Академический Документы
Профессиональный Документы
Культура Документы
Quantum Mechanics
applied to Atoms and Light
Ph.W. Courteille
Universidade de São Paulo
Instituto de Fı́sica de São Carlos
03/05/2019
2
Content
0 Preface 1
0.1 Possible topics of partial courses . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
0.2 Organization of the course . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
0.3 Recommended bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
I Quantum Mechanics 5
3
4 CONTENT
2.2.1 Superposition principle (1. postulate) . . . . . . . . . . . . . . . . . . . . 42
2.2.2 Interpretation of the wave function (2. postulate) . . . . . . . . . . . . . . 42
2.2.3 Dirac bra-ket notation and vector representation . . . . . . . . . . . . . . 43
2.2.4 Observables (3. postulate) . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.2.5 Representation of operators as matrices . . . . . . . . . . . . . . . . . . . 44
2.2.6 Correspondence principle (4. postulate) . . . . . . . . . . . . . . . . . . . 45
2.2.7 Schrödinger equation and quantum measurements (5. postulate) . . . . . 45
2.2.8 Stationary Schrödinger equation . . . . . . . . . . . . . . . . . . . . . . . 46
2.3 Abstract formalism of quantum mechanics . . . . . . . . . . . . . . . . . . . . . . 47
2.3.1 Lie algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.3.2 Complete bases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.3.3 Degeneracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.3.4 Bases as unitary operators . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.3.5 Complete set of commuting operators . . . . . . . . . . . . . . . . . . . . 50
2.3.6 Uncertainty relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.3.7 Representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.3.8 Setting up a Hilbert space with several degrees of freedom . . . . . . . . . 53
2.4 Time evolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4.1 Unitary transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4.2 Schrödinger picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
2.4.3 Heisenberg picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.4.4 Interaction picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.4.5 Ehrenfest’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.5 Symmetries in quantum mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.5.1 Translation and rotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.5.2 Galilei and Lorentz boosts . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.5.3 Gauge transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.5.4 Noether’s theorem and conservation laws . . . . . . . . . . . . . . . . . . 67
2.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.6.1 Basic notions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.6.2 Postulates of quantum mechanics . . . . . . . . . . . . . . . . . . . . . . . 70
2.6.3 Abstract formalism of quantum mechanics . . . . . . . . . . . . . . . . . . 71
2.6.4 Time evolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
2.6.5 Symmetries in quantum mechanics . . . . . . . . . . . . . . . . . . . . . . 73
Preface
This script represents a synthesis of several postgraduate courses given at the Institute of Physics
of São Carlos (IFSC) of the University of São Paulo (USP). The courses are Quantum Mechanics
(SFI5774), Atomic and Molecular Physics (SFI5814), Quantum Mechanics B (SFI5707), Inter-
action of Light and Matter (SFI5877), and Atomic Optics (SFI5887). The topics of the courses
are, of course, closely intertwined. The purpose of this composite script is to emphasize the
interconnection of topics and facilitate the understanding of how they are related. In part I we
introduce the quantum mechanics, which represents the fundamental theory for the rest of the
book. In the second part we focus on the structure of the atom. In the third and fourth part
we study the properties of light, its interaction with individual atoms and atomic ensembles and
how the interaction is influenced by cavities and surfaces. Finally, in part V we introduce the
optics of matter wave.
The course is intended for masters and PhD students in physics. The script is a preliminary
version continually being subject to corrections and modifications. Error notifications and sug-
gestions for improvement are always welcome. The script incorporates exercises the solutions of
which can be obtained from the author.
1
2 CHAPTER 0. PREFACE
Sections QM QO AM AO
Antecedents 1.1.1-1.3.7 x
Repetition of basic notions of quantum mechanics 2.1.1-2.5.4 x x x
Motion in separable potentials 3.1.1-3.3.5 x
The harmonic oscillator 3.4.1-3.4.7 x x
Rotations, central potentials and spin coupling 4.1.1-4.4.4 x x
Stationary perturbation theory 5.1.1-5.3.2 x x
Time-dependent perturbation theory 5.4.1-5.4.6 x x x
Periodic systems 6.1.1-6.1.6 x
Spin of the electron and the (hyper-)fine structure 7.1.1-7.4.4 x x
Charged particles in electromagnetic fields 8.1.1-8.1.2 x x x
Atoms in stationary and electromagnetic fields 8.2.1-8.3.1 x x
Atoms with multiple electrons 9.1.1-9.4.4 x
Dimeric molecules 10.1.1-10.3.3 x
Collisions 11.1.1-11.2.3 x
Absorption and emission of radiation and spectral lines 12.1.1-12.3.2 x
Susceptibility and polarization 12.4.1-12.4.4 x x
Density operator and Bloch equations 13.1.1-13.2.4 x
Spontaneous emission and saturation 13.3.1-13.3.4 x
Line broadening processes 13.4.1-13.4.5 x
Multilevel systems 13.5.1-13.5.4 x
Quantization of radiation and dressed states 14.1.1-14.1.3 x
Jaynes-Cummings model and quantum correlations 14.2.1-14.2.3 x x
Spontaneous emission and the Markov approximation 14.3.1-14.3.1 x
Fluorescence and Rayleigh and Raman scattering 14.3.2-14.3.4 x
Beam splitting and quantum amplification 14.4.1-14.4.6
The laser 14.5.1-14.5.2
Squeezed states 14.6.1-14.6.5
Photon counting statistics 14.7.1-14.7.3
Forces exerted by electromagnetic fields 15.1.1-15.1.3 x x x
Cooling and trapping 15.2.1-15.2.6 x x
Quantum measurement 16.1.1-16.7.1 x x
Quantum correlations and entanglement 17.1.1-17.1.3 x
Creation of quantum correlations 17.2.1-17.2.6 x
Quantum gates 17.3.1-17.3.5 x
The structure factor 18.1.1-18.1.1 x
Coupled dipoles model 18.2.1-18.1.2 x
Bragg scattering and photonic band gaps 18.2.1-18.3.3 x
Interaction of atoms with cavities 19.1.1-19.2.5
CARL & consorts in the classical and quantum regime 20.1.1-20.3.4 x
Statistical models of CARL 21.1.1-21.3.2 x
Manipulation of atomic gases 22.1.1-22.6.5 x
Thermodynamics of degenerate gases 23.1.1-23.2.8 x
Bose-Einstein condensation 24.1.1-24.6.4 x
Superfluid properties of condensed gases 25.1.1-25.4.3 x
Coherent properties of condensed gases 25.5.1-25.6.3 x
Theories on the interaction of light and BECs 26.1.1-26.2.2 x
Bragg diffraction by BECs and matter wave superradiance 26.3.1-26.4.4 x
BECs in superpositions of states coupled by light 26.5.1-26.5.2 x
Interaction of BECs with optical cavities 26.6.1-26.6.10 x
Condensates in reduced dimensions 27.1.1-27.1.3 x
0.2. ORGANIZATION OF THE COURSE 3
Quantum Mechanics
5
Chapter 1
Derived constants:
Fine-structure constant α = e2 /4πε0 ~c ≈ 1/137
Avogadro constant NA = 1/uA × 1g/mol = 6.0221367 × 1023 1/mol
Gas constant R = NA kB = 8.314510 J/mol K
Vacuum permittivity ε0 = 1/µ0 c2 = 8.8542 × 10−12 As/Vm
Bohr radius aB = 4πε0 ~2 /me e2 = 0.529 × 10−10 m
Bohr magneton µB = e~/2me = 9.27 × 10−24 J/T
Classical electron radius re = α2 aB
Rydberg constant R∞ = me cα2 /2h = 13.7 eV
Compton wavelength λC = h/me c
Thomson cross section σe = (8π/3)re2
Muon mass mµ = 105.658389 MeV
Proton mass mp = 938.27231 MeV
Neutron mass mp = 939.56563 MeV
Deuteron mass md = 1875.61339 MeV
7
8 CHAPTER 1. ANTECEDENTS OF QUANTUM MECHANICS
With this fixing we give the energy in terms of e2cgs /aB , the wavevector in terms of 1/aB , the
distance in terms of aB and the mass in terms of me , such that,
Indefinite integrals:
Z p Z
∞
−ax2
∞
2 Γ( n+1
2 )
e dx = π/a e xn e−ax dx = n+1 . (1.9)
−∞ 0 2a 2
Higher momenta:
Z x1 x1 Z x1
−ax2 −ax2 2
e dx = xe + 2a x2 e−ax dx . (1.10)
x0 x0 x0
Definition:
Z ∞
2 dn −x2 2n 2
Hn (x) = (−1)n ex n
e =√ (x + it)n e−t dt (1.11)
dx π −∞
Hen (x) ≡ 2−n/2 Hn (x) .
Orthogonality:
Z ∞
2
e−x Hm (x)Hn (x)dx = 0 for m 6= n . (1.12)
−∞
Recursion:
d
Hn (x) = 2Hn−1 (x) (1.13)
dx
d −x2 2
e Hn (x) = e−x Hn+1 (x)
dx
Hn+1 (x) = 2xHn (x) − 2nHn−1 (x) .
Particular values:
Series:
∞
X (2n)! (−1)n−k
H2n (x) = (2x)2k (1.15)
(2k)! (n − k)!
k=0
X∞
(2n + 1)! (−1)n−k
H2n−1 (x) = (2x)2k+1
(2k + 1)! (n − k)!
k=0
int(n/2)
X 1 (−1)k
Hn (x) = n! (2x)n−2k .
k! (n − 2k)!
k=0
10 CHAPTER 1. ANTECEDENTS OF QUANTUM MECHANICS
Series:
n
X
n + m (−x)k
L(m)
n (x) = . (1.17)
n−m k!
k=0
Recursion:
d (m) (m+1)
L (x) = −Ln−1 (x) . (1.18)
dx n
Related functions:
q
2
umn (ε) ≡ e−ε · (iε)n−m · m! n−m 2
n! · Lm (ε ) (1.19)
q
umn (0) ≈ (iε)n−m · m!(n−m)!
n!
2
√
un+1,n (0) ≈ iε · n + 1 .
Fourier transforms:
Z ∞ √ √
2 (−1)k π 2 /4a
e−ax x2k cos xp · dx = √ · ep · He2k (p/ 2a) (1.20)
2k a2k+1
−∞
Z ∞ √
2 /2 (−1)m π 2 /2
e−x x2m L2m 2
n (x ) cos xp · dx =
√
2n!
· e−p · Hen (p)Hen+2m (p)
−∞
Z ∞ √
2 /2 (−1)m π 2 /2
e−x x2m+1 L2m+1
n (x2 ) sin xp · dx = √
2n!
· e−p · Hen (p)Hen+2m+1 (p)
−∞
Z ∞ Z ∞ Z ∞
−ax−bp 1 −ax −bp
e f (|x − p|) · dxdp = a+b e f (x)dx + e f (p)dp .
−∞ −∞ −∞
where,
∞
X Z ∞
(±z)t 1 xη−1 dx
gη (z) = ± = , (1.25)
tη Γ(η) 0 z −1 ex ± 1
t=1
is also called the Bose/Fermi function. The upper sign holds for bosons, the lower for fermions.
The Sommerfeld expansion,
Z ∞ Z y Z ∞ Z x
g(x)dx g(y + x)η−1 dx g(y − x)η−1 dx
x−y
= g(x)dx + x
− (1.26)
0 e +1 0 0 e +1 0 ex + 1
Z y
π2
≈ g(x)dx + g 0 (x) + ...
0 6
holds for z 1 and yields,
xη π 2 η(η − 1) 7π 4 η(η − 1)(η − 2)(η − 3)
fη (ey ) ≈ 1+ + + ... . (1.27)
Γ(η + 1) 6x2 360x4
For small z both functions converge towards,
Z ∞ η−1
1 x dx
cη (z) = = cη−1 (z) = z . (1.28)
Γ(η) 0 z −1 ex
The derivative is,
∞ ∞ ∞
∂fη (Z) ∂ X (−Z)t X (−Z)t−1 1 X (−Z)t fη−1 (Z)
= − η = η−1
= − η−1 = , (1.29)
∂Z ∂Z t t Z t Z
t=1 t=1 t=1
or,
Z ∞ Z ∞ η−1 −2 x
∂fη (Z) 1 ∂ xη−1 dx 1 x Z e dx
= −1
= (1.30)
∂Z Γ(η) ∂Z 0 Z e + 1 x Γ(η) 0 (Z −1 ex + 1)2
Z ∞ Z
1 d −1 η−1 ∞ xη−2 fη−1 (Z)
= xη−1 −1 x
dx = −1
dx = .
ZΓ(η) 0 dx Z e + 1 ZΓ(η) 0 Z ex + 1 Z
At the end of the 19th century, the physical world seemed simple: matter and light was all
that existed. Matter was made up of atoms and light was a wave. Therefore, to describe a real
system, it was enough to calculate the trajectories of its elementary particles, the propagation
of light and the way they interact. We now know that life is not as simple, and that atoms are
also waves and light also behaves like particles.
Frictions between the old notions and new observations appeared in the late 19th century,
such as the ultraviolet divergence of black-body radiation. The pioneer of the new ideas was
Max Planck who, in 1905, with a little help from Einstein quantized the electromagnetic field,
and therefore the light, into small harmonic oscillators. This was the starting point for the
development of a new theory called quantum mechanics. Soon, this theory was applied to
explain the photoelectric effect. The second important step was initialized by Niels Bohr, who
quantized the hydrogen atom in 1913 into discrete excitation levels.
Observing dust particles in a whirling motion within a ray of sunlight, Democritus was led to
the idea that atoms would behave in the same way, randomly colliding, some crowding, others
dispersing, others never yet joining with another atom.
The consistency of clusters of atoms, which makes something look solid, liquid, gaseous, or
animated (which is the state of the soul) would then be determined by the shape of the atoms
involved and their spatial arrangement. In this sense, water atoms are smooth and slippery;
the atoms of steel have shapes with sharp edges that hold them solidly together; the atoms of
salt, as their taste shows, are harsh and pointed; the atoms of air are small and little connected,
penetrating all other materials; and the atoms of soul and fire are spherical and very delicate.
Figure 1.2: Atoms of steel and air, atoms of the soul, and Bohr’s atom model.
We know nowadays that Democritus’ first theory of the structure of matter was very close
to the truth: There really are indivisible particles called atoms composed of a nucleus and an
electronic shell, and the space between the atomic nuclei is, in fact, quite empty.
