superconductors in real materials. These atomic
gas systems also provide the opportunity to study aspects of quantum systems
that are not typically accessible in solid state materials. For example, experiments
such as those in Ch. 8 utilize dynamics for measurements. The ability to study
the physics of the BCS-BEC crossover in real time may provide new insight into
this many-body quantum system.

122

CHAPTER 11. CONCLUSIONS AND FUTURE DIRECTIONS

11.2

123

Future work

In the immediate future there are a plethora of experiments that can be done to
study BCS-BEC crossover physics with this new system. This list includes both
more precise versions of previous measurements as well as experiments designed
to study entirely new phenomena. In our 40 K experiments there are a number
of possible direct extensions of measurements discussed in this thesis. For example, a better measurement of the Feshbach resonance position B0 than that of
Fig. 6.9 would produce more precise values of 1/kF a. To make a more precise
measurement of B0 one could use the magnetic field modulation technique for
dissociating molecules to measure Eb of a low density gas. The phase diagram of
the BCS-BEC crossover measured through N0 /N could be improved with better
temperature measurements and better understanding of the systematics on N0 /N .
With more precise measurements of experimental observables in the crossover, we
will be able to test more sophisticated crossover theories.
Possible extensions of this study of BCS-BEC crossover physics are also numerous. Studies of higher partial wave pairing in an atomic system [57, 164, 188]
would be an important contribution due to the relevance to cuprate superconductors, which have been shown to have d-wave symmetry. Predictions have been
made for the behavior of an atomic Fermi gas in the BCS-BEC crossover with
unequal particle number in each spin state of the two-component gas [189, 190].
Such a magnetized system has been the topic of a number of recent experimental pursuits [191, 192]. Further, quantum Fermi gases have now been studied in
optical lattice potentials [193, 40]; studying crossover physics in the presence of
such a lattice could more closely mimic conditions in real solids.

Bibliography
[1] H. K. Onnes, On the sudden rate at which the resistance of mercury disappears, Akad. van Wetenschappen 14(113), 818 (1911).
[2] J. F. Allen and A. D. Misener, Flow of liquid helium II, Nature 141, 75
(1938).
[3] P. Kapitza, Viscosity of liquid helium at temperatures below the lambda
point, Nature 141, 74 (1938).
[4] D. D. Osheroff, R. C. Richardson, and D. M. Lee, Evidence for a new phase
of solid 3 He, Phys. Rev. Lett. 28, 885 (1972).
[5] J. G. Bednorz and K. Mueller, Possible high-Tc superconductivity in the
Ba-La-Cu-O system, Z. Physik B64, 189 (1986).
[6] D. W. Snoke and G. Baym, in Bose-Einstein Condensation, edited by A.
Griffin, D. W. Snoke, and S. Stringari (Cambridge Univ., Cambridge, UK,
1995), pp. 111.
[7] M. Holland, S. J. J. M. F. Kokkelmans, M. L. Chiofalo, and R. Walser,
Resonance superfluidity in a quantum degenerate Fermi gas, Phys. Rev.
Lett. 87, 120406 (2001).
[8] M. Randeria, in Bose-Einstein Condensation, edited by A. Griffin, D. W.
Snoke, and S. Stringari (Cambridge Univ., Cambridge, UK, 1995), pp. 355
392.
[9] S. Bose, Plancks Gesetz und Lichtquantenhypothese, Z. Phys. 26, 178
(1924).
[10] A. Einstein, Quantentheorie des einatomigen idealen Gases. Zweite Abhandlung, Sitzungsber. Preuss. Akad. Wiss. 1925, 3 (1925).
[11] For a nice pedagogical description of BEC see Ch. 1 of J. F. Annett, Superconductivity, Superfluids, and Condensates (Oxford Univ. Press, Oxford,
UK, 2004).
124

BIBLIOGRAPHY

125

[12] F. London, On the Bose-Einstein condensation, Phys. Rev. 54, 947 (1938).
[13] J. M. Blatt and S. T. Butler, Superfluidity of a boson gas, Phys. Rev. 96,
1149 (1954).
[14] P. E. Sokol, in Bose-Einstein Condensation, edited by A. Griffin, D. W.
Snoke, and S. Stringari (Cambridge Univ., Cambridge, UK, 1995), pp. 51
85.
[15] J. Bardeen, L. N. Cooper, and J. R. Schrieffer, Microscopic theory of superconductivity, Phys. Rev. 106, 162 (1957).
[16] J. Bardeen, L. N. Cooper, and J. R. Schrieffer, Theory of superconductivity,
Phys. Rev. 108, 1175 (1957).
[17] L. N. Cooper, Bound electron pairs in a degenerate Fermi gas, Phys. Rev.
104, 1189 (1956).
[18] Q. Chen, J. Stajic, S. Tan, and K. Levin, BCS-BEC crossover: From high
temperature superconductors to ultracold superfluids, Phys. Rep. 412, 1
(2005).
[19] W. D. Philllips, Laser cooling and trapping of neutral atoms, Rev. Mod.
Phys. 70, 721 (1998).
[20] W. Ketterle, D. S. Durfee, and D. M. Stamper-Kurn, in Proceedings of the
International School of Physics - Enrico Fermi, edited by M. Inguscio, S.
Stringari, and C. E. Wieman (IOS Press, Amsterdam, 1999), p. 67.
[21] E. A. Cornell, J. R. Ensher, and C. E. Wieman, in Proceedings of the International School of Physics - Enrico Fermi, edited by M. Inguscio, S.
Stringari, and C. E. Wieman (IOS Press, Amsterdam, 1999), p. 15.
[22] M. H. Anderson, J. R. Ensher, M. R. Matthews, C. E. Wieman, and E. A.
Cornell, Observation of Bose-Einstein condensation in a dilute atomic vapor,
Science 269, 198 (1995).
[23] K. B. Davis, M.-O. Mewes, M. R. Andrews, N. J. van Druten, D. S. Durfee,
D. M. Kurn, and W. Ketterle, Bose-Einstein condensation in a gas of sodium
atoms, Phys. Rev. Lett. 75, 3969 (1995).
[24] M. R. Andrews, C. G. Townsend, H.-J. Miesner, D. S. Durfee, D. M. Kurn,
and W. Ketterle, Observation of interference between two Bose-Einstein
condensates, Science 275, 637 (1997).