The atomic hypothesis came to be reborn in the modern age with the scientists Boyle,
Clausius, Maxwell, and Boltzmann due to their successful explanations of the properties of
gases based on the so-called kinetic theory, where they assumed a gas being constituted of
identical molecules that collided elastically with each other and with the walls of the recipient
containing them. The discovery of the atom through the laws of proportions in chemistry and
the establishment of Avogadro’s number considerably strengthened the atomic hypothesis. The
hypothesis was definitely consecrated with the various experiments that established the charge
of the electron and the mass ratio between electrons and protons.
By the beginning of the 19th century the atomic nature of matter had definitely been es-
tablished, and the basic composition of the atoms was already relatively well known. It was
known, through experiments, that electrons could be removed from neutral atoms thus creating
positively charged ions and that only a certain number of electrons could be removed from each
14 CHAPTER 1. ANTECEDENTS OF QUANTUM MECHANICS
atom. This number proved to be dependent on the atomic species and was called the atomic
number Z. This information was fundamental for establishing the basic composition of atoms.
The question that arose at this point concerned the dimensions and configurations of the atomic
system. How would loads and masses be distributed in this entity?
Figure 1.3: Comparison of Rutherford scattering by free electrons and electrons strongly bound
to small nuclei. (a) Thomson’s ’raisin-in-a-pudding’ type atom; (b) Rutherford’s ’planetary’
atom. (c) Rutherford scattering by a raisin pudding atom and (d) by a planetary atom.
The atomic model proposed by Joseph John Thomson suggests a structure resembling a pud-
ding with raisins: the electrons would be homogeneously distributed within an extended nucleus
(size 0.1 nm) of positive charge thus compensating for the negative charge of the electrons. The
α-particles would penetrate the gold nucleus, perceived as almost homogeneous, but would suf-
fer multiple deflections due to collisions with the disordered electrons within the nucleus. Since
electrons are very light, the angle of deflection θ would be small, even after many collisions.
For this model we expect a Gaussian dependence of the particles’ deflection angle given by the
scattering cross section [see Fig. 1.3(a-b)],
dσ 2 2
∝ e−θ /θ0 , (1.31)
dΩ
where θ0 is a small angle.
1.2. THE DISCOVERY OF THE ATOM 15
However, the measurements performed on this Rutherford scattering showed different results:
• For a fixed scattering angle, the amount of particles scattered into a solid angle element
dΩ is proportional to the thickness of the metal foil.
• For a given fixed angle and a given metal sheet the amount of scattered particles in dΩ
2 , where E
varies inversely with Ekin kin is the kinetic energy of the α-particles.
• For a given energy and a given metal sheet, the number of particles scattered into dΩ is
proportional to (sin 2θ )−4 .
• For a given energy and sheet thickness, the number of particles scattered into dΩ in a
2 , where Z is the atomic number of the element that
given direction is proportional to Ztg tg
constitutes the sheet.
The extremely rare deflection of α-particles and their angular distribution can be understood
by the assumption that the positive charge is concentrated in a very small volume (∼ 1 fm, that
is 10000 times less than the size of the atom itself). This volume is called the atomic nucleus,
hence the denomination of nuclear model. Since most of the particles pass through the gold
sheet without hindrance, there must be a large gap between the nuclei. The electrons, which
move within a large (in comparison with the diameter of the nucleus) empty space (the vacuum)
around the nucleus, shield the positive nuclear charge, so that the atom appears outwardly
neutral.
Figure 1.4: (Left:) Trajectory of an α-particle. (Right:) Illustration of the scattering cross
section.
We now derive Rutherford’s scattering formula from the hypothesis of a point-like nucleus.
Due to the repulsive action of the Coulomb force,
Zα Ztg e2
F = , (1.32)
4πε0 r2
we have for the trajectory of the α-particle (Zα = 2) a hyperbola [see Fig. 1.4(a)]. The large
half-axis of the hyperbola can be determined from the following ansatz,
Zα Ztg e2 1
Ekin = , (1.33)
4πε0 2a
where 2a is the minimum distance of the particle α, when it collides with the nucleus in a central
collision 3 . The distance a depends on the kinetic energy and can also be used for non-central
3
In a central collision, when the α-particle reaches the minimum distance 2a, its initial kinetic energy, Ekin is
fully converted into potential energy.
16 CHAPTER 1. ANTECEDENTS OF QUANTUM MECHANICS
collisions. The collision parameter b is the minimum distance of the α-particle to the nucleus, if
it continued to fly in a straight line. In fact the α-particle will be deflected by an angle θ. From
the geometry of the hyperbola, as 2φ + θ = 180◦ , we obtain the following equation:
tan φ = ab = tan 90◦ − 2θ = cot 2θ , (1.34)
and therefore
b 8πε0 Ekin
cot 2θ = = b, (1.35)
a Zα Ztg e2
replacing a with the formula (1.33). Taking the derivative of this latter formula, we obtain a
relation between the width db of the hollow cone and the pertinent width dθ of the deflection
angle θ,
1 8πε0 Ekin
− 2 θ
dθ = db . (1.36)
2 sin 2 Zα Ztg e2
N
Let ntg = Vtg be the density of the particles in the target (Ntg atoms per volume V ) and x
the film thickness. Then σ = NAtg = VN/x tg
= ntg1 x is the average cross-section per atom sensed by
the α-particle on its way through the film. The probability P (θ)dθ for the α-particle of being
within a ring at distance b from the nucleus (whose area is 2πbdb) and being scattered into the
angle θ is then given by,
2πbdb
P (θ)dθ = = ntg x2πbdb . (1.37)
σ
These particles, i.e., dN of the N particles, are deflected into the hollow cone with the probability,
dN Zα2 Ztg
2 e4
1
= ntg x 2 2 · dΩ . (1.40)
N 256π ε0 Ekin sin4 2θ
2
That is Rutherford’s scattering formula. Often, the formula is expressed with the differential
dσ
cross section dΩ . We get,
dN dσ
= = ntg xdσ , (1.41)
N σ
and therefore 2
dσ Zα Ztg e2 1
= , (1.42)
dΩ 4πε0 · 4Ekin sin4 2θ
with
dN dσ
= N ntg x . (1.43)
dΩ dΩ
Here, we have to make some comments:
1.2. THE DISCOVERY OF THE ATOM 17
• The angle θ = 0 is not defined, since there exists a minimum deflection angle θmin . This
angle is reached, when the α-particle moves at the distance b = bmax from the atom, that
is, at the edge of the circular area of the cross section. For a greater collision parameter
b, the α-particle traverses the field of the next neighboring atom, and the deflection angle
increases again. We have:
A θmin Zα Ztg e2
σ= = πb2max e 2 ' tan θmin
2 = , (1.44)
Ntg 8πε0 Ekin · bmax
simply by inverting the formula (1.35). For very large impact parameters, that is, when
the α-particle passes the atom outside its electronic layer, the electrons of the atom shield
the charge of the nucleus, an effect called screening.
• For very high energies, the distribution of the nuclear charge over a finite volume influences
the scattering, calling for corrections in the Rutherford formula. Moreover, at short inter-
nuclear distances, nuclear forces appear additionally to the electromagnetic interaction.
Zπ
P (θ)dθ = 1 . (1.45)
θmin
-25
10
N
-30
10
0 50 100 150
φ
Rutherford derived the formula (1.42) describing the scattering of α-particles within classical
physics. A derivation from the laws governing quantum mechanics using the Born approximation
shows that Rutherford’s formula describes scattering correctly in first order, and that purely
quantum effects present only minor corrections. We will review the Rutherford scattering in
Excs. 1.4.1.1 and 1.4.1.2 and discuss the screening effect in Exc. 1.4.1.3.
18 CHAPTER 1. ANTECEDENTS OF QUANTUM MECHANICS
p2 e2 me ω 2 r2 e2
E= − = − , (1.49)
2me 4πε0 r 2 4πε0 r
dE dr e2 dr dr
−P = = me ω 2 r + 2
= 2me ω 2 r . (1.50)
dt dt 4πε0 r dt dt
The latter equation supposes an equilibrium between the centrifugal force and the Coulomb
force,
e2
me ω 2 r = , (1.51)
4πε0 r2
allowing link the revolution frequency ω to the instantaneous radius of the orbit r(t). Resolving
the Eq. (1.50) by ṙ and replacing the power by the relation (1.48) and the frequency ω by the
relation (1.51), we obtain,
dr P µ0 ω 2 e2 e4 1
=− 2
= − r = − 2 . (1.52)
dt 2me ω r 24πme c 96π ε0 me c r2
2 2 3
Insertion of the numerical values gives the decay time τ ∼ 10−10 s. This is the effect called
radiation collapse of the classical atomic model.
4
This type of model had already been proposed by Jean Perrin in 1901 and by Hantaro Nagaoka in 1903,
around the same time when Thompson developed his model. The planetary model was later on rescued by John
William Nicholson in 1911.
1.2. THE DISCOVERY OF THE ATOM 19
The z-direction is not influenced. With the ansatz x = aeiωt e y = beiωt we obtain the system
of equations,
which has a non-trivial solution for a and b only when the determinant of the coefficients of a
and b vanishes: 2
ω0 − ω 2 2iΩL ω
0= = ω 4 − (2ω02 + 4Ω2L )ω 2 + ω04 . (1.60)
−2iΩL ω ω02 − ω 2
We get, r q
1 Ω2L
ω = ω1,2 = ω02 + 2Ω2L ± 2ΩL ω02 + Ω2L = ω0 ± ΩL + + ... , (1.61)
2 ω0
20 CHAPTER 1. ANTECEDENTS OF QUANTUM MECHANICS
or, as ΩL ω, we get ω1,2 = ω0 ∓ ΩL . The result is a splitting of the energy levels proportional
to the magnetic field,
~e
∆E = 2~ΩL = B = 2µB B , (1.62)
me
where the abbreviation µB = e~/2me ' 9.27 · 10−24 JT−1 is called the magneton Bohr.
Although the classical derivation shows quantitative deviations from experimental observa-
tions, it is quite interesting, as it illustrates several aspects which have a quantum mechanical
equivalence.
L
~ = −µB
µ .
~
In the presence of a magnetic field the dipole undergoes an interaction W = −~
µ · B, and
therefore a feels a force,
F = −~µ · ∇B .
By subjecting beams of atoms to the gradient of a magnetic field and detecting this force,
Stern and Gerlach were able to measure the magnetic moment produced by the rotation of
the electrons around the atomic nuclei.
where m and n are integers. RH = (1/4π)2 (me e4 /4π~3 c) is the Rydberg constant and µ =
me mat /(me + mat ) the reduced mass.
The discrete nature of spectral lines and the problem of the radiation collapse led Niels Bohr
to formulate the following postulates:
1. There are specific stationary orbits, where electrons do not emit energy.
3. The laws of mechanics can describe the dynamic equilibrium of electrons in stationary
states, but fails to describe the transition of electrons between stationary orbits.
1.3. THE DISCOVERY OF THE PHOTON 21
Thus, Bohr’s model predicts the quantization of energy levels, known as first quantization of
quantum mechanics. The radii of the possible orbits can be calculated from the postulate that
the orbital angular momentum be quantized in units of ~, that is, the electrons form stationary
de Broglie waves along the orbits 5 . We discuss Bohr’s model in Excs. 1.4.1.6 and 1.4.1.7.
In the picture proposed by Bohr, the radiative decay happens as an abrupt transition of an
electron between an outer (more energetic) orbit and an inner (less energetic) orbit. Since the
energies of stationary orbits are very well defined, the emitted radiation is mono-energetic, i.e.,
the spectrum consists of discrete characteristic lines.
We note here that the picture of an abrupt transition of electrons between discrete states,
called the quantum jump, did not receive Schrödinger’s blessing. He rather imagined for elec-
trons, within his theory of quantum wave mechanics, wave-shaped orbitals instead of planetary
trajectories, thus avoiding the problem of radiation due to charge deceleration and the quantum
jump concept. According to him, during a transition between electronic orbits, the energy is
transformed into radiation gradually 6 .
dipole is often treated like a classical oscillator. The exposition of such a dipole to simple
boundary conditions prepares the analogous development of a quantum oscillator and provides
a direct path to quantization of the radiation field.
Even if we rarely do experiments by throwing light into a small hole in a metallic box, the
electromagnetic fields obtained by solving Maxwell’s equation are particularly simple for bound-
ary conditions, where the fields disappear on the inner surfaces of the box. Before discussing the
physics of radiation in a perfectly conducting cavity, we have to introduce some basic relations
between electromagnetic amplitudes, stored energy, and intensity.
The electric field of a plane wave oscillating with frequency ω and propagating through
vacuum in the direction of propagation defined by the wave vector,
2π
k= k̂ , (1.64)
λ
can be written,
E = E0 ei(k·r−ωt) , (1.65)
where E0 = E0 ê consists of an amplitude E0 and a polarization ê. Since the field E0 is transverse
to the direction of propagation, the polarization has two components perpendicular to k. The
magnetic induction field associated with the wave is,
1
B0 = E0 . (1.66)
c
For a propagating wave E and B are in phase, while for a standing wave they are out of phase.
For a given cavity mode we can express the standing wave in this mode as,
E = E0 (r)e−iωt . (1.67)
The energy of the electromagnetic field of a standing wave, averaged over one oscillation of the
frequency ω is, Z
1 ε0 2 1
Ū = |E| + |B|2 dV . (1.68)
2 2 2µ0
Now, the energy density of the oscillating electromagnetic field is given by,
dŪ 1 1
ū = = ε0 |E| + |B|2
2
. (1.69)
dV 4 µ0
From the equation (1.66) we can see that the contributions of the electric and magnetic fields
are equal. Therefore, Z
1 1
Ū = ε0 |E|2 dV e ū = ε0 |E|2 . (1.70)
2 2
Another important quantity is the flux of electromagnetic energy through a surface. The
Poynting vector describing this flux is defined by,
1
I= E×B . (1.71)
µ0
Again using the equation (1.66), we find the value averaged over a time period,
1
I¯ = ε0 c|E|2 . (1.72)
2
1.3. THE DISCOVERY OF THE PHOTON 23
This quantity, called intensity, describes the fact that the flux is a density of energy multiplied
with the velocity of propagation in vacuum,
1
ūc = ε0 c|E|2 = I¯ . (1.73)
2
1V2
W = . (1.75)
2 R
1 ∂2E
∇2 E = and ∇·E=0 . (1.76)
c2 ∂t2
The stationary waves solutions separate into terms oscillating in time and in space. Now,
Figure 1.6: (Left) Cavity in position space showing the thermal motion of the electrons inside
the walls. (Center and right) Density-of-states in a cavity in momentum space.
respecting the boundary conditions for a three-dimensional box of length L, we have for the
24 CHAPTER 1. ANTECEDENTS OF QUANTUM MECHANICS
components of E 9 ,
E(r, t) = e−iωt [êx cos(kx x) sin(ky y) sin(kz z) (1.77)
+êy sin(kx x) cos(ky y) sin(kz z)
+êz sin(kx x) sin(ky y) cos(kz z)] ,
with the components,
πnx
kx = para nx = 0, 1, 2, ... (1.78)
L
and similar for ky e kz . Note, that for each component Ex,y,z the transverse amplitudes disappear
in 0 and L. By inserting this solution into Helmholtz’s equation (1.76), we obtain,
ω2
kx2 + ky2 + kz2 =
. (1.79)
c2
The states kx,y,z (enumerated by nx,y,z ) form a three-dimensional orthogonal lattice of points in
space k separated by a distance along the axes kx , ky , kz of Lπ , as shown in Fig. 1.6. In principle,
the number of states that can be placed within a sphere of radius k in the momentum space is,
Z
N= dnx dny dnz . (1.80)
sphere
However, the periodic boundary conditions for |k| limit the components kx , ky , kz to positive
values (n ≥ 0), that is, the volume under consideration is limited to an octant. On the other
hand, we must multiply the number of states by two because of the degeneracy of polarizations,
Z n 3 Z k
2 L 4L3 k 3 4L3 ω 3
4N = 4πn dn = 4πk 2 dk = 2 = . (1.81)
0 π 0 π 3 3π 2 c3
With this, we obtain the mode density,
N ω3
= . (1.82)
L3 3π 2 c3
The spectral density of modes % can be given in several units,
Z Z Z
N
%(n)dn = %(k)dk = %(ω)dω = 3 , (1.83)
L
such that,
πn2 k2 ω2
%(n) =
3
ou %(k) = 2 or %(ω) = 2 3 . (1.84)
L π π c
The density of oscillating modes within the cavity grows like the square of the frequency.
Now, the mean energy per mode in a sample of oscillators in thermal equilibrium is, following
the equipartition law, equal to,
Ē = kB T , (1.85)
where kB is the Boltzmann constant. We conclude that the spectral energy density J in the
cavity is,
ω2
JRJ (ω)dω = kB T %(ω)dω = kB T 2 3 dω . (1.86)
π c
This law is known as the Rayleigh-Jeans law of black-body radiation. As seen in Fig. 1.7, this law
suggests the physically impossible fact, called ultraviolet catastrophe, that the energy storage in
the cavity grows without limits like the square of frequency.
9
See the script Electrodynamics by the same author [55].
1.3. THE DISCOVERY OF THE PHOTON 25
2000
T = 3000 K
1500
u (eV)
1000
500
0
0 200 400 600
ν (THz)
e−En /kB T
pn = . (1.87)
q
Now Planck hypothesized that the energy be quantized, that is, it must be assigned in discrete
portions, proportional to the frequency, such that,
En = n~ω , (1.88)
−~ω/k T
P∞ ~ isn called Planck’s
where n = 0, 1, 2, .. and the proportionality constant
−1
constant. With the
abbreviation U ≡ e B and using the rule n=0 U = (1 − U ) , we find the average
number,
X X ∂ X n U 1
n̄ = npn = (1 − U ) nU n = (1 − U )U U = = ~ω/k T . (1.89)
n n
∂U n 1−U e B −1
n̄n
pn = (1 − U )U n = , (1.90)
(1 + n̄)1+n
Finally, we obtain Planck’s expression for the energy density inside the cavity by substituting
¯ for the factor kB T in Rayleigh-Jeans’ law (1.85),
ω2 ~ω
J (ω)dω = Ēρω dω = 2 3 ~ω/k
dω . (1.92)
π c e BT − 1
This result, drawn in Fig. 1.7, is much more satisfactory, because now the energy density has an
upper bound, and coincides with the results of experiments. Note, that for high temperatures or
low excitation energies, ~ω kB T , Planck’s distribution converges to that of Rayleigh-Jeans’,
J (ω) → JRJ (ω).
Solve the Excs. 1.4.2.1 to 1.4.2.4.
E = hν . (1.93)
Assuming a fixed exit work A for the extraction of an electron, we can measure the constant ~:
mv 2 eV
hν = A + 2 = A + eV → h= . (1.94)
ν − νg
The energy of the fastest electrons is measured through the decelerating voltage by varying ν
and I. We will discuss the photoelectric effect quantitatively later in the Exc. 5.5.4.5.
wider than the spectral width of the atomic transition and where the spectral distribution of
the light flux from the source, I¯ω , is weak compared to the saturation intensity of the atomic
transition. Even if modern light sources generally have very narrow and intense spectral emission
bands, Einstein’s coefficients are often used in the spectroscopic literature to characterize the
light-matter interaction with atoms and molecules.
The Einstein rate equations describe the energy flux between atoms and the optical modes
of the cavity,
dN1 dN2
=− = −R1→2 + R2→1 + S2→1 (1.95)
dt dt
= −N1 B12 J (ω) + N2 B21 J (ω) + N2 A21 .
R1→2 is the absorption rate, R2→1 the stimulated emission rate and S2→1 the spontaneous emis-
sion rate. The assumption of a third type of transition, called spontaneous emission, is necessary,
if B12 = B21 but N1 > N2 in thermal equilibrium. In thermal equilibrium we have the condition
of stationarity, dN dN2
dt = − dt = 0 for a given energy density value J (ω) = Jth (ω), such that,
1
A21
Jth (ω) = . (1.96)
N1
N2 B12 − B21
The Boltzmann distribution law controlling the distribution of the number of atoms in the lower
and upper states is given by,
N1 g1
= e−(E1 −E2 )/kB T , (1.97)
N2 g2
where g1,2 are the degeneracies of the lower and upper states and E2 − E1 = ~ω0 . We find,
A21
Jth (ω) = g1 ~ω0 /kB T . (1.98)
g2 e B12 − B21
But this result must be consistent with the Planck’s distribution (1.92). Therefore, by comparing
this equation with the equation (1.98), it must be,
g1 B12
=1. (1.99)
g2 B21
and also,
A21 ~ω 3
= 2 03 . (1.100)
B21 π c
This equation shows that, once we know one of the three transition rates, we can always calculate
the others.
It is useful to compare the rate A21 with B21 from the equation (1.98) inserting the equation
(1.99),
A21
= e~ω0 /kB T − 1 . (1.101)
B21 Jth (ω)
This expression shows that, when ~(ω2 − ω1 ) kB T , that is, for optical, UV, or X-ray frequen-
cies, spontaneous emission dominates. But in low-frequency regimes, that is, IR, microwave, or
radio waves, stimulated emission is more important. Note that even when stimulated emission
dominates, spontaneous emission is always present and plays an important role, for example, in
processes ultimately limiting the emission bandwidth of lasers.
1.3. THE DISCOVERY OF THE PHOTON 29
P̃ = ε0 χ̃e Ẽ . (1.102)
The tilde ∼ ornamenting the quantities means that they are complex, and that measurable
quantities must be taken from the real parts. χ is the linear complex electric susceptibility,
which is an intrinsic property of the medium responding to the light field. It is related to the
medium’s dielectric constant,
ε̃ = ε0 (1 + χ̃) . (1.103)
Near resonances, the susceptibility is a strong function of frequency and can be spatially
anisotropic. It is a complex quantity with a real dispersive part, χ0e , and an imaginary absorptive
part, χ00e ,
χ̃e = χ0e + iχ00e . (1.104)
Several expressions that we already know become modified in a dielectric medium,
2
kc
= 1 + χ̃e , (1.105)
ω
The relationship between the amplitudes of the electric and magnetic fields is,11
(χ) √ √ p 1
B0 = εµE0 = ε0 µ0 1 + χ̃e E0 = (η + iκ)E0 = (η + iκ)B0 . (1.109)
c
We use the subscript (χ) to mark quantities within the dielectric medium. The average energy
density is,
1
ū(χ) = ε0 η 2 |E|2 = η 2 ū . (1.110)
2
11
In a dielectric medium, µ ' µ0 .
30 CHAPTER 1. ANTECEDENTS OF QUANTUM MECHANICS
Multiplying the two sides with ~ω0 , we see that the left side describes the rate of energy scattered
out of the beam by spontaneous emission, while the right side describes the energy loss of the
beam, i.e., the difference between energy removed by absorption and energy returned to the
beam by stimulated emission,
(χ)
dŪ (χ) ūω
= −N2 A21 ~ω0 = − 2 B12 (N1 − N2 gg12 )~ω0 . (1.119)
dt η
γ0 dω
Lγ 0 (ω − ω0 )dω = 0 2 , (1.121)
2π γ
(ω − ω0 )2 + 2
with the spectral width γ 0 . The differential L(ω − ω0 )dω can be considered as the probability
of finding the emitted light in the frequency range between ω and ω + dω. In fact, we will
retrieve the Lorentzian profile, when discussing other contributions to the spectral width, such
as excitations with strong radiation fields (power broadening) or broadening due to collisions.
The width γ 0 is generally caused by several physical sources.
Note that L(ω − ω0 )dω is dimensionless. Now, by generalizing the equation (1.119), the
energy loss rate of the light beam becomes,
The absorption probability distribution is convoluted with the energy distribution of the light
(χ)
source, ūω = dū(χ) (ω)/dω. The distribution of energy density is,
Figure 1.9: Absorption spectrum (blue) and spectral energy distribution of the source (red).
where V is the volume of the cavity. Assuming that light propagates toward z and converting
the time dependence into a spatial dependence,
I (χ)
Iω(χ) = . (1.129)
A21 /2π
Now, Eq. (1.128) becomes, with a bit of algebra,
dI¯(χ) B12 2π~ω0
=− ndz . (1.130)
I¯(χ) A21 cη
The right-hand parenthesis of Eq. (1.130) can be considered as a cross-section for absorption
of resonant light. Using Eq. (1.100) we can express this cross section as,
g2 πλ20
σ0 = , (1.131)
g1 2η
such that Eq. (1.130) becomes
dI¯(χ)
= −σ0 ndz (1.132)
I¯(χ)
and
I¯(χ)
(χ)
= e−σ0 nz0 . (1.133)
I¯
0
1.3. THE DISCOVERY OF THE PHOTON 33
Here, z0 is the total distance, over which absorption takes place. The last equation is the
integral version of the Lambert-Beer law for light absorption. It is very useful for measuring
atomic densities in gas cells or of atomic beams. Comparing the Eqs. (1.113) and (1.133), we
see that the absorption coefficient can be written as a product of the absorption cross section
and the density of the gas,
2ωκ ω
K= = χ00e = σ0 n . (1.134)
c ηc
Solve the Excs. 1.4.2.5 to 1.4.2.8.
gives the specific heat. The disappearance of the specific heat at low temperatures,
3N (~ω)2 −~ω/(kB T )
CV ' e , (1.138)
kB T 2
which is related to the finite localization energy of harmonic oscillators, does not describe ex-
perimental observations very well, and the model had to be refined by Debye, later on.
where v is the velocity of sound propagation, the formula is totally equivalent to the density-
of-states for photons in a cavity. Assuming
R νm that there is an upper bound νm for the vibrational
frequencies, we normalize as 3N0 = 0 ρ(ν)dν. The energy now is 13 ,
Z νm Z θ/T
~ω 4πV T4 x3 dx
E= νdν = 9N k B . (1.140)
0 e~ω/kB T − 1 v3 θ3 0 ex − 1
The Debye temperature θ = hνm /kB is characteristic for the metal. The derivative is then,
" Z #
T 3 θ/T x3 dx θ 1
CV = 9N kB 4 − . (1.141)
θ 0 ex − 1 T eθ/T − 1
1.4 Exercises
1.4.1 The discovery of the atom
1.4.1.1 Ex: Analysis of Rutherford scattering
a. What conclusions can be drawn from the observation that Rutherford’s formula describes well
the scattering of charged particles traveling through matter over a wide range of parameters?
b. Why do we see a deviation from Rutherford’s formula for large energies?
c. The scattering of protons with energy E crossing a thin film of thorium is well described up
to energies of E = 4.3 MeV by Rutherford’s formula. Estimate for this case the range of nuclear
forces.
d. For small scattering angles θ we observe large deviations from Rutherford’s formula. Explain
why?
e. Assume the thorium atoms of item (c) to form a periodic crystal with the lattice constant
d = 10aB . At which minimum angle θ Rutherford’s formula loses its validity.
Since light is a wave, in the most general form, assuming the validity of the superposition
R i(k·r−ωt)
principle, it can be described by a wave packet, A(r, t) = e a(k)d3 k. It is easy to verify
that the wave equation,
∂2
A − c2 ∇2 A = 0 , (2.2)
∂t2
reproduces the dispersion relation.
On the other hand, we have slow massive particles possessing kinetic energy,
p2
E= . (2.3)
2m
With the hypothesis of de Broglie that even a massive particle has wave quality, we can try
an ansatz 1 of a wave equation satisfying the dispersion relation (2.3). From Planck’s formula,
E = ~ω, and
R the formula of Louis de Broglie, p = ~k, describing the particle by a wave packet
ψ(r, t) = ei(k·r−ωt) ϕ(k)d3 k not subject to external forces, it is easy to verify that the equation,
∂ ~2 2
i~ ψ = − ∇ ψ, (2.4)
∂t 2m
reproduces the dispersion relation. If the particle is subject to a potential, its total energy is
E = p2 /2m + V (r, t). This dispersion relation corresponds to the famous Schrödinger equation,
∂ ~2 2
i~ ψ = − ∇ + V (r, t) ψ . (2.5)
∂t 2m
1
Trial, working hypothesis.
37
38 CHAPTER 2. FOUNDATIONS OF QUANTUM MECHANICS
m2 c4
ω 2 − c2 k 2 = . (2.6)
~2
This dispersion relation can be obtained from the differential equation,
∂2 2 2 m2 c4
A − c ∇ A = − A, (2.7)
∂t2 ~2
R
inserting, for example, the already proposed wave packet A(r, t) = ei(k·r−ωt) a(k)d3 k, supposed
not to be subject to external forces. The equation (2.7) is a wave equation called Klein-Gordon
equation. For particles without rest mass, as in the case of photons, the equation is reduced to
the wave equation of light (2.2).
Now, making the transition to non-relativistic velocities, v c, we can expand the dispersion
relation,
p
2 v2 ~2 k 2
2 4 2 2 2
E = m c + c m v = mc 1 + 2 + .. or ~ω ' mc2 + . (2.8)
2c 2m
In analogy with the Klein-Gordon equation we can derive the approximate dispersion relation
(2.8) from a wave equation,
∂ 2 ~2 2
i~ A = mc − ∇ A. (2.9)
∂t 2m
2 t/~
With the transformation ψ = e−imc A, we rediscover the Schrödinger equation (2.4),
∂ ~2
i~ ψ=− ∆ψ (2.10)
∂t 2m
as the non-relativistic limit of the Klein-Gordon equation.