BIBLIOGRAPHY

126

[25] M. R. Matthews, B. P. Anderson, P. C. Haljan, D. S. Hall, C. E. Wieman,

and E. A. Cornell, Vortices in a Bose-Einstein Condensate, Phys. Rev. Lett.
83, 2498 (1999).
[26] C. Raman, M. Kohl, R. Onofrio, D. S. Durfee, C. E. Kuklewicz, Z. Hadzibabic, and W. Ketterle, Evidence for a critical velocity in a Bose-Einstein
condensed gas, Phys. Rev. Lett. 83, 2502 (1999).
[27] F. Dalfovo, S. Giorgini, L. P. Pitaevskii, and S. Stringari, Theory of BoseEinstein condensation in trapped gases, Rev. Mod. Phys. 71, 463 (1999).
[28] D. R. Tilley, Superfluidity and Superconductivity (Adam Hilger, Ltd., Accord, MA, 1986).
[29] S. Inouye, M. R. Andrews, J. Stenger, H.-J. Miesner, D. M. Stamper-Kurn,
and W. Ketterle, Observation of Feshbach resonances in a Bose-Einstein
condensate, Nature 392, 151 (1998).
[30] J. L. Roberts, N. R. Claussen, J. P. Burke, Jr., C. H. Greene, E. A. Cornell,
and C. E. Wieman, Resonant magnetic field control of elastic scattering of
cold 85 Rb, Phys. Rev. Lett. 81, 5109 (1998).
[31] M. Greiner, O. Mandel, T. Esslinger, T. W. Hansch, and I. Bloch, Quantum
phase transition from a superfluid to a Mott insulator in a gas of ultracold
atoms, Nature 415, 39 (2002).
[32] B. DeMarco and D. S. Jin, Onset of Fermi degeneracy in a trapped atomic
gas, Science 285, 1703 (1999).
[33] A. G. Truscott, K. E. Strecker, W. I. McAlexander, G. B. Partridge, and
R. G. Hulet, Observation of Fermi pressure in a gas of trapped atoms,
Science 291, 2570 (2001).
[34] F. Schreck, L. Khaykovich, K. L. Corwin, G. Ferrari, T. Bourdel, J. Cubizolles, and C. Salomon, Quasipure Bose-Einstein condensate immersed in a
Fermi sea, Phys. Rev. Lett. 87, 080403 (2001).
[35] G. Roati, F. Riboli, G. Modugno, and M. Inguscio, Fermi-Bose quantum
degenerate 40 K-87 Rb mixture with attractive interaction, Phys. Rev. Lett.
89, 1804 (2002).
[36] S. R. Granade, M. E. Gehm, K. M. OHara, and J. E. Thomas, All-optical
production of a degenerate Fermi gas, Phys. Rev. Lett. 88, 120405 (2002).

BIBLIOGRAPHY

127

[37] Z. Hadzibabic, C. A. Stan, K. Dieckmann, S. Gupta, M. W. Zwierlein, A.

Gorlitz, and W. Ketterle, Two species mixture of quantum degenerate Bose
and Fermi gases, Phys. Rev. Lett. 88, 160401 (2002).
[38] J. Goldwin, S. Inouye, M. Olsen, B. Newman, B. D. DePaola, and D. S.
Jin, Measurement of the interaction strength in a Bose-Fermi mixture with
87
Rb and 40 K, Phys. Rev. A 70, 021601 (2004).
[39] M. Bartenstein, A. Altmeyer, S. Riedl, S. Jochim, C. Chin, J. Hecker Denschlag, and R. Grimm, Crossover from a molecular Bose-Einstein condensate
to a degenerate Fermi gas, Phys. Rev. Lett. 92, 120401 (2004).
[40] M. Kohl, H. Moritz, T. Stoferle, K. G
unter, and T. Esslinger, Fermionic
atoms in a 3D optical lattice: Observing Fermi-surface, dynamics, and interactions, Phys. Rev. Lett. 94, 080403 (2005).
[41] C. Ospelkaus, S. Ospelkaus, K. Sengstock, and K. Bongs, Interaction-driven
dynamics of 40 K / 87 Rb Fermi-Bose gas mixtures in the large particle number
limit, cond-mat/0507219 (2005).
[42] H. T. C. Stoof, M. Houbiers, C. A. Sackett, and R. G. Hulet, Superfluidity
of spin-polarized 6 Li, Phys. Rev. Lett. 76, 10 (1996).
[43] R. Combescot, Trapped 6 Li: A high Tc superfluid?, Phys. Rev. Lett. 83,
3766 (1999).
[44] E. Timmermans, K. Furuya, P. W. Milonni, and A. K. Kerman, Prospect of
creating a composite Fermi-Bose superfluid, Phys. Lett. A 285, 228 (2001).
[45] Y. Ohashi and A. Griffin, BCS-BEC crossover in a gas of Fermi atoms with
a Feshbach resonance, Phys. Rev. Lett. 89, 130402 (2002).
[46] D. M. Eagles, Possible pairing without superconductivity at low carrier
concentrations in bulk and thin-film superconducting semiconductors, Phys.
Rev. 186, 456 (1969).
[47] A. J. Leggett, Cooper pairing in spin-polarized Fermi systems, J. Phys. C.
(Paris) 41, 7 (1980).
[48] P. Nozieres and S. Schmitt-Rink, Bose condensation in an attractive fermion
gas: From weak to strong coupling superconductivity, J. Low-Temp. Phys.
59, 195 (1985).
[49] M. Randeria, J. M. Duan, and L. Y. Shieh, Superconductivity in twodimensional Fermi gas: Evolution from Cooper pairing to Bose condensation, Phys. Rev. B 41, 327 (1990).

128

BIBLIOGRAPHY

[50] M. Dreschler and W. Zwerger, Crossover from BCS-superconductivity to

Bose-condensation, Ann. Physik 1, 15 (1992).
[51] R. Haussman, Properties of Fermi liquid at the superfluid transition in the
crossover region between BCS superconductivity and Bose-Einstein condensation, Phys. Rev. B 49, 12975 (1994).
[52] M. Houbiers, H. T. C. Stoof, W. I. McAlexander, and R. G. Hulet, Elastic
and inelastic collisions of Li-6 atoms in magnetic and optical traps, Phys.
Rev. A 57, R1497 (1998).
[53] J. L. Bohn, Cooper pairing in ultracold
Phys. Rev. A 61, 053409 (2000).

40

K using Feshbach resonances,

[54] T. Loftus, C. A. Regal, C. Ticknor, J. L. Bohn, and D. S. Jin, Resonant

control of elastic collisions in an optically trapped Fermi gas of atoms, Phys.
Rev. Lett. 88, 173201 (2002).
[55] K. M. OHara, S. L. Hemmer, S. R. Granade, M. E. Gehm, J. E. Thomas,
V. Venturi, E. Tiesinga, and C. J. Williams, Measurement of the zero crossing in a Feshbach resonance of fermionic 6 Li, Phys. Rev. A 66, 041401(R)
(2002).
[56] K. Dieckmann, C. A. Stan, S. Gupta, Z. Hadzibabic, C. H. Schunck, and
W. Ketterle, Decay of an ultracold fermionic lithium gas near a Feshbach
resonance, Phys. Rev. Lett. 89, 203201 (2002).
[57] C. A. Regal, C. Ticknor, J. L. Bohn, and D. S. Jin, Tuning p-wave interactions in an ultracold Fermi gas of atoms, Phys. Rev. Lett. 90, 053201
(2003).
[58] K. M. OHara, S. L. Hemmer, M. E. Gehm, S. R. Granade, and J. E.
Thomas, Observation of a strongly interacting degenerate Fermi gas of
atoms, Science 298, 2179 (2002).
[59] C. A. Regal and D. S. Jin, Measurement of positive and negative scattering
lengths in a Fermi gas of atoms, Phys. Rev. Lett. 90, 230404 (2003).
[60] S. Gupta, Z. Hadzibabic, M. W. Zwierlein, C. A. Stan, K. Dieckmann, C. H.
Schunck, E. G. M. van Kempen, B. J. Verhaar, and W. Ketterle, Radiofrequency spectroscopy of ultracold fermions, Science 300, 1723 (2003).
[61] M. E. Gehm, S. L. Hemmer, K. M. OHara, and J. E. Thomas, Unitaritylimited elastic collision rate in a harmonically trapped Fermi gas, Phys. Rev.
A 68, 011603(R) (2003).

BIBLIOGRAPHY

129

[62] T. Bourdel, J. Cubizolles, L. Khaykovich, K. M. F. Magalhaes, S. J. J. M. F.

Kokkelmans, G. V. Shlyapnikov, and C. Salomon, Measurement of the interaction energy near a Feshbach resonance in a 6 Li Fermi gas, Phys. Rev.
Lett. 91, 020402 (2003).
[63] C. A. Regal, C. Ticknor, J. L. Bohn, and D. S. Jin, Creation of ultracold
molecules from a Fermi gas of atoms, Nature 424, 47 (2003).
[64] K. E. Strecker, G. B. Partridge, and R. G. Hulet, Conversion of an atomic
Fermi gas to a long-lived molecular Bose gas, Phys. Rev. Lett. 91, 080406
(2003).
[65] J. Cubizolles, T. Bourdel, S. J. J. M. F. Kokkelmans, G. V. Shlyapnikov,
and C. Salomon, Production of long-lived ultracold Li2 molecules from a
Fermi gas of atoms, Phys. Rev. Lett 91, 240401 (2003).
[66] S. Jochim, M. Bartenstein, A. Altmeyer, G. Hendl, C. Chin, J. Hecker
Denschlag, and R. Grimm, Pure gas of optically trapped molecules created
from fermionic atoms, Phys. Rev. Lett. 91, 240402 (2003).
[67] C. A. Regal, M. Greiner, and D. S. Jin, Lifetime of molecule-atom mixtures
near a Feshbach resonance in 40 K, Phys. Rev. Lett. 92, 083201 (2004).
[68] M. Greiner, C. A. Regal, and D. S. Jin, Emergence of a molecular BoseEinstein condensate from a Fermi gas, Nature 426, 537 (2003).
[69] S. Jochim, M. Bartenstein, A. Altmeyer, G. Hendl, S. Riedl, C. Chin, J.
Hecker Denschlag, and R. Grimm, Bose-Einstein condensation of molecules,
Science 302, 2101 (2003).
[70] M. W. Zwierlein, C. A. Stan, C. H. Schunck, S. M. F. Raupach, S. Gupta,
Z. Hadzibabic, and W. Ketterle, Observation of Bose-Einstein condensation
of molecules, Phys. Rev. Lett. 91, 250401 (2003).
[71] T. Bourdel, L. Khaykovich, J. Cubizolles, J. Zhang, F. Chevy, M. Teichmann, L. Tarruell, S. Kokkelmans, and C. Salomon, Experimental study of
the BEC-BCS crossover region in 6 Li, Phys. Rev. Lett. 93, 050401 (2004).
[72] G. B. Partridge, K. E. Strecker, R. I. Kamar, M. W. Jack, and R. G. Hulet,
Molecular probe of pairing in the BEC-BCS crossover, Phys. Rev. Lett. 95,
020404 (2005).
[73] C. A. Regal, M. Greiner, and D. S. Jin, Observation of resonance condensation of fermionic atom pairs, Phys. Rev. Lett. 92, 040403 (2004).