It is interesting to note that in all cases discussed, obviously the dispersion relations and the
differential equations can be interconverted by the substitutions,
∂
E −→ i~ and p −→ −i~∇ . (2.11)
∂t
We will discuss this later in the context of Ehrenfest’s theorem in Secs. 2.1.6, 2.1.7, and 2.4.5.
−Q̇ = I . (2.18)
The more general solution of the Schrödinger equation can be written as a superposition of
plane waves ei(r·k−ωt) with frequencies ω = p2 /2~m and wave vectors k = p/~. Each plane
wave has an individual amplitude ϕ(p), such that,
Z Z
2
1
ψ(r, t) = h3/2 d3 pϕ(p)ei(r·k−ωt) = d3 p h3/2
1
ϕ(p)ei(r·p/~−p t/2m~) , (2.19)
with h ≡ 2π~. At time t = 0, this expansion is nothing more than a Fourier transform,
Z
ψ(r, 0) = h3/2 d3 pϕ(p)eir·k ,
1
(2.20)
In the absence of forces the momentum distribution becomes stationary. We can now use
the momentum distribution ϕ(p) as coefficients of the expansion of the temporal wavefunction
ψ(r, t), as shown above. Thus, the expansion represents a general solution of the time-dependent
Schrödinger equation. The magnitude |ϕ(p)|2 is the probability density in momentum space.
Example 2 (Normalization of the wave function in momentum space): It is easy
to show that the probability density in momentum space is also normalized:
Z Z Z Z
0
∗
2 3 1
|ϕ(p)| d p = h3 3 3
d p d rψ (r)e ir·k
d3 r0 ψ(r0 )e−ir ·k
Z Z Z
0
= d3 r d3 r0 ψ ∗ (r)ψ(r0 ) (2π)
1
3 d3 keik·(r−r )
Z Z Z
3 0 ∗ 0 3 0
= d r d r ψ (r)ψ(r )δ (r − r ) = |ψ(r)|2 d3 r = 1 ,
3
knowing that the Fourier transform of a plane wave is nothing more than the Dirac distri-
bution.
Since the probability distributions |ψ(r)|2 and |ϕ(p)|2 are interconnected by Fourier trans-
form, we already know that we can not localize 2 both simultaneously. If one is well localized,
the other is necessarily delocalized. Do the Exc. 2.6.1.2.
2.1.6 Eigenvalues
We have already seen that the position and momentum distributions of a particle are spread.
We calculate the mean values of these distributions, denoted by hri e hpi, as first moments of
the respective distributions:
Z Z
3 2
hri = d r|ψ(r, t)| r and hpi = d3 p|ϕ(p, t)|2 p . (2.22)
Z Z Z
= h3/2 ψ (r) i ∇ ϕ(p)e d pd r = ψ ∗ (r) ~i ∇ψ(r)d3 r .
1 ∗ ~ ik·r 3 3
2
Localize: Restrict the distribution volume indefinitely.
2.1. BASIC NOTIONS 41
This calculation shows that the expectation value, called eigenvalue, of the momentum can be
expressed through an operator p̂ ≡ (~/i)∇ acting on the wavefunction 3,4 .
More generally, we can compute the eigenvalue of a function in r e p via,
Z
hf (r, p)i = d3 rψ ∗ (r)f (r, p̂)ψ(r) . (2.23)
However, it is important to note that the operators r̂ and p̂ do not necessarily commute.
introducing the commutator [â, b̂] ≡ âb̂ − b̂â as an abbreviation. After that,
Z h i Z
~ ~
i i i
~ h[Ĥ, p̂]i = ~ h[V̂ , p̂]i = ~ d rψ V̂ i ∇ψ − i ∇(V ψ) = − d3 rψ ∗ ψ∇V = hF̂i .
3 ∗
(2.28)
3
From now on, the hat over a physical magnitude will denote quantum operators.
4
We note here that the rules hψ|x̂|ψi ↔ hφ|− ~i ∇p |φi and hψ| ~i ∇r |ψi ↔ hφ|p̂|φi from the Fourier transformation
are useful
2 for numerical simulations of the Schrödinger equation: Instead of calculating the spatial derivative
~
i
∇ of the wavefunction, one makes a Fast Fourier Transform (FFT) to momentum space, multiplies with p,
and transforms back.
42 CHAPTER 2. FOUNDATIONS OF QUANTUM MECHANICS
are possible states as well. This is called superposition principle, and means, for example, that
a particle may be simultaneously in several places or that an atom may be at the same time
excited and deexcited.
There are systems that can only exist in a restricted number of states, such as the two-level
atom. Others may exist in an infinite number of states or even in a continuous distribution of
states.
But the notation has other advantages. For example, let us suppose that we know the three
possible states of a system, |1i, |2i, and |3i, which are linearly independent. Then we can define
the states as vectors:
1 0 0
|1i = 0 , |2i = 1 , |3i = 0 . (2.35)
0 0 1
These three states can be interpreted as the basis of a vector space representing the system.
Now, each wavefunction can be expanded on this basis and expressed by a vector. An arbitrary
ket state of this system will then be,
c1
|ψi = c2 . (2.36)
c3
Now we can easily calculate the probability for a system to be in a state |ψi,
c1
||ψi|2 = hψ|ψi = c∗1 c∗2 c∗3 · c2 = |c1 |2 + |c2 |2 + |c3 |2 . (2.38)
c3
To find the current values aψ of any observable  in a specific situation given by a wave
function ψ, we need to solve an equation of eigenvalues,
We can rewrite the equation as aψ = hψ|Â|ψi. The values an are real numbers, if the observable
is a Hermitian operator, that is,
The eigenvalue of a projector, hP̂k i = hψ|P̂k |ψi = |hk|ψi|2 , is nothing more than the probability
of finding a system, whose general state is |ψi, in the particular state, since expanding as done
in (2.33), we have,
X
hP̂k i = c∗m cn hm|kihk|ni = |ck |2 . (2.44)
m,n
5
Note that there are theoretical attempts to generalize the concept of observables to non-Hermitian operators
[24, 25] only displaying PT -symmetry.
2.2. POSTULATES OF QUANTUM MECHANICS 45
Using the matrix formalism we can define other interesting operators and verify their prop-
erties,
1 0 0 0
|1ih1| = , |2ih2| = ,
0 0 0 1 . (2.45)
0 1 0 0
|1ih2| = = σ− , |2ih1| = = σ+
0 0 1 0
The rising and lowering operators, σ ± , are also called Pauli spin matrices, because they were
introduced by Wolfgang Pauli. The vector,
+
σ + σ−
~σ ≡ i(σ − − σ + ) (2.46)
+ −
[σ , σ ]
is called Bloch vector 6,7 . The eigenvalue of the Bloch vector has a fixed length (see Exc. 2.6.2.2).
The representation of physical quantities by matrices allows the description of quantum
superposition states. The purpose of Exc. 2.6.2.3 is to illustrate this at the example of a particle
passing through a double slit.
even in a superposition of states, while after the measurement we know exactly the state. This
amounts to a reduction of entropy, which is not allowed in a closed system.
The famous postulate of state reduction or projection of the wavefunction formulated by John
von Neumann describes the quantum measurement process as a sequence of two distinct steps 9 .
In a first step, the measuring apparatus projects the measured operator  on an eigenvector
basis. That is, if the measurement is compatible with the operator 10 , we obtain a distribution
of probability amplitudes of the results,
X X X
 y hÂi = hψ|Â|ψi = hψ|Â| ck |ki = ak ck hψ|ki = ak |ck |2 , (2.49)
k k k
P
with hψ|ψi = k |ak |2 = 1. Therefore, we can understand |hk|ψi|2 as the probability of the
system to be in the eigenstate |ki 11 .
In a second step, the observing scientist will read the measuring device and note the result,
which will necessarily be one of the possible ak ,
hÂi y ak . (2.50)
If the state is stationary, it will never change any more. That is, each subsequent measurement
will yield the same result. The Exc. 2.6.2.4 illustrates the process of quantum measurement at
the example of a measurement of the excitation energy of a two-level atom.
∂
ĤΨ(t, x) = i~ Ψ(t, x) , (2.51)
∂t
p̂ 2
∂
with Ĥ ≡ 2m + V (x, t) and p̂ ≡ −i~ ∂x . If the potential is independent of time, V (x, t) = V (x),
we can do the following ansatz, Ψ(x, t) ≡ ψ(x)f (t). Insertion into the Schrödinger equation
yields,
1 ~2 d2 i~ d
− 2
+ V (x) ψ(x) = f (t) = const. ≡ E . (2.52)
ψ(x) 2m dx f (t) dt
The solution of the right-hand side of the equation is i~(ln f − ln f0 ) = E(t − t0 ). Hence,
is nothing more than an eigenvalue equation. This means that the Schrödinger wave mechanics
is equivalent to the mechanics of the Heisenberg matrices. The Excs. 2.6.2.5 and 2.6.2.6 are first
simple calculations of the eigenvalues and eigenvectors of a two-level system.
9
For simplicity, we only consider pure state, here.
10
To understand the meaning of compatible, we must establish a more complete theory of measurement including
the reservoir in the quantum description.
11 P
Alternatively, Â −→ k |kihk|Â|kihk|.
2.3. ABSTRACT FORMALISM OF QUANTUM MECHANICS 47
For a Hermitian operator,  = † , we have hψ|Â|ψi = hÂψ|ψi or hÂi ≡ hψ|Â|ψi = hÂi∗ , using
the Dirac bra-ket notation,
hψ|† ≡ |ψi . (2.57)
There are identity and nullity operators,
which can be 6= 0. The sum of two hermitian operators is hermitian, but the product is not,
since,
( + B̂)† = † + B̂ † =  + B̂ but (ÂB̂)† = B̂ † † = B̂  6= ÂB̂ . (2.60)
On the other hand, the following relations of hermitian operators are always hermitian,
the functions are called linearly independent. A set of linearly independent functions may form
a basis. The space opened by a set of linearly independent functions is called Hilbert space.
An operator  is completely characterized by its eigenvalues and eigenfunctions. If a set
of eigenfunctions |ni is complete, every allowed state of the system can be expanded in these
eigenfunctions, X
|ψi = cn |ni and Â|ni = an |ni . (2.66)
n
2.3.3 Degeneracy
The eigenvectors form a natural basis for the Hilbert space. However, a problem arises in the case
of degeneracy, that is, when some eigenvalues are equal, an = am . In this case, the eigenvectors
that correspond to degenerate eigenvalues are not completely defined, and we have to construct
a basis verifying that all constructed eigenvectors are orthogonal. For this, there exists the
2.3. ABSTRACT FORMALISM OF QUANTUM MECHANICS 49
method of orthogonalization by Schmidt, which works like this: We assume that we have already
solved the eigenvalue equation, that we found a degenerate eigenvalue, Â|ak i = a|ak i for every
k = 1, .., gk , where gk is the degree of degeneracy, and that we also found a complete basis of
eigenvalues |am i, but which is not orthogonal, that is, ∃{m, n} with han |am i 6= 0. The task is
to build another basis |bm i satisfying hbn |bm i = δn,m .
The first vector of the orthogonal base can be chosen freely, e.g.,
|b1 i ≡ |a1 i . (2.72)
Since de basis {|ak i} is assumed to be complete, the second vector is necessarily a linear combina-
tion of vectors |ak i, that is, |b2 i = |a2 i+λ|b1 i. With the condition hb1 |b2 i = 0 = hb1 |a2 i+λhb1 |b1 i
we can determine the parameter λ, and obtain for the second vector,
hb1 |a2 i
|b2 i ≡ |a2 i − |b1 i . (2.73)
hb1 |b1 i
In the same way, we can derive for a third vector, |b3 i = |a3 i + µ|b1 i + ν|b2 i, the conditions,
hb1 |b3 i = 0 = hb1 |a3 i + µhb1 |b1 i and hb2 |b3 i = 0 = hb2 |a3 i + νhb2 |b2 i, and obtain,
hb1 |a3 i hb2 |a3 i
|b3 i ≡ |a3 i − |b1 i − |b2 i . (2.74)
hb1 |b1 i hb2 |b2 i
An overall way of writing this down is,
|b1 ihb1 | |b2 ihb2 | |bk−1 ihbk−1 |
|bk i ≡ 1 − − − ... − |ak i . (2.75)
hb1 |b1 i hb2 |b2 i hbk−1 |bk−1 i
In the Exc. 2.6.3.2 we practice the orthogonalization of a set of three linearly independent
but non-orthogonal vectors, and in the Exc. 2.6.3.3 we find an orthogonal basis for a partially
degenerate three-level system.
Also,
Â|ni = Ê|ni and ÂU = U Ê . (2.80)
That is, by knowing the unitary matrix (or transformation matrix) U , we can solve the eigenvalue
problem simply by Ê = U −1 ÂU .
Note, that this does not apply to a non-orthonormal basis. In this case, we need to do a
Schmidt orthogonalization and use the condition det Û = 1. We apply the technique detailed in
this section to solve Excs. 2.6.3.4 and 2.6.3.5.
In addition, we observe that if two operators commute, the orthonormal basis constructed
for one of the operators is also orthonormal for the other. That is, if two operators  and B̂
commute and if |ψ1 i and |ψ2 i are two eigenvectors of  with different eigenvalues, then the
matrix element hψ1 |B̂|ψ2 i is equal to zero:
Finally, we affirm that, if two operators  and B̂ commute, we can construct an orthonormal
basis {|ψa,b i} with common eigenvectors of  and B̂:
The fact that commuting operators have a common system of eigenvectors authorizing sharp
eigenvalues can be used to construct and characterize a state.
are the plane waves eipx x/~ and eipy y/~ . Therefore, the total state of the particle can be
described by,
|ψpx ,py ,pz i = |ψpx i|ψpy i = e(i/~)(px x+py y) f (z) .
2.3. ABSTRACT FORMALISM OF QUANTUM MECHANICS 51
However, these eigenfunctions are infinitely degenerate, since the linear momentum in z-
direction is not specified. A third operator p̂z |ψi = pz |ψi commutes with the others,
[p̂k , p̂m ] = 0 .
Hence,
|ψpx ,py ,pz i = e(i/~)(px x+py y+pz z) ,
is a possible state of the system.
∂2
On the other hand, choosing p̂2z = −~2 ∂z 2
2 as the third operator, giving the eigenvalues pz ,
|ψpx ,py ,p2z i = e(i/~)(px x+py y) cos p~z z or |ψpx ,py ,p2z i = e(i/~)(px x+py y) sin p~z z . (2.84)
Therefore, there are two solutions with the same eigenvalues, px , py , p2z . To lift this degener-
acy, we need to introduce yet another observable. This observable can be, for example, the
parity P̂ , that is, the behavior of the wave function upon mirroring z −→ −z in the x-y plane.