BIBLIOGRAPHY

130

[74] M. W. Zwierlein, C. A. Stan, C. H. Schunck, S. M. F. Raupach, A. J.

Kerman, and W. Ketterle, Condensation of pairs of fermionic atoms near a
Feshbach resonance, Phys. Rev. Lett. 92, 120403 (2004).
[75] M. Bartenstein, A. Altmeyer, S. Riedl, S. Jochim, C. Chin, J. Hecker Denschlag, and R. Grimm, Collective excitations of a degenerate gas at the
BEC-BCS crossover, Phys. Rev. Lett. 92, 203201 (2004).
[76] J. Kinast, S. L. Hemmer, M. E. Gehm, A. Turlapov, and J. E. Thomas,
Evidence for superfluidity in a resonantly interacting Fermi gas, Phys. Rev.
Lett. 92, 150402 (2004).
[77] J. Kinast, A. Turlapov, and J. E. Thomas, Breakdown of hydrodynamics in
the radial breathing mode of a strongly interacting Fermi gas, Phys. Rev. A
70, 051401 (2004).
[78] C. A. Regal, M. Greiner, S. Giorgini, M. Holland, and D. S. Jin, Momentum
distribution of a Fermi gas of atoms in the BCS-BEC crossover, Phys. Rev.
Lett. 95, 250404 (2005).
[79] J. Kinast, A. Turlapov, J. E. Thomas, Q. Chen, J. Stajic, and K. Levin,
Heat capacity of a strongly interacting Fermi gas, Science 307, 1296 (2005).
[80] C. Chin, M. Bartenstein, A. Altmeyer, S. Riedl, S. Jochim, J. H. Denschlag,
and R. Grimm, Observation of the pairing gap in a strongly interacting
Fermi gas, Science 305, 1128 (2004).
[81] M. Greiner, C. A. Regal, and D. S. Jin, Probing the excitation spectrum of
a Fermi gas in the BCS-BEC crossover regime, Phys. Rev. Lett. 94, 070403
(2005).
[82] M. Zwierlein, J. Abo-Shaeer, A. Schirotzek, C. Schunck, and W. Ketterle,
Observation of high-Tc superfluidity, Nature 435, 1047 (2005).
[83] B. DeMarco, Quantum behavior of an atomic Fermi gas, Ph.D. thesis, University of Colorado - Boulder, 2001.
[84] J. Stenger, S. Inouye, M. R. Andrews, H.-J. Miesner, D. M. Stamper-Kurn,
and W. Ketterle, Strongly enhanced inelastic collisions in a Bose-Einstein
condensate near Feshbach resonances, Phys. Rev. Lett. 82, 2422 (1999).
[85] J. L. Roberts, N. R. Claussen, S. L. Cornish, and C. E. Wieman, Magnetic
field dependence of ultracold inelastic collisions near a Feshbach resonance,
Phys. Rev. Lett. 85, 728 (2000).

BIBLIOGRAPHY

131

[86] P. O. Fedichev, M. W. Reynolds, and G. V. Shlyapnikov, Three-body recombination of ultracold atoms to a weakly bound s-level, Phys. Rev. Lett.
77, 2921 (1996).
[87] E. Nielsen and J. H. Macek, Low-energy recombination of identical bosons
by three-body collisions, Phys. Rev. Lett. 83, 1566 (1999).
[88] B. D. Esry, C. H. Greene, and J. P. Burke, Recombination of three atoms
in the ultracold limit, Phys. Rev. Lett. 83, 1751 (1999).
[89] E. Braaten, H.-W. Hammer, and S. Hermans, Nonuniversal effects in the
homogeneous Bose gas, Phys. Rev. A 63, 063609 (2001).
[90] B. D. Esry, C. H. Greene, and H. Suno, Threshold laws for three-body
rescombination, Phys. Rev. A 65, 010705(R) (2001).
[91] D. S. Petrov, Three-body problem in Fermi gases with short-range interparticle interaction, Phys. Rev. A 67, 010703(R) (2003).
[92] P. G. de Gennes, Superconductivity of Metals and Alloys (Addison-Wesley,
California, 1966).
[93] M. Tinkham, Introduction to Superconductivity (Krieger, Malabar, Florida,
1980).
[94] J. J. Sakurai, Modern quantum mechanics (Addison-Wesley, Reading, MA,
1994).
[95] D. S. Petrov, C. Salomon, and G. V. Shlyapnikov, Weakly bound dimers of
fermionic atoms, Phys. Rev. Lett 93, 090404 (2004).
[96] G. A. Baker, Neutron matter model, Phys. Rev. C 60, 054311 (1999).
[97] H. Heiselberg, Fermi systems with long scattering lengths, Phys. Rev. A 63,
043606 (2001).
[98] J. Kinast, A. Turlapov, and J. E. Thomas, Damping of a unitary Fermi gas,
Phys. Rev. Lett. 94, 170404 (2005).
[99] J. E. Thomas, J. Kinast, and A. Turpalov, Virial theorem and universality
in a unitary Fermi gas, Phys. Rev. Lett. 95, 120402 (2005).
[100] M. Marini, F. Pistolesi, and G. C. Strinati, Evolution from BCS superconductivity to Bose condensation: Analytic results for the crossover in three
dimensions, Eur. Phys. J. B 1, 151 (1998).

BIBLIOGRAPHY

132

[101] C. J. Pethick and H. Smith, Bose-Einstein condensation in dilute gases

(Cambridge Univ. Press, Cambridge, UK, 2002).
[102] J. Carlson, S.-Y. Chang, V. R. Pandharipande, and K. E. Schmidt, Superfluid Fermi gases with large scattering length, Phys. Rev. Lett. 91, 050401
(2003).
[103] G. E. Astrakharchik, J. Boronat, J. Casulleras, and S. Giorgini, Equation
of state of a Fermi gas in the BEC-BCS crossover: A quantum Monte Carlo
study, Phys. Rev. Lett. 93, 200404 (2004).
[104] M. J. Holland, C. Menotti, and L. Viverit, The role of boson-fermion correlations in the resonance theory of superfluids, cond-mat/0404234 (2005).
[105] L. Vichi and S. Stringari, Collective oscillations of an interacting trapped
Fermi gas, Phys. Rev. A 60, 4734 (1999).
[106] U. Fano, Effects of configuration interaction on intensities and phase shifts,
Phys. Rev. 124, 1866 (1961).
[107] H. Feshbach, A unified theory of nuclear reactions II, Ann. Phys. 19, 287
(1962).
[108] W. C. Stwalley, Stability of spin-aligned hydrogen at low temperatures and
high magnetic fields: New field-dependent scattering resonance and predissociations, Phys. Rev. Lett. 37, 1628 (1976).
[109] E. Tiesinga, B. J. Verhaar, and H. T. C. Stoof, Threshold and resonance phenomena in ultracold ground-state collisions, Phys. Rev. A 47, 4114 (1993).
[110] J. P. Burke, Theoretical investigations of cold alkali atom collisions, Ph.D.
thesis, University of Colorado - Boulder, 1999, and references therein.
[111] M. H. Szyma
nska, K. Goral, T. Kohler, and K. Burnett, Conventional character of the BCS-BEC crossover in ultracold gases of 40 K, Phys. Rev. A 72,
013610 (2005).
[112] K. Goral, T. Kohler, S. A. Gardiner, E. Tiesinga, and P. S. Julienne, Adiabatic association of ultracold molecules via magnetic-field tunable interactions, J. Phys. B 37, 3457 (2004).
[113] A. Derevianko, W. R. Johnson, M. S. Safronova, and J. F. Babb, Highprecision calculations of dispersion coefficients, static dipole polarizabilities,
and atom-wall interaction constants for alkali-metal atoms, Phys. Rev. Lett.
82, 3589 (1999).

BIBLIOGRAPHY

133

[114] M. H. Szyma
nska, Private communication; the full coupled channels calculation was done by P. Julienne.
[115] A. Corney, Atomic and Laser Spectroscopy (Clarendon Press, Oxford, UK,
1977).
[116] C. Chin, A simple model of Feshbach molecules, cond-mat/0506313 (2005).
[117] G. F. Gribakin and V. V. Flambaum, Calculation of the scattering length
in atomic collisions using the semiclassical approximation, Phys. Rev. A 48,
546 (1993).
[118] C. Ticknor, private communication.
[119] R. B. Diener and T.-L. Ho, The condition for universality at resonance
and direct measurement of pair wavefunctions using rf spectroscopy, condmat/0405174 (2004).
[120] G. M. Bruun and C. J. Pethick, Effective theory of Feshbach resonances and
many-body properties of Fermi gases, Phys. Rev. Lett. 92, 140404 (2004).
[121] S. De Palo, M. L. Chiofalo, M. J. Holland, and S. J. J. M. F. Kokkelmans,
Resonance effects on the crossover of bosonic to fermionic superfluidity,
Phys. Lett. A 327, 490 (2004).
[122] M. Bartenstein et al., Precise determination of 6 Li cold collision parameters
by radio-frequency spectroscopy on weakly bound molecules, Phys. Rev.
Lett. 94, 103201 (2005).
[123] C. J. Myatt, Bose-Einstein condensation experiments in a dilute vapor of
Rubidium, Ph.D. thesis, University of Colorado - Boulder, 1997.
[124] J. R. Ensher, The first experiments with Bose-Einstein condensation of 87 Rb,
Ph.D. thesis, University of Colorado - Boulder, 1998.
[125] B. DeMarco, J. L. Bohn, J. P. Burke, Jr., M. Holland, and D. S. Jin,
Measurement of p-wave threshold law using evaporatively cooled fermionic
atoms, Phys. Rev. Lett. 82, 4208 (1999).
[126] B. DeMarco, S. B. Papp, and D. S. Jin, Pauli blocking of collisions in a
quantum degenerate atomic Fermi gas, Phys. Rev. Lett. 86, 5409 (2001).
[127] S. Chu, J. E. Bjorkholm, A. Ashkin, and A. Cable, Experimental observation
of optically trapped atoms, Phys. Rev. Lett. 57, 314 (1986).