The fact that the set of operators px , py , pz on one hand and px , py , p2z , P̂ on the other are
equivalent, shows that the required number of observables for a complete characterization
depends on their judicious choice.
Also, the number needed for a complete set of commuting operators (CSCO) depends on the
number of degrees of freedom and the symmetry of the system. In the case of the free particle
in one dimension it is enough to consider one observable only, for example, x̂ or p̂. In three
dimensions, we already need at least three commuting observables. In Exc. 2.6.3.7 we will try
to find a CSCO for a matrix with partially degenerate eigenvalues.
2.3.7 Representations
2.3.7.1 Spatial representation
A Hilbert space can be discrete or, as in the case of the momentum of a free particle, continuous.
In this latter case, the eigenvalues are continuously distributed, since the equation,
has solutions for each value of E. The eigenfunctions are ψ(r) = aeip·r/~ . Eq. (2.86) clearly has
the form of an eigenvalue equation, for which we have already introduced the Heisenberg matrix
formalism. The question now is how these descriptions combine.
52 CHAPTER 2. FOUNDATIONS OF QUANTUM MECHANICS
Observables that do not commute correspond to expansions on different bases and generate
alternative representations. For example, we can represent quantum mechanics in position space
or linear momentum space. If |ri is a basis of the space of the particles’ state,
Z
0 3 0
r̂|ri = r|ri , hr |ri = δ (r − r) , |rihr|d3 r = 1̂ , (2.87)
R3
The quantities hr|ψ(t)i = ψ(t, r) Schrödinger wave functions. We can also say that the wave-
functions are the coordinates of the state in the particular base |ri. Consequently,
hr|r̂|r0 i = rδ 3 (r − r0 ) (2.89)
0 3 0
hr|f (r̂)|r i = f (r)δ (r − r ) . (2.90)
where the quantity A(r, r0 ) ≡ hr|Â|r0 i is called kernel of the operator. The transition from
Heisenberg’s abstract mechanics to Schrödinger’s wave mechanics is done by the substitutions
|ψ(t)i → ψ(t, r) and  → A(r, r0 ).
where the prefactor ~−3/2 is introduced to take account of the unit of the states 12 . ψ and ϕ are
different representations of the same quantum state related by,
Z Z
3
hr|ψ(t)i = 1
hr|pihp|ψ(t)id p = h3/2 eir·p/~ ϕ(p, t)d3 p = ψ(r, t) (2.95)
ZR3
Z R3
defining the function ϕ̃(k) ≡ ~3/2 ϕ(p). Applying the Fourier transform to functions of operator
we can calculate,
Z Z
hr|G(p̂)|r i = d phr|G(p̂)|pihp|r i = d3 pG(p)hr|pihp|r0 i
0 3 0
(2.98)
Z
0
1
= ~3/2 d3 pG(p)eik·(r−r ) = ~13 (FG)(r − r0 ) .
In Exc. 2.6.3.10 we will show hr|p̂|ψi = (~/i)∇r hr|ψi, thus justifying that we can understand
an operator as a rule to determine what happens to a function. For example, the rule p̂x asks
for a derivation of the wavefunction by x.
acting on wavefunctions |ψi ∈ H which can be expanded on a basis |ii of H. Now, we consider
a subspace R ⊂ H defined by the base |ki. Then the projector P̂R can be represented by,
0 0 0
1
X
P̂R ≡ |kihk| = 0 .. 0
. (2.100)
k 1
|ki∈R
0 0 0
54 CHAPTER 2. FOUNDATIONS OF QUANTUM MECHANICS
The projector,
1 0 0
P̂ = 0 1 0
0 0 0
Obviously, the concatenation (2.103) only serves to increase the Hilbert space of a given
degree of freedom described by a given observable, e.g., when we add one more level of energy
to the spectrum of an atom described by a Hamiltonian. If, in contrast, we want to add another
degree of freedom, we need the external sum or external product discussed below.
2.3. ABSTRACT FORMALISM OF QUANTUM MECHANICS 55
The symbol ⊗ denotes the external tensorial product of two vectors (states). Now, in order to
represent the multidimensional space HA ⊗ HB by a matrix, we use the fact that it is isomorphic
to the space HI ⊗ HA⊗B , that is, we proceed to a reorganization of the quantum numbers
identifying,
α 1 β 1
α1 |βi α1 β2
|γi ≡ |αi|βi = |αi ⊗ |βi = |αβi = =
α2 β1 = (γm )m ∈ HA ⊗ HB , (2.106)
α2 |βi
α2 β 2
where m = 1, 2, 3, 4 is identified with (i, k) = (1, 1), (1, 2), (2, 1), (2, 2). The new vector is element
of the 4-dimensional vector space HA ⊗ HB . If {|αii } and {|βik } are bases in their respective
spaces HA and HB , then {|γim } is a basis of the product space HA ⊗ HB .
For observables we proceed in the same way: The external product of two commutators
spans a Hilbert product space with the dimension corresponding to product of the dimensions
of the sub-spaces. Assuming that,
X X
 ≡ |iiAij hj| and B̂ ≡ |kiBkl hl| and [Â, B̂] = 0 . (2.107)
i,j k,l
then X
 ⊗ B̂ ≡ |ikiAij Bkl hjl| , (2.108)
(ik)(jl)
such that
dim  ⊗ B̂ = dim  dim B̂ .
For example, |iihj| ⊗ |kihl| = |ikihjl|.
56 CHAPTER 2. FOUNDATIONS OF QUANTUM MECHANICS
For two two-dimensional operators  and B̂, the external product is defined by,
A11 B11 A11 B12 A12 B11 A12 B12
A11 B21 A11 B22 A12 B21 A12 B22
 ⊗ B̂ =
A21 B11 A21 B12 A22 B11 A22 B12 ,
(2.109)
A21 B21 A21 B22 A22 B21 A22 B22
Obviously, the external product is associative (Â⊗ B̂)⊗C = Â⊗(B̂ ⊗C), but does not commute.
Nevertheless, we can reverse the order of the product of two operators using,
1
0 1
 ⊗ B̂ = Ŝ(B̂ ⊗ Â)Ŝ com Ŝ =
1 0 .
(2.111)
1
We note, that it is important to distinguish from what space the vector came from. In our
notation, the vector before the symbol of the tensorial product (⊗) is belongs to the space HA ,
and the one after the ⊗ belongs to the space HB . With the definition (2.107) we can verify that
the operators only act on their respective states:
We can check the relationship (2.112) by the definitions (2.106) and (2.109) of the external
product,
A11 B11 A11 B12 A12 B11 A11 B12 α11 β11
A11 A12 B11 B12 α1 β1 A11 B21 A11 B22 A12 B21 A11 B22
⊗ ⊗ = α11 β21
A21 A22 B21 B22 α2 β2 A21 B11 A11 B12 A22 B11 A22 B12 α21 β11
A21 B21 A11 B22 A22 B21 A22 B22 α21 β21
A11 B11 α11 β11 + A11 B12 α11 β21 + A12 B11 α21 β11 + A11 B12 α21 β21
A11 B21 α11 β11 + A11 B22 α11 β21 + A12 B21 α21 β11 + A11 B22 α21 β21
=A21 B11 α11 β11 + A11 B12 α11 β21 + A22 B11 α21 β11 + A22 B12 α21 β21
(2.113)
A21 B21 α11 β11 + A11 B22 α11 β21 + A22 B21 α21 β11 + A22 B22 α21 β21
(A11 α1 + A12 α2 )(B11 β1 + B12 β2 )
(A11 α1 + A12 α2 )(B21 β1 + B22 β2 )
=(A21 α1 + A22 α2 )(B11 β1 + B12 β2 )
that is,
A11 + B11 A11 + B12 A12 + B11 A12 + B12
A11 + B21 A11 + B22 A12 + B21 A12 + B22
 ⊕ B̂ =
A21 + B11
. (2.115)
A21 + B12 A22 + B11 A22 + B12
A21 + B21 A21 + B22 A22 + B21 A22 + B22
It can be decomposed as,
1 1 1 1
 ⊕ B̂ =  ⊕ O + O ⊕ B̂ =  ⊗ + ⊗  . (2.116)
1 1 1 1
Again, using the definition (2.111) of the unitary operator Ŝ, we can reverse the order of the
operator by,
 ⊕ B̂ = Ŝ(B̂ ⊕ Â)Ŝ . (2.117)
Example 7 (Direct external sum of two diagonal Hamiltonians): As an example we
consider a two-level atom excited by radiation and trapped in an external harmonic potential.
We assume that the degrees of freedom do not interact. As the Hamiltonian of the HO is
diagonal, the total Hamiltonian is organized into a diagonal matrix of quadratic subspaces,
..
~ω
2
~ω
2 + ~Ω 0 0 .
~ω ..
2 + ~Ω ~ω
+ ~∆ 0 0 .
2
0 ~Ω .
Ĥ = ~ω(n + 2 ) ⊕
1
= .
. .
2 + ~Ω
3~ω 3~ω
~Ω ~∆ 0 0 2
..
2 + ~Ω 2 + ~∆
3~ω 3~ω
0 0 .
..
··· ··· ··· ··· .
It acts on the product state |ni|ii, where the first ket denotes the vibrational level and the
second ket the electronic excitation of the atom.
2.3.8.4 Trace
The trace of an operator over a subspace reduces its domain to the remaining dimensions (the
·-symbol is a place holder for the dimension over which we do NOT want to trace):
X X
TrB Â ⊗ B̂ = h· m|ikiAij Bkl hjl| · mi = |iiAij Bkl hj|δkm δlm (2.118)
(ik)(jl)(·m) (ik)(jl)(·m)
X X
= |iiAij Bmm hj| = Â Bmm = ÂTrB B̂ .
(i)(j)(m) m
For example, Trρ Â ⊗ ρ̂ = Â. See the Excs. 2.6.3.14 and 2.6.3.15.
It can be shown,
TrÂB̂ = TrB̂ Â . (2.119)
58 CHAPTER 2. FOUNDATIONS OF QUANTUM MECHANICS
|ψS (t)i −→ Û (t)† |ψS (t)i ≡ |ψH i and ÂS (t) −→ Û (t)† ÂS (t)Û (t) ≡ ÂH (t) (2.127)
with the transformation defined by equation (2.125), is equivalent. The subscript H means the
Heisenberg picture. In particular, we obviously have,
Thus, the matrix element of the operator ÂS in Schrödinger’s picture with the time-dependent
base {|ψS i} is equal to the matrix element of the operator ÂH = U † ÂS U in Heisenberg’s picture
with the time-independent base {|ψH i}. In this picture the wavefunctions are independent of
time,
d d
|ψH i = |ψS (0)i = 0 , (2.129)
dt dt
but the operators depend im- and explicitly on time,
∂ ∂ d i
p̂S = 0 =⇒ p̂H = 0 =⇒ p̂H = [Ĥ, p̂H ] .
∂t ∂t dt ~
In the Exc. 2.6.4.3 we will use the Heisenberg picture to derive the equations of motion for
a particle confined to a potential.
60 CHAPTER 2. FOUNDATIONS OF QUANTUM MECHANICS
Analogously to Eq. (2.125), we define a time evolution operator in terms of the time-independent
part of the total Hamiltonian,
|ψI (t)i = eiĤ0 t/~ |ψS (t)i and AI (t) = eiĤ0 t/~ ÂS e−iĤ0 t/~ . (2.134)
Now we are interested in the temporal dependence of quantum states and operators in the
interaction picture. Replacing the inverse function |ψS (t)i = e−iĤ0 t/~ |ψI (t)i in the Schrödinger
equation (2.124) we immediately see,
∂
V̂ (t)|ψI (t)i = i~ |ψI (t)i . (2.135)
∂t
Apparently, in the interaction picture, only the perturbative term in Hamiltonian controls the
temporal evolution. Taking the time derivative of both sides of the equation by setting the
operator ÂI in the interaction representation to Eq. (2.134) results in,
dÂI i
= [Ĥ0 , ÂI ] . (2.136)
dt ~
Therefore, we see that the time derivative can be expressed in the form of a commutator, resem-
bling the Heisenberg equation (2.131), except that only the unperturbed term of the Hamiltonian
appears in the argument of the commutation operator.
for all m. Indeed, this situation can always be arranged by absorbing the diagonal elements
of V̂ into Ĥ0 , thus modifying its eigenvalues to,
0
Em = Em + hm|V̂ |mi . (2.138)
When V̂ is small, the generator S of the transformation will likewise be small. The trans-
formation can then be expanded in S using the Baker-Campbell-Haussdorf formula,
Ĥ 0 = Ĥ0 + V̂ + [iS, Ĥ0 ] + [iS, V̂ ] + 21 [iS, [iS, Ĥ0 ]] + 12 [iS, [iS, V̂ ]] + . . . . (2.141)
The Hamiltonian can be made diagonal to first order in V̂ by choosing the generator S such
that,
[Ĥ0 , iS] = V̂ . (2.142)
This equation always has a definite solution under the assumption that V̂ is off-diagonal in
the eigenbasis of Ĥ0 . Substituting this choice in the previous transformation yields:
This expression is the standard form of the Schrieffer-Wolff transformation. Note that all the
operators on the right-hand side are now expressed in a new basis ’dressed’ by the interaction
V̂ to first order.