BIBLIOGRAPHY

134

[128] R. Grimm, M. Weidem

uller, and Y. B. Ovchinnikov, in Advances in Atomic,
Molecular, and Optical Physics (PUBLISHER, ADDRESS, 2000), Vol. 42,
Chap. Optical dipole traps for neutral atoms, pp. 95170.
[129] D. A. Butts and D. S. Rokhsar, Trapped Fermi gases, Phys. Rev. A 55,
4346 (1997).
[130] L. Viverit, S. Giorgini, L. P. Pitaevskii, and S. Stringari, Adiabatic compression of a trapped Fermi gas, Phys. Rev. A 63, 033603 (2001).
[131] Q. Chen, J. Stajic, and K. Levin, Thermodynamics of interacting fermions
in atomic traps, cond-mat/0411090 (2005).
[132] B. Jackson, P. Pedri, and S. Stringari, Collisions and expansion of an ultracold dilute Fermi gas, Europhys. Lett. 67(4), 524 (2004).
[133] C. R. Monroe, E. A. Cornell, C. A. Sackett, C. J. Myatt, and C. E. Wieman,
Measurement of Cs-Cs elastic scattering at T = 30 K, Phys. Rev. Lett. 70,
414 (1993).
[134] J. L. Roberts, J. P. Burke, Jr., N. R. Claussen, S. L. Cornish, E. A. Donley,
and C. E. Wieman, Improved characterization of elastic scattering near a
Feshbach resonance in 85 Rb, Phys. Rev. A 64, 024702 (2001).
[135] J. P. Burke, Jr., C. H. Greene, J. L. Bohn, H. Wang, P. L. Gould, and W. C.
Stwalley, Determination of 39 K scattering lengths using photoassociation
spectroscopy of the Og state, Phys. Rev. A 60, 4417 (1999).
[136] C. Menotti, P. Pedri, and S. Stringari, Expansion of an interacting Fermi
gas, Phys. Rev. Lett. 89, 250402 (2002).
[137] Y. Kagan, E. L. Surkov, and G. V. Shlyapnikov, Evolution of a Bose gas in
anisotropic time-dependent traps, Phys. Rev. A 55, R18 (1997).
[138] S. Gupta, Z. Hadzibabic, J. R. Anglin, and W. Ketterle, Collisions in zero
temperature Fermi gases, Phys. Rev. Lett. 92, 100401 (2004).
[139] S. L. Cornish, N. R. Claussen, J. L. Roberts, E. A. Cornell, and C. E.
Wieman, Stable 85 Rb Bose-Einstein condensates with widely tunable interactions, Phys. Rev. Lett. 85, 1795 (2000).
[140] J. M. Gerton, D. Strekalov, I. Prodan, and R. G. Hulet, Direct observation of growth and collapse of a Bose-Einstein condensate with attractive
interactions, Nature 408, 692 (2000).

BIBLIOGRAPHY

135

[141] J. L. Roberts, N. R. Claussen, S. L. Cornish, E. A. Donley, E. A. Cornell,

and C. E. Wieman, Controlled collapse of a Bose-Einstein condensate, Phys.
Rev. Lett. 86, 4211 (2001).
[142] D. M. Harber, H. J. Lewandowski, J. M. McGuirck, and E. A. Cornell, Effect
of cold collisions in spin coherence and resonance shifts in a magnetically
trapped ultracold gas, Phys. Rev. A 66, 053616 (2002).
[143] C. Ticknor, C. A. Regal, D. S. Jin, and J. L. Bohn, Multiplet structure
of Feshbach resonances in nonzero partial waves, Phys. Rev. A 69, 042712
(2004).
[144] E. A. Donley, N. R. Claussen, S. T. Thompson, and C. E. Wieman, Atommolecule coherence in a Bose-Einstein condensate, Nature 417, 529 (2002).
[145] N. R. Claussen, E. A. Donley, S. T. Thompson, and C. E. Wieman, Microscopic dynamics in a strongly interacting Bose-Einstein condensate, Phys.
Rev. Lett. 89, 010401 (2002).
[146] E. Hodby, S. T. Thompson, C. A. Regal, M. Greiner, A. C. Wilson, D. S.
Jin, E. A. Cornell, and C. E. Wieman, Production efficiency of ultracold
Feshbach molecules in bosonic and fermionic systems, Phys. Rev. Lett. 94,
120402 (2005).
[147] E. Timmermans, P. Tommasini, M. Hussein, and A. Kerman, Feshbach resonances in atomic Bose-Einstein condensates, Phys. Rep. 315, 199 (1999).
[148] F. A. Abeelen and B. J. Verhaar, Time-dependent Feshbach resonance scattering and anomalous decay of a Na Bose-Einstein condensate, Phys. Rev.
Lett. 83, 1550 (1999).
[149] F. H. Mies, E. Tiesinga, and P. S. Julienne, Manipulation of Feshbach resonances in ultracold atomic collisions using time-dependent magnetic fields,
Phys. Rev. A 61, 022721 (2000).
[150] L. B. Ratcliff, J. L. Fish, and D. D. Konowalow, Electronic transition dipole
moment functions for a transitions among the twenty-six lowest-lying states
of Li2 , J. Mol. Spectrosc. 122, 293 (1987).
[151] C. Chin and P. S. Julienne, Radio-frequency transitions on weakly-bound
ultracold molecules, Phys. Rev. A 71, 012713 (2005).
[152] K. G. Petrosyan, Fermionic atom laser, JETP Lett. 70, 11 (1999).