In the general case, the difficult step of the transformation is to find an explicit expression
for the generator S. Once this is done, it is straightforward to compute the Schrieffer-Wolff
Hamiltonian by computing the commutator [S, V̂ ]. The Hamiltonian can then be projected
on any subspace of interest to obtain an effective projected Hamiltonian for that subspace. In
order for the transformation to be accurate, the eliminated subspaces must be energetically
well separated from the subspace of interest, meaning that the strength of the interaction V̂
must be much smaller than the energy difference between the subspaces. This is the same
regime of validity as in standard second-order perturbation theory.
leaves the physics of a system unchanged. The question is now, how the Schrödinger equation,
d
Ĥ|ψi = i~ |ψi (2.145)
dt
transforms into the new system, that is, what will the Hamiltonian Ĥ 0 look like in the trans-
formed equation,
? d
Ĥ 0 |ψU i = i~ |ψU i . (2.146)
dt
We calculate,
d d
i~ |ψU i = i~U † |ψi + i~U̇ † |ψi = (U † Ĥ + i~U̇ † )|ψi (2.147)
dt dt
= (U † Ĥ + i~U̇ † )U |ψU i = (U † ĤU + i~U̇ † U )|ψU i = Ĥ 0 |ψU i . (2.148)
Hence,
Ĥ 0 = U † ĤU + i~U̇ † U . (2.149)
62 CHAPTER 2. FOUNDATIONS OF QUANTUM MECHANICS
Example 11 (Interaction picture): The above derivation is general and holds for any
unitary transformation. We will now apply it to transform the Hamiltonian Ĥ = Ĥ0 + V̂ (t)
into the interaction picture via the transformation U = e−(i/~)Ĥ0 t . From
i i
U̇ † = Ĥ0 e(i/~)Ĥ0 t = Ĥ0 U † (2.150)
~ ~
we calculate,
h i i
Ĥ 0 = U † ĤU + i~U̇ † U = U † Ĥ0 + V̂ (t) U + i~ Ĥ0 U † U (2.151)
h i ~
= U † Ĥ0 + V̂ (t) U − Ĥ0 = U † V̂ (t)U , (2.152)
which confirms the validity of the Schrodinger equation ... in the interaction picture provide
the Hamiltonian is taken to be the perturbation part V̂ (t), only. In the Heisenberg picture
V̂ (t) = 0, such that,
Ĥ 0 = 0 . (2.153)
δF (r̂)
[p̂, F (r̂)] = −i~ . (2.155)
δr̂
The momentum observable is not singularly defined by the commutation relation, because
each unitarily transformed operator satisfies the relation as well. We can expand a unitarily
equivalent momentum as p̃ = U pU + = eiF (r) pe−iF (r) = p + i[F (r), p] + 2!1 [F (r), [F (r), p]] + ...
using the relation (2.199).
The observables in the Heisenberg picture follow the same equations of motion as the cor-
responding classical quantities. This correspondence principle is called Ehrenfest theorem. For
~2 2
example, when working with position and momentum variables [x̂, k̂] = i and Ĥ = 2m k̂ + V (x̂),
we obtain,
δ Ĥ δ Ĥ
[x̂, Ĥ] = i~ and [p̂, Ĥ] = −i~ , (2.156)
δ p̂ δ x̂
and using the Heisenberg equation (2.131),
δ Ĥ δ Ĥ
x̂˙ = and p̂˙ = − . (2.157)
δ p̂ δ x̂
We will demonstrate this in Exc. 2.6.4.5 for the case of a harmonic potential.
In the Schrödinger picture the equation of motion for the eigenvalues of the observables takes
the form,
d ∂ i
hÂS i = h∂t ψ|ÂS |ψi + hψ|∂t ÂS |ψi + hψ|ÂS |∂t ψi = hÂS i + h[Ĥ, ÂS ]i . (2.158)
dt ∂t ~
2.5. SYMMETRIES IN QUANTUM MECHANICS 63
The eigenvalues behave as Heisenberg observables in Eq. (2.130), that is, they follow the laws
of Hamilton’s and Newton’s mechanics.
The important result now is that the equations that govern the eigenvalues of the observables
are identical in the both pictures, since from the Heisenberg picture we obtain with Eq. (2.130),
d ∂ i
hÂH i = hÂH i + h[Ĥ, ÂH ]i .
dt ∂t ~
The temporal evolution has already been discussed extensively in Sec. 2.4.
Ttr r ≡ a + r . (2.161)
Before this, we need to derive the following calculation rule for commutators, which will be done
in Exc. 2.6.5.1:
e B̂e− = B̂ + [Â, B̂] + 2!1 [Â, [Â, B̂]] + ... . (2.162)
Applying this formula to the two operators p̂ and r̂ related by the commutation rule (2.47),
we obtain,
0
e(i/~)a·p̂ r̂e(−i/~)a·p̂ = r̂ + [(i/~)a · p̂, r̂] + 1
2! [(i/~)a · p̂, a] + ... = r̂ + a . (2.163)
and forms a group since Utr (a)Utr (b) = Utr (a + b). Summarizing the impact of the translation
on the operators of space,
† †
Utr (a)r̂Utr (a) = r̂ + a , Utr (a)p̂Utr (a) = p̂ , (2.166)
Hence,
Utr (a)|ri = e(−i/~)a·p̂ |ri = |r + ai . (2.168)
Finally, comparing the expansion of the translation operator
(−i/~)a·p̂ i 1 (a · p̂)2
e |ri = 1 − a · p̂ − 2 + .. |ri , (2.169)
~ ~ 2!
with the Taylor expansion of the translated state,
(a · ∇)2
|r + ai = 1 + a · ∇ + + .. |ri . (2.170)
2!
we obtain
~
p̂|ri = − ∇|ri . (2.171)
i
We calculate,
X (~
α×)n
eα~ × r = ~ × r + 21 α
r=r+α ~ × (~
α × r) + .. (2.173)
n
n!
= êα (êα · r) + êα × r sin α − êα × (êα × r) cos α ,
as we will see in Exc. 2.6.5.2. We define the unitary rotational transformation by,
†
Urt (~ α) = eα~ × r̂
α)r̂Urt (~ , α)|ri = |eα~ × ri .
Urt (~ (2.174)
To derive the explicit form of the rotation operator, we consider two rotations about the
same axis α
~ = λ1 êα + λ2 êα , such that
Urt (λ1 êα )Urt (λ2 êα ) = Urt (λ1 êα + λ2 êα ) . (2.175)
α) = e(−i/~)L·~α .
Urt (~ (2.177)
The explicit form of L follows from its action on a state |ri. Comparing the expansion of the
operator (2.177),
Urt (α)|ri = 1 − ~i L · α
~ + ... |ri (2.178)
with the Taylor expansion of the state,
|eα~ × ri = |r + α
~ × r + ...i = |ri + (~
α × r) · ∇r |ri + ... , (2.179)
we find,
− ~i L̂ · α
~ |ri = (~ ~ · (r × ∇r )|ri = − ~i r̂ × p̂|ri ,
α × r) · ∇r |ri = α (2.180)
that is,
L̂ = r̂ × p̂ . (2.181)
Therefore, the observable L̂ is the orbital angular momentum of the particle producing the
rotations.
Tv r = r + vt and Tv p = p + mv . (2.182)
Holds Tv1 Tv2 = Tv1 +v2 . We derive the expression for the operator of this transformation from
its actions on the position and momentum states. For the position we have with (2.164),
That is, by defining G = p̂t − r̂m = i~∇v UG (v)|v=0 , the unitary transformation
and analogously,
Z Z
(−i/~)v·pt (−i/~)v·pt 3
e |pi = e |rihr|pid p = 1
h3/2
|rie(i/~)p·(r−pt) d3 p = |ri . (2.187)
R3 R3
13
We need to keep in mind that, following Glauber’s rule, e eB̂ = eÂ+B̂+[Â,B̂]/2 with  ≡ v · p̂t and B̂ ≡ v ·r̂m.
However, the commutator [v · p̂t, v · r̂m] = −i~mtv2 is independent of p̂ and r̂, such that it only contributes an
unimportant phase.
66 CHAPTER 2. FOUNDATIONS OF QUANTUM MECHANICS
[G · a, G · b] = 0 , (2.188)
Obviously, for very high velocities, the Galilei-boost should be replaced by the Lorentz trans-
form (or Lorentz boost) [139]. Here, we only note, that the additivity of velocities expressed by
equation (2.188) does not hold for non-collinear relativistic velocities 14
Also, we know that the fields are invariant under the substitution,
Φ → Φ0 ≡ Φ − ∂t χ and A → A0 ≡ A + ∇χ , (2.191)
obviously keeps the Schrödinger equation invariant. Transforming operators and wave functions
as,
Ĥ → Ucl ĤUcl−1 ≡ ĤU and |ψi → Ucl |ψi ≡ |ψU i , (2.193)
we calculate for the energy,
d −1 −iq −1 dχ
ĤU |ψU i = Ucl i~ dt Ucl |ψU i = Ucl i~Ucl−1 dt
d
|ψU i + Ucl i~ ~ Ucl dt |ψU i
= i~ dtd
− iq~ dχdt |ψU i , (2.194)
d −1 d dχ
Ucl i~ Ucl = i~ + q and Ucl p̂Ucl−1 = p̂ − q∇χ . (2.196)
dt dt dt
14
See the script Electrodynamics of the same author [55].
15
See the script Electrodynamics of the same author [55].
16
In quadrivetorial notation i~∂µ −→ i~∂µ + q∂µ χ.
2.5. SYMMETRIES IN QUANTUM MECHANICS 67
This shows that the gauge transformation applies to the minimum coupling rule,
Ucl Ucl
Ĥ = Ĥkin + qΦ y Ĥkin + qΦ + q∂t χ and mv = p − qA y p − qA − q∇χ , (2.197)
confirming the rules (2.191). That is, the Hamiltonian of a particle carrying the charge q and
interacting with an electric potential Φ and a potential magnetic vector A is,
Ĥ = 1
2m (p − qA − q∇χ)2 + qΦ + q∂t χ . (2.198)
Therefore, to find a conservation law, i.e., an invariable observable (also called constant of
motion), we must verify that the observable and the transformed wavefunctions simultaneously
satisfy the same fundamental equations (that is, Schrödinger’s or Heisenberg’s equation) as
the original observable and wavefunctions. For example, if the wavefunction |ψi satisfies the
Schrödinger equation, the wave function U |ψi must do this too,
! d dU d dU
ĤU |ψi = i~ U |ψi = i~ |ψi + i~U |ψi = i~ |ψi + U Ĥ|ψi . (2.200)
dt dt dt dt
Consequently, we obtain the relation,
As shown in (2.158) and (2.159), an operator that commutes with the Hamiltonian does not
explicitly depend on time, that is, it is conserved.
with
P̂ 2 = P̂ . (2.206)
We talk about even parity when P̂ |ψ(r)i = |ψ(r)i and odd parity when P̂ |ψ(r)i = −|ψ(r)i. See
Exc. 2.6.5.4.
17
Imagine that the forces attracting two bodies to each other are not equal: Contrary to Newton’s third law,
body A attracts body B, more than the body B attracts the body A. In that case after a while the two bodies
have different momenta. With the unitary transformation Utr (a) = e−ip·a/~ ' 1 − ip · a/~ + ... we have,
L(xµ ) = 1 µν µ ε0 2 1 2
4µ0 F Fµν − Aµ j = 2 E − 2µ0 B − Aµ j
µ
(2.208)
ε0 2 1 2
= 2 [∇Φ + ∂t A] − 2µ0 [∇ × A] − Φρ + A · j ,
From the Lagrangian formulation, Maxwell’s equations can be derived by requiring the action
to be minimal, δS = 0, which yields the Euler-Lagrange equations. As the field equations do
not change under gauge transformation, this implies that the action is also unchanged.
To find the relation with charge conservation, we simply have to compare the actions in
different gauges. First, we express the Lagrangian transformed into the old gauge,
0
L0 (xµ ) = ε0
2 {∇[Φ − ∂t χ ] + ∂t [A + ∇χ ]}2 − 1
2µ0 {∇ × [A + ∇χ ]}2 (2.210)
− [Φ − ∂t χ]ρ + [A + ∇χ] · j
= L + (∂t χ)ρ + ∇χ · j .
With this result, we can calculate the difference between the actions under gauge transformation
and recall, that they can not be different:
Z Z
! 0
0 = S − S = [(∂t χ)ρ + ∇χ · j]dV dt = − χ[∂t ρ + ∇ · j]dV dt , (2.211)
using partial integration 18 and choosing volumes so large, that every charge is inside. This is
the continuity equation derived from the gauge invariance of the action. The calculation really
is nothing more than an application of Noether’s theorem from which we could have derived
directly the continuity equation, ∂µ j µ = 0.
In summary, the conservation of charge means invariance with respect to gauge transforma-
tions,
Ucl (χ) ≡ e−iqχ(r,t)/~ , (2.212)
where χ is the gauge field. We note that q and χ are conjugated observables. Therefore, if
[χ, Ĥ] = 0, then the charge q is a conserved quantity.
Transformations can be combined. For example, we believe that nowadays all laws are
invariant with respect to CPT transformation, that is, a combination of charge conjugation,
parity inversion, and θ transform.
18
R
Think about the argument, because ∂t |χρdt = 0!
70 CHAPTER 2. FOUNDATIONS OF QUANTUM MECHANICS
2.6 Exercises
2.6.1 Basic notions
2.6.1.1 Ex: Conservation of probability
Demonstrate the conservation of local probability through the definitions of probability densities,
ρ(r, t), and probability current j(r, t).
Using the stationary Schrödinger equation, calculate the eigenvalues and eigenvectors.
that we can describe it using the basis formed by |1i and |2i. On this basis the Hamiltonian Ĥ
of the system is given by,
E0 −E1
Ĥ = .
−E1 E0
a. If the system is initially in state |1i, will it remain in that state at a later time? How about
if the initial state is |2i?
b. Obtain the eigenvalues EI and EII and the respective eigenvectors |Ii and |IIi of Ĥ, express-
ing them in terms of |1i and |2i.
c. What is the probability of measuring an energy EI in the following state,
d. Based on the above result, we can predict at least one possible electromagnetic radiation
emission frequency for an ammonia sample. What is this frequency?
EUF 2014, 1 Semestre, Mecânica Quântica - Q1 1
/ )
)
)
)
)
/
)
|1 > |2>
Figure 2.2: The two states of the ammonium molecule.
(a) Não, já que estes estados não são autoestados do hamiltoniano. A evolução temporal do sistema
mistura os dois e temos duas combinações específicas com energia bem definida, os autoestados de
H. Começando em |1> ou |2> teremos a possibilidade de medir qualquer uma das duas energias
possíveis dependendo do tempo esperado.
~2 d2
Ĥ = − . (3.1)
2m dx2
Therefore, the general solution of the Schrödinger stationary equation,
is, q
ψ(x) = Aeikx + Be−ikx with k= 2mE
~2
. (3.3)
R∞ R∞
Note that the eikx functions are not quadratically integrable, since −∞ |eikx |2 dx = −∞ dx → ∞.
On the other side, they do not represent actual physical systems. In practice, we need to consider
wave packets or specify a finite volume for the particle. Note also that the spectrum of eigenvalues
is continuous. Do the Exc. 3.5.1.1.
This means that the wavefunction can not be infinite inside a finite volume. But it can be
infinite within an infinitely small volume. Also, since the Schrödinger equation contains the
second derivative by position, the wavefunction must be continuous and have a continuous
derivative.
75
76 CHAPTER 3. LINEAR MOTION / SEPARABLE POTENTIALS
This is the case, for example, for a rectangular well with Vx (x) = Vy (y) = Vz (z) = V0 /3 inside
the well and V (x, y, z) = 0 outside. It also holds for a harmonic potential,
m 2 2
V (r) = ωx x + ωy2 y 2 + ωz2 z 2 . (3.6)
2
In these cases, the following ansatz for the wavefunction is generally useful,
As the potential barriers are high, the walls are hard, that is, the particle, even being a quantum
particle, can not penetrate. The wavefunction and the possible energy values are,
r
2 nπx n2 ~2 π 2
ψ(x) = sin and En = . (3.11)
L L 2mL2
The Exc. 3.5.2.1 asks to demonstrate the result (3.11) illustrated in Fig. 3.1 1 .
Obviously the spectrum of eigenvalues is now discrete. They can be enumerated by an integer
n called quantum number. Note that the energy levels are not equidistant.
2 2
~ π
Example 14 (Localization energy ): There is a minimal energy E1 = 2mL 2 which is called
zero point energy or localization energy. This energy can be understood as a consequence of
Heisenberg’s uncertainty principle. We can make the following gross estimation of the zero
point energy. Obviously, the particle is localized with an uncertainty lower than ∆x < L.
Hence, ∆p > ~/∆x > ~/L. The average kinetic energy is,
40
35
30
25
E , ψ (μm)
20
15
10
0
0 0.5 1
x / L (μm)
Figure 3.1: (Code: QM M otion SquareW ell.m) Wavefunctions and energies in the rectangular
potential well.
Figure 3.3: Scheme of a two-sided (left) and one-sided (right) potential well.
For confined particles, E < 0, the problem is totally symmetric. In addition, the wavefunction
must disappear for x → ±∞. Therefore, we can simplify,
A1 = 0 = B 3 and A3 = B1 . (3.17)
k2
B1 eik1 L/2 = A2 e−ik2 L/2 + B2 eik2 L/2 = −A2 e−ik2 L/2 + B2 eik2 L/2 . (3.18)
k1
We now consider the quotient B2 /A2 . Using the right part of equation (3.18),
Since the amplitudes are real, the imaginary part of the quotient (3.19) should disappear, which
is the case when,
−0.2
E , En (V0)
−0.4
−0.6
−0.8
−1
−10 0 10
2κ1 k2
tan k2 L ,
−κ21 + κ22
Figure 3.4: (Code: QM M otion SquareF inite.m) Graphical solution for a finite bilateral po-
tential well. The red dotted curves represent the tangents (left side of the equation (3.21)),
the solid green curves the hyperbolas (right side of the equation), the circles in cyan are the
eigenenergies. When 0 < E − V0 E, they converge to the eigenenergies of the infinitely deep
well (black crosses and vertical black line).
At the bottom of deep potentials, that is, when 0 < E − V0 E, or equivalently, E ' V0 ,
we have k2 κ1 and hence, tan k2 L → 0 =⇒ k2 L = nπ. The energies are then,
~2 ~2 π 2 2
E − V0 = 2m = n . (3.22)
k22 2mL2
Apply the notions obtained in this section to solve Exc. 3.5.2.3.
propagating particle waves. From here on we will use the letter A (B) to denote the amplitudes
of waves propagating in direction ∞ (−∞).
In locations where the potential changes abruptly, the particle can be partially reflected.
with the scattering matrix T for a particle with energy E (see Fig. 3.2),
1 + kk12 ei(k1 −k2 )L 1 − kk21 ei(−k1 −k2 )L
T = 12 (3.26)
1 − kk21 ei(k1 +k2 )L 1 + kk21 ei(−k1 +k2 )L
−ik L !
k1 k1
1 e 2 0 1 + k 1 − k eik1 L 0
=2 2 2 .
0 eik2 L 1 − kk21 1 + kk21 0 e−ik1 L
If there are more zones with different depths, we may concatenate the scattering matrices.
Denoting by Tm→n the scattering matrix describing a transition at position Lm,n of a potential
of the depth Vm to another potential Vn , we write,
Multiplying the first line with T22 and the second with −T12 and adding them,
This equation resolved by A2 along with the second equation (3.29) resolved by B1 give,
A2 B2 T12 /T22 T11 − T12 T21 /T22 B2
=S = . (3.31)
B1 A1 1/T22 −T21 /T22 A1
3.3. POTENTIAL BARRIER 81
The matrix S describes the causality of scattering process more adequately: The amplitude
A2 in region (2) results from the superposition of a wave B2 being reflected by the barrier and
a wave A1 being transmitted by the barrier. The amplitude B1 in region (1) results from the
superposition of a wave A1 being reflected by the barrier and a wave B2 being transmitted by
the barrier. Therefore, the matrix S is more appropriate for the description of the quantum
reflection, as we will discuss in the next section. However, it has the disadvantage that it can
not be concatenated in the same way as the T matrices.
Unlike the T matrix the S matrix is unitary, since
T11
det S = S11 S22 − S12 S21 = − = −e2ik1 L . (3.32)
T22
Also, it is possible to show,
∗ ∗
S11 S21 S11 S12 1 0
S †S = ∗ ∗ = . (3.33)
S12 S22 S21 S22 0 1
The Exc. 3.5.3.1 asks to calculate the transmission and reflection of a particle at a potential
barrier.
We use B2 = 0, since no wave comes from the side of region (2), and A1 = 1, because it simplifies
the formulas and does not affect the generality of the results. The interesting results are:
• Even with E2 > V2 , the particle has a probability of being reflected at the step, |B1 | > 0.
Example 15 (Contrast of a partially reflected wave): Defining K± ≡ 21 max |ψ1 |2 ± min |ψ1 |2 ,
the contrast of the wavefunction in region (1) is given by K− /K+ . Writing the function as
ψ1 = eik1 x + B1 e−ik1 x it is easy to show, that
p p
K+ + K− − K+ − K− K−
|B1 | = p p ' . (3.36)
K+ + K− + K+ − K− 2K+
82 CHAPTER 3. LINEAR MOTION / SEPARABLE POTENTIALS
This formula can be understood as an analogue of Fresnel’s formula for matter waves 2 .
In Exc. 3.5.3.2 we calculate the behavior of a Broglie wave passing through a potential step
and entering a classically forbidden region. In Exc. 3.5.3.3 we investigate a model describing the
collision between attracting or repelling particles via a partial reflection at a potential step.
1 = |B1 |2 + k2
k1 |A2 |
2
=R+T , (3.39)
3 3
2.5 2.5
2 2
E/V0
E/V0
1.5 1.5
1 1 β = 10
β=3
0.5 0.5
0 0
0 0.5 1 0 0.5 1
R,T T
Figure 3.5: (Code: QM Motion Reflection.m) Left: Tunnel effect and quantum reflection at a
potential barrier. Right: Coefficients of transmission and reflection (horizontal) through the
shown potential barrier as a function of the energy normalized√to the height of the barrier E/V0 .
The dashed red curve corresponds to a low barrier, β ≡ ~1 L 2mV0 = 3, the blue solid curve
corresponds to a deep barrier β = 10.
its width while maintained the product of its width and its depth constant. Here, for simplicity,
we only consider a one-dimensional potential well, but the analysis can be expanded to more
dimensions.
The time-independent Schrödinger equation for the wavefunction ψ(x) of a particle in one
dimension is,
~2 d2 ψ
− (x) + αδ(x)ψ(x) = Eψ(x) , (3.41)
2m dx2
The potential is called a δ-potential well if α is negative and a δ-potential barrier if α is positive.
The potential splits the space in two parts (x < 0 and x > 0). In each of these parts the
potential energy is zero, and the Schrödinger equation reduces to,
d2 ψ 2mE
=− 2 ψ . (3.42)
dx2 ~
This is a linear differential equation with constant coefficients, whose solutions are linear
√ com-
binations of eikx and e−ikx , where the wavenumber k is related to the energy by k = 2mE ~ . In
general, due to the presence of the δ-potential in the origin, the coefficients of the solution need
not be the same in both half-spaces:
(
ψ1 (x) = A1 e−ikx + B1 eikx for x < 0
ψ(x) = , (3.43)
ψ2 (x) = A2 e−ikx + B2 eikx for x > 0
where, in the case of positive energies (real k), eikx represents a wave traveling to the right, and
e−ikx one traveling to the left. One obtains a relation between the coefficients by imposing that
the wavefunction be continuous at the origin,
A second relation can be found by studying the derivative of the wavefunction. Normally,
we could also impose differentiability at the origin, but this is not possible because of the δ-
potential. However, if we integrate the Schrödinger equation around x = 0 over an interval
[−, +]: Z + Z + Z +
~2
− ψ 00 (x)dx + V (x)ψ(x)dx = E ψ(x)dx . (3.45)
2m − − −
In the limit → 0, the right-hand side of this equation vanishes; the left-hand side becomes,
~2 0
− [ψ (0) − ψ10 (0)] + αψ(0) . (3.46)
2m 2
Substituting the definition of ψ into this expression, we obtain,
~2
− ik(−A1 + B1 + A2 − B2 ) + α(A1 + B1 ) = 0 . (3.47)
2m
The boundary conditions thus give the following restrictions on the coefficients,
A1 + B1 − A2 − B2 = 0
2mα . (3.48)
−A1 + B1 + A2 − B2 = (A1 + B1 )
ik~2
Figure 3.6: (a) The δ-potential (green) and the bound state wavefunction (blue). (b) Double
δ-potential (green).
from which follows that α must be negative, that is the bound state only exists for the well, and
not for the barrier. The Fourier transform of this wavefunction is a Lorentzian function. The
energy of the bound state is then,
~2 κ2 mα2
Eb = − =− 2 . (3.51)
2m 2~
The δ-potential well and its wavefunction are exhibited in Fig. 3.6(a).
3.3.6.2 Scattering
For positive energies, the particle is free to move in either half-space: x < 0 or x > 0, but it may
be scattered at the delta function potential. The quantum case can be studied in the following
situation: a particle incident on the barrier from the left side (B1 ). It may be reflected (A1 ) or
transmitted (B2 ). To find the amplitudes for reflection and transmission for incidence from the
left, we put in the above equations B1 = 1 (incoming particle), A1 = r (reflection), A2 = 0 (no
incoming particle from the right), and B2 = t (transmission), and solve for r and t even though
we do not have any equations in t. The result is,
1 1
t= , r= . (3.52)
mα i~2 k
1− 2 −1
i~ k mα
Due to the mirror symmetry of the model, the amplitudes for incidence from the right are the
same as those from the left. The result is that there is a non-zero probability,
1 1
R = |r|2 = = . (3.53)
~4 k 2 2~2 E
1+ 1+
m2 α 2 mα2
for the particle to be reflected. This does not depend on the sign of λ, that is, a barrier has
the same probability of reflecting the particle as a well. This is a significant difference from
classical mechanics, where the reflection probability would be 1 for the barrier (the particle
simply bounces back), and 0 for the well (the particle passes through the well undisturbed).
The probability for transmission is,
1 1
T = |t|2 = 1 − R = = . (3.54)
m2 α 2 mα2
1+ 4 2 1+ 2
~ k 2~ E
An application example regards the interfaces between two conducting materials. In the bulk
of the materials, the motion of the electrons is quasi-free and can be described by the kinetic
term in the above Hamiltonian with an effective mass m. Often, the surfaces of such materials
are covered with oxide layers or are not ideal for other reasons. This thin, non-conducting layer
may then be modeled by a local δ-function potential. Electrons may then tunnel from one
material to the other giving rise to a current.
Figure 3.7: (Code: QM M otion DeltaT unnelling.m) Transmission (red) and reflection (blue)
probability of a δ-potential well. The energy E > 0 is in units of Eb = mα2 /2~2 .
where 0 < R < ∞ is the ’internuclear’ distance with δ-function (negative) peaks located at
x = ±R/2 (shown in brown in the diagram). Keeping in mind the relationship of this model
with its three-dimensional molecular counterpart, we use atomic units and set ~ = m = 1.
Here 0 < λ < 1 is a formally adjustable parameter. From the single well case, we can infer
the ’ansatz’ for the solution to be:
which has two solutions d = d± . For the case of equal charges (symmetric homonuclear
case), λ = 1 and the pseudo-quadratic reduces to:
d± = q[1 ± e−d± R ] .
The ’+’ case corresponds to a wave function symmetric about the midpoint (shown in red
in the diagram) where A = B and is called gerade. Correspondingly, the ’-’ case is the
wavefunction that is anti-symmetric about the midpoint where A = −B is called ungerade
(shown in green in the diagram). They represent an approximation of the two lowest discrete
energy states of the three-dimensional H+2 and are useful in its analysis. Analytical solutions
for the energy eigenvalues for the case of symmetric charges are given by:
d± = q + W (±qRe−qR )/R ,
3.4. HARMONIC OSCILLATOR 87
where W is the standard Lambert function. Note that the lowest energy corresponds to
the symmetric solution d+ . In the case of unequal charges, and for that matter the three-
dimensional molecular problem, the solutions are given by a generalization of the Lambert
function (see section on generalization of Lambert function and references herein).
One of the most interesting cases is when qR ≤ 1, which results in d− = 0. Thus, one has a
non-trivial bound state solution with E = 0. For these specific parameters, there are many
interesting properties that occur, one of which is the unusual effect that the transmission
coefficient is unity at zero energy.
~2 d2 m
Ĥ = − 2
+ ω 2 x̂2 (3.56)
2m
" dx 2 ! ! #
r r r r
mω 1 mω 1
= ~ω x̂ − i p̂ x̂ + i p̂ + 12 = ~ω ↠â + 12 ,
2~ 2m~ω 2~ 2m~ω
p mω q
1
with the abbreviation â ≡ 2~ x̂ + i 2m~ω p̂ and its Hermitian transposition ↠. Now let’s try
to find out the properties of the operators ↠and â. First of all, the commutator is,
"r r r r #
mω 1 mω 1 i i
[â, ↠] = x̂ + i p̂, x̂ − i p̂ = [x̂ + p̂, x̂ − p̂] = [p̂, x̂] = 1 .
2~ 2m~ω 2~ 2m~ω 2~ ~
(3.57)
Knowing Ĥ|ψi = E|ψi is it clear that â â is an observable with the eigenvalue n ≡ ~ω − 12 ,
† E
↠â|ψi = E
~ω − 1
2 |ψi ≡ n|ψi =⇒ |ψi = |ni . (3.58)
Now, we show that the states â|ψi are eigenstates of the operator defined as n̂ ≡ ↠â, since,
↠ââ|ψi = (â↠− [â, ↠])â|ψi = (â↠â − â)|ψi = â(↠â − 1)|ψi = (n − 1)â|ψi (3.59)
=⇒ â|ψi ∝ |n − 1i
=⇒ n = hn|↠â|ni = C 2 hn − 1|n − 1i
√
=⇒ C = n .