BIBLIOGRAPHY

136

[153] P. Torma and P. Zoller, Laser probing of atomic Cooper pairs, Phys. Rev.
Lett. 85, 487 (2000).
[154] H. Moritz, T. Stoferle, K. G
unter, M. Kohl, and T. Esslinger, Confinement
induced molecules in a 1D Fermi gas, Phys. Rev. Lett. 94, (2005).
[155] M. Greiner, C. A. Regal, J. T. Stewart, and D. S. Jin, Probing paircorrelated fermionic atoms through correlations in atom shot noise, Phys.
Rev. Lett. 94, 110401 (2005).
[156] S. T. Thompson, E. Hodby, and C. E. Wieman, Ultracold molecule production via a resonant oscillating magnetic field, Phys. Rev. Lett. 95, 190404
(2005).
[157] J. P. DIncao and B. D. Esry, Scattering length scaling laws for ultracold
three-body collisions, Phys. Rev. Lett. 94, 213201 (2005).
[158] H. Suno, B. D. Esry, and C. H. Greene, Recombination of three ultracold
Fermionic atoms, Phys. Rev. Lett. 90, 053202 (2003).
[159] S. T. Thompson, E. Hodby, and C. E. Wieman, Spontaneous dissociation
of 85 Rb Feshbach molecules, Phys. Rev. Lett. 94, 020401 (2005).
[160] J. P. DIncao and B. D. Esry, Enhancing observability of the Efimov effect
in ultracold atomic gas mixtures, cond-mat/0508474 (2005).
[161] N. Balakrishnan, R. C. Forrey, and A. Dalgarno, Quenching of H2 vibrations
in ultracold 3 He and 4 He collisions, Phys. Rev. Lett. 80, 3224 (1998).
[162] R. C. Forrey, N. Balakrishnan, A. Dalgarno, M. R. Haggerty, and E. J.
Heller, Quasiresonant energy transfer in ultracold atom-diatom collisions,
Phys. Rev. Lett. 82, 2657 (1999).
[163] P. Soldan, M. T. Cvitas, J. M. Hutson, P. Honvault, and J.-M. Launay,
Quantum dynamics of ultracold Na and Na2 collisions, Phys. Rev. Lett. 89,
153201 (2002).
[164] J. Zhang, E. G. M. van Kempen, T. Bourdel, L. Khaykovich, J. Cubizolles,
F. Chevy, M. Teichmann, L. Tuarruell, S. J. J. M. F. Kokkelmans, and
C. Salomon, Expansion of a lithium gas in the BEC-BCS crossover, condmat/0410167 (2004).
[165] K. Xu, T. Mukaiyama, J. R. Abo-Shaeer, J. K. Chin, D. E. Miller, and W.
Ketterle, Formation of quantum-degenerate sodium molecules, Phys. Rev.
Lett. 91, 210402 (2003).

137

BIBLIOGRAPHY

[166] S. D
urr, T. Volz, A. Marte, and G. Rempe, Observation of molecules produced from a Bose-Einstein condensate, Phys. Rev. Lett. 92, 020406 (2003).
[167] T. Weber, J. Herbig, M. Mark, H.-C. Nagerl, and R. Grimm, Three-body
recombination at large scattering lengths in an ultracold atomic gas, Phys.
Rev. Lett. 91, 123201 (2003).
[168] C. Chin and R. Grimm, Thermal equilibrium and efficient evaporation of
an ultracold atom-molecule mixture, Phys. Rev. A 69, 033612 (2004).
[169] L. D. Carr, G. V. Shlyapnikov, and Y. Castin, Achieving a BCS transition
in an atomic Fermi gas, Phys. Rev. Lett. 92, 150404 (2004).
[170] S. Giorgini, L. P. Pitaevskii, and S. Stringari, Condensate fraction and critical temperature of a trapped interacting Bose gas, Phys. Rev. A 54, R4633
(1996).
[171] M. W. Zwierlein, C. H. Schunck, C. A. Stan, S. M. F. Raupach, and W.
Ketterle, Formation dynamics of fermion pair condensate, Phys. Rev. Lett.
94, 180401 (2005).
[172] A. V. Avdeenkov and J. L. Bohn, Pair wave functions in atomic Fermi
condensates, Phys. Rev. A 71, 023609 (2005).
[173] E. Altman and A. Vishwanath, Dynamic projection on Feshbach molecules:
a probe of pairing and phase flucutations, Phys. Rev. Lett. 95, 110404
(2005).
[174] A. Perali, P. Pieri, and G. C. Strinati, Extracting the condensate density
from projection experiments with Fermi gases, Phys. Rev. Lett. 95, 010407
(2005).
[175] L. Viverit, S. Giorgini, L. Pitaevskii, and S. Stringari, Momentum distribution of a trapped Fermi gas with large scattering length, Phys. Rev. A 69,
013607 (2004).
[176] M. L. Chiofalo, S. Giorgini, and M. Holland, Released momentum distribution of a Fermi gas in the BCS-BEC crossover, in progress (2005).
[177] E. Arimondo, M. Inguscio, and P. Violino, Experimental determinations of
the hyperfine structure in the alkali atoms, Rev. Mod. Phys. 49, 31 (1977).
[178] B. DeMarco, H. Rohner, and D. Jin, An enriched
atom studies, Rev. Sci. Instrum. 70, 1967 (1999).