88 CHAPTER 3. LINEAR MOTION / SEPARABLE POTENTIALS
We note that the quantum number of the new |n − 1i is decreased by 1. Similarly, we show for
the state ↠|ψi,
↠â↠|ψi = ↠([â, ↠] + ↠â)|ψi = ↠(1 + ↠â)|ψi = (n + 1)↠|ψi (3.60)
=⇒ ↠|ψi ∝ |n + 1i
=⇒ n + 1 = hn|↠â + [â, ↠]|ni = C 2 hn + 1|n + 1i
√
=⇒ C = n + 1 .
Therefore, this new state is also an eigenvector |n + 1i, with a quantum number increased by
one unit. ↠e â are creation and annihilation operators of an energy packet,
√ √
↠|ni = n + 1|n + 1i e â|ni = n|n − 1i . (3.61)
Now it is clear, that n̂ can be understood as a number operator 3 . The energy spectrum of the
harmonic oscillator is equidistant,
En = ~ω n + 21 . (3.63)
From the results of the previous item we obtain the uncertainty relation ∆x∆p for the OH
in the state |ni,
~
∆p∆x = (2n + 1) . (3.69)
2
Example 17 (Localization energy ): The non-vanishing energy of the fundamental state of
the harmonic oscillator, E0 = ~ω/2, is an immediate consequence of the Heisenberg principle
∆x∆p ≥ ~, because in analogy with Example 14 we calculate,
hp2 i ∆p2 ~2 ~2 ~ω
= > > = .
2m 2m 2m∆x2 2ma2ho 2
Now we start looking for asymptotic solutions. For x̃ → ±∞, that is, when the particle
enters the classically forbidden region, we can neglect the total energy of the particle,
d2 2
− 2 + x̃ ψ̃∞ (x̃) ' 0 . (3.71)
dx̃
2
The solution of this equation is ψ̃∞ (x̃) = Ce−x̃ /2 , since
d2 2 d 2 2 2 2 2 2
− 2 + x̃ e−x̃ /2 = − (−x̃)e−x̃ /2 +x̃2 e−x̃ /2 = −x̃2 e−x̃ /2 +e−x̃ /2 +x̃2 e−x̃ /2 = e−x̃ /2 ' 0 .
2
dx̃ dx̃
(3.72)
2
This motivates the ansatz ψ̃(x̃) = e−x̃ /2 H(x̃) for the complete differential equation (3.70),
2 2
d2 −x̃2 /2
2
−x̃2 /2 d H(x̃) de−x̃ /2 dH(x̃) d2 e−x̃ /2 2
− + x̃ 2
e H(x̃) = −e − 2 − H(x̃) + x̃2 e−x̃ /2 H(x̃)
dx̃2 dx̃2 dx̃ dx̃ dx̃2
(3.73)
2
2
d H(x̃) 2 dH(x̃) h 2 2
i 2
= −e−x̃ /2 2
− 2(−x)e−x̃ /2 + −x̃2 e−x̃ /2 + e−x̃ /2 H(x̃) + x̃2 e−x̃ /2 H(x̃)
dx̃ dx̃
2
≡ (2n + 1)e−x̃ /2 H(x̃) .
where the constant C is determined by the normalization condition, hψm |ψn i = δm,n . The
Hermite functions, Hn , are found in mathematical tables. Here we will only show the graphical
representation of |ψ|2 in Fig. 3.8. The Exc. 3.5.4.1 asks to evaluate OH in a classically forbidden
region and in Exc. 3.5.4.2 we will calculate the spectrum of a semi-harmonic OH.
7
5
E/h̄ωho , ψ(x)
0
−5 0 5
x/aho
Figure 3.8: (Code: QM M otion Harmonic.m) Wavefunctions and energies for a rectangular
well.
where PE is the probability density of finding the oscillating particle at position x. The proba-
bility of finding the particle in a range dx close to the location x is easily calculated,
We see that for high energy values the wavefunction approaches the classical expectation.
We already mentioned that there exist solutions only for certain energies En = ~ω(2n + 1).
Consequently, the energy levels are equidistant, En+1 − En = ~ω, as if there were a box into
E
n+3
n+2
n+1
n
which we add, one after the other, particles with the energy ~ω until we have accumulated
n portions of energy. These particles are called phonons in the case of vibrations of massive
particles, and photons in the case of a radiation field.
The fact that the energy distribution is the same as the one proposed by Planck for the black-
body radiation suggests the use of the harmonic oscillator to describe the second quantization.
we can write, X
e−iĤt/~ = |nie−iωt(n+1/2) hn| . (3.83)
n
P
If the initial state is |ψ(0)i = m cm |mi, the final state and the eigenvalue of any observable
will be,
X X
|ψ(t)i = |nie−iωt(n+1/2) hn|ψ(0)i = e−iωt(n+1/2) cn |ni (3.84)
n n
X X X
hψ(t)|Â|ψ(t)i = hm|eiωt(m+1/2) c∗m |Â| e−iωt(n+1/2) cn |ni = c∗m cn eiωt(m−n) hm|Â|ni .
m n m,n
that is, the state remains stationary. Motion needs non-diagonal elements of Â.
Another observation is that the populations do not change, even in the case of an initial
superposition, since,
We conclude that
• movement of an observable  is possible, but only due to variations of the phase factors;
• to carry out transitions between the vibrational states is necessary to perturb the oscillator,
e.g. by applying fields of electromagnetic radiation.
That is, the total energy of the oscillator is the sum of the energies of the states weighted
with the populations of those states. In the case of the position operator,
X
hψ(t)|x̂|ψ(t)i = a√ho
2
c∗m cn eiωt(m−n) hm|â + ↠|ni
m,n
X √ √
= a√ho
2
c∗n−1 cn e−iωt n + c∗n+1 cn eiωt n + 1
n
m,n→∞ √ X√
−→ aho 2 n|cn |2 cos ωt .
n
That is, the particle can only oscillate, if there are populations in consecutive states. If this
is not the case, hψ(t)|x̂|ψ(t)i = 0. The oscillation frequency is always ω, independent of
the energy of the particle. The Excs. 3.5.4.3 and 3.5.4.4 analyze the temporal evolution of
oscillators subject to a sudden perturbation.
we can separate the spatial directions and obtain a one-dimensional equation for each coordinate,
such that the coordinates can be considered separately. Each function ψk (xk ) is of the form (3.78)
and the energies are,
Ek = ~ωk (nk + 21 ) , (3.89)
where k = x, y, z.
3.4. HARMONIC OSCILLATOR 93
which holds when  and B̂ commute with their commutator, i.e., [Â, [Â, B̂]] = [B̂, [Â, B̂]] = 0.
Ĝ0 (τ ) = (Â + B̂)eτ (Â+B̂) e−τ B̂ e−τ Â − eτ (Â+B̂) B̂e−τ B̂ e−τ Â − eτ (Â+B̂) e−τ B̂ Âe−τ Â
h i h i
= eτ (Â+B̂) Âe−τ B̂ − e−τ B̂ Â e−τ Â = eτ (Â+B̂) Â − e−τ B̂ Âeτ B̂ e−τ B̂ e−τ Â
h i
= eτ (Â+B̂) Â − Â + [−τ B̂, Â] + 2! 1
[−τ B̂, [−τ B̂, Â]] + .. e−τ B̂ e−τ Â ,
using the formula (2.162). If now [Â, [Â, B̂]] = 0 = [B̂, [Â, B̂]], then,
Ĝ0 (τ ) = eτ (Â+B̂) τ [B̂, Â]e−τ B̂ e−τ Â = −τ [Â, B̂]eτ (Â+B̂) e−τ B̂ e−τ Â = −τ [Â, B̂]Ĝ(τ ) .
The state resulting from the action of the operator D̂(α) onto the estado fundamental of the
HO is,
∞
X
2 /2 † ∗ 2 /2 (α↠)n
|αi ≡ D̂(α)|0i = e−|α| eαâ e−α â |0i = e−|α||0i (3.94)
n!
n=0
−|α|2 /2 † (α↠)2 −|α|2 /2 α√ α2 √
=e 1 + αâ + + .. |0i = e |0i + 1|1i + 2!|2i + .. , (3.95)
2! 1! 2!
that is, the state |αi is a superposition distributed according to the Poisson distribution,
X∞
2 /2 αn
|αi = e−|α| √ |ni . (3.96)
n=0 n!
X∞ X∞ X∞
−|α|2 /2 αn −|α|2 /2 αn √ −|α|2 /2 αn
â|αi = e √ â|ni = e √ n|n − 1i = e p |n − 1i ,
n=0 n! n=0 n! n=0 (n − 1)!
(3.97)
that is,
â|αi = α|αi . (3.98)
The state |αi is called coherent state or Glauber state 5,6 . We note that, in spite of its appearance,
the equation (3.98) is not an eigenvalue equation, since â is not observable.
Using the formula (2.162), we verify immediately,
∂ ∂
↠− α∗ = e ↠− α = .
∂α ∂ α∗
−
→ ←−
With the bosonic operators we can construct the observables,
√
x̂ + ip̂ = 2â , ∂x − i∂p = 2∂α , ∂x + i∂p = 2∂α∗ .
−
→ ←−
−
In two dimensions [87], d2 α = d(Re α)d(Im α) = dxdp.
3.4. HARMONIC OSCILLATOR 95
Figure 3.10: Illustration of the Glauber states. Here, x̂1 ≡ â + ↠e x̂2 ≡ i(â − ↠).
3.5 Exercises
3.5.1 Translational motion
3.5.1.1 Ex: Trapped particle
Consider the problem of a particle of mass m forced to move in a single direction and completely
confined to a box, with walls placed at the positions x = 0 and x = a. a. The particle be in the
ground state, what is its energy and its wavefunction?
b. Suppose the particle has the following wavefunction:
π π
1 2π
ψI (x) = √ 2 cos (6x − a) − 3i sin x + cos (2x − a) ,
7a 2a a 2a
~ 2 2
what is the probability that a measurement of the energy and yields the result E = 2π ma2
?
c. Considering again the state of item (a) (the ground state), what is the probability distribution
for the momentum of this particle?
d. Still starting from the ground state, suppose we remove (instantaneously) the walls, leaving
the particle free (Ĥ = p̂2 /2m). What is the energy of this free particle?
Formulae:
Z L nπx nπL[1 − (−1)n eiBL ]
eiBx sin dx = para n = 1, 2, 3, ...
0 L n2 π 2 − B 2 L2
Z ∞
x2 2 πx π2
2 2
cos 2 dx =
−∞ (1 − x ) 4
~2 00
− ψ (x) + αδ(x)ψ(x) = Eψ(x) .
2m
The energy spectrum may be a discrete spectrum of bound states and a continuum of free states.
a. Calculate the transmission coefficient for the case of a particle with energy E thrown against
the potential energy barrier V (x) = αδ(x). Does the result change for the case when V (x) =
−αδ(x), with α > 0?
b. For this last potential, find the energy of the bound state and its corresponding wavefunction.
98 CHAPTER 3. LINEAR MOTION / SEPARABLE POTENTIALS
eE
a. With the abbreviation b ≡ mω 2
we can rewrite the Hamiltonian
p2 m m
Ĥ (1) = + ω 2 (x − b)2 − ω 2 b2 ,
2m 2 2
3.5. EXERCISES 99
c. The temporal evolution is given by the time-dependent Schrödinger equation. Since the jump
is finite, the solution must be well behaved in time t = 0,
( (1)
(1) (1) (1)
(1) e−(i/~)Ĥ t ψn (x) = e−(i/~)En t ψn (x) para t < 0
ψn (x, t) = (1) .
e−(i/~)Ĥt ψn (x) para t ≥ 0
For times t ≥ 0 we write,
† †
ψn(1) (x, t) = e−(i/~)Ht e−β(â−â ) ψn (x) = e−(i/~)Ĥt e−β(â−â ) e(i/~)Ĥt e−(i/~)Ĥt ψn (x) .
which is easy to show by expansion of eB̂ . The expression e− B̂e can be evaluated through its
action on the complete system of eigenfunctions,
(0) (0) (0)
e−(i/~)Ĥt âe(i/~)Ĥt |ψn i = e−(i/~)Ĥt âe(i/~)En t |ψn i = e−(i/~)En−1 t e(i/~)En t â|ψn i .
Hence,
e−(i/~)Ĥt âe(i/~)Ĥt = âeiωt e e−(i/~)Ĥt ↠e(i/~)Ĥt = ↠e−iωt .
Now we can write the temporal solution,
iωt −↠e−iωt ) † )−iβ (0)
sin ωt(â+↠) −(i/~)En t
ψn(1) (x, t) = e−β(âe e−(i/~)Ĥt ψn (x) = e−β cos ωt(â−â e ψn (x) .
Finally,
|ψn(1) (x, t)|2 = |ψn (x − x̄(t))|2 ,
(1)
where x̄(t) ≡ b cos ωt. This means that the spatial distribution of ψn around x̄(t) is the same
as of ψn around x̄ = 0. The entire distribution oscillates without deformation. The momentum
distribution follows from the Fourier transform,
Z Z
1 −(i/~)px (1) 1
(1)
φn (p, t) = √ dxe ψn = √ due−(i/~)pu e−(i/~)umωb sin ωt eiγ(p,t) ψn
2π~ 2π~
= eiγ(p,t) φn (p + mωb sin ωt) ,
with γ = γ ∗ . We obtain,
|φ(1) 2 2
n (p, t)| = |φn (p − p̄(t))| ,
d
ĤΞ(t, r1 , r2 ) = i~ Ξ(t, r1 , r2 ) , (4.2)
dt
becomes stationary,
−~2 2 −~2 2
∇ + ∇ + V (r1 − r2 ) Ξ(r1 , r2 ) = Etot Ξ(r1 , r2 ) . (4.3)
2m1 r1 2m2 r2
Now we transform into the center-of-mass system making for the total wavefunction the ansatz
Ξ(r1 , r2 ) = e−iP·R/~ Ψ(r) with R ≡ m m2
M r1 + M r2 and r ≡ r1 −r2 and introducing the abbreviation
1
−~2 2 −iP·R/~
∇ e Ψ(r) (4.4)
2m1 r1
−~2 h −iP·R/~ 2 −iP·R/~ 2 −iP·R/~
i
= e ∇r1 Ψ(r) + 2(− im
~M
1P
)e ∇ r1 Ψ(r) + Ψ(r)(− im1
~2 M
P) e
2m1
2
−~ 2 i~P m 1 P2
= e−iP·R/~ ∇r1 Ψ(r) + ∇r1 Ψ(r) − Ψ(r) .
2m1 M 2M 2
101
102 CHAPTER 4. ROTATIONS / CENTRAL POTENTIALS
P2 −~2 2 −~2 2
Ψ(r) + ∇