40

K source for fermionic

BIBLIOGRAPHY

138

[179] D. Sesko, T. Walker, C. Monroe, A. Gallagher, and C. Wieman, Collisional

losses from a light-force atom trap, Phys. Rev. Lett. 63, 961 (1989).
[180] G. Modugno, C. Benko, P. Hannaford, G. Roati, and M. Inguscio, SubDoppler laser cooling of fermionic 40 K atoms, Phys. Rev. A 60, R3373
(1999).
[181] K. L. Corwin, Z. Lu, C. F. Hand, R. J. Epstein, and C. E. Wieman,
Frequency-stabilized diode laser with Zeeman shift in an atomic vapor, Applied Optics 37, 3295 (1998).
[182] J. Goldwin, Quantum degeneracy and interactions in the 87 Rb-40 K BoseFermi mixture, Ph.D. thesis, University of Colorado - Boulder, 2005.
[183] K. B. MacAdam, A. Steinbach, and C. Wieman, A narrow band tunable
diode laser system with grating feedback, and a saturated absorption spectrometer for Cs and Rb, Am. J. Phys. 60, 1098 (1992).
[184] M. E. Gehm, K. M. OHara, T. A. Savard, and J. E. Thomas, Dynamics of
noise-induced heating in atom traps, Phys. Rev. A 58, 3914 (1998).
[185] N. R. Claussen, Dynamics of Bose-Einstein condensates near a Feshbach
resonance in 85 Rb, Ph.D. thesis, University of Colorado - Boulder, 2003.
[186] D. M. Pozar, Microwave Engineering (John Wiley and Sons, Inc., New York,
1998).
[187] J. L. Roberts, Bose-Einstein condensates with tunable atom-atom interactions: The first experiments with 85 Rb BECs, Ph.D. thesis, University of
Colorado - Boulder, 2001.
[188] C. H. Schunck, M. W. Zwierlein, C. A. Stan, S. M. F. Raupach, W. Ketterle, A. Simoni, E. Tiesinga, C. J. Williams, and P. S. Julienne, Feshbach
resonances in 6 Li, Phys. Rev. A 71, 045601 (2005).
[189] T. Mizushima, K. Machida, and M. Ichioka, Direct imaging of spatially
modulated superfluid phases in atomic fermion systems, Phys. Rev. Lett.
94, 060404 (2005).
[190] D. E. Sheehy and L. Radzihovsky, BEC-BCS crossover in magnetized
Feshbach-resonantly paired superfluids, cond-mat/0508430 (2005).
[191] M. W. Zwierlein, A. Schirotzek, C. H. Schunck, and W. Ketterle, Fermionic
superfluidity with imbalanced spin populations and quantum phase transitin
to the normal state, cond-mat/0511197 (2005).

BIBLIOGRAPHY

139

[192] G. B. Partridge, W. Li, R. I. Kamar, Y. Liao, and R. G. Hulet, Pairing and

phase separation in a polarized Fermi gas, cond-mat/0511752 .
[193] H. Ott, E. de Mirandes, F. Ferlaino, G. Roati, G. Modugno, and M. Inguscio,
Collisionally induced transport in periodic potentials, Phys. Rev. Lett. 92,
140405 (2004).
[194] M. Greiner, C. A. Regal, C. Ticknor, J. L. Bohn, and D. S. Jin, Detection
of spatial correlations in an ultracold gas of fermions, Phys. Rev. Lett. 92,
150405 (2004).

Appendix A
Journal articles
In writing this thesis my goal was to discuss the main objective of my graduate
work, which was to access BCS-BEC crossover physics with a Fermi gas of atoms.
To adhere to this subject I necessarily omitted some of the work to which I
contributed during my graduate career. For example, I devoted time to the study
of p-wave Feshbach resonances in Fermi gases [57, 143] and to a feasibility study
of atom shot noise as a probe of ultracold gases [155]. Thus, for completeness I list
in Table A.1 journal articles I contributed to in varying extents in my graduate
work.

140

141

APPENDIX A. JOURNAL ARTICLES

Table A.1: Articles I co-authored as a graduate student.

Paper Title
1
Resonant control of elastic collisions in an
optically trapped Fermi gas of atoms
2
Tuning p-wave interactions in an ultracold
Fermi gas of atoms
3
Measurement of positive and negative scattering lengths in a Fermi gas of atoms
4
Creation of ultracold molecules from a Fermi
gas of atoms
5
Lifetime of molecule-atom mixtures near a
Feshbach resonance in 40 K
6
Detection of spatial correlations in an ultracold gas of fermions
7
Emergence of a molecular Bose-Einstein condensate from a Fermi gas
8
Multiplet structure of Feshbach resonances
in nonzero partial waves
9
Observation of resonance condensation of
fermionic atom pairs
10
Probing the excitation spectrum of a Fermi
gas in the BCS-BEC crossover regime
11
Production efficiency of ultracold Feshbach
molecules in bosonic and fermionic systems
12
Probing pair-correlated fermionic atoms
through correlations in atom shot noise
13
Momentum distribution of a Fermi gas of
atoms in the BCS-BEC crossover

Published Citation
2002
[54]
2003

[57]

2003

[59]

2003

[63]

2004

[67]

2004

[194]

2003

[68]

2004

[143]

2004

[73]

2005

[81]

2005

[146]

2005

[155]

2005

[78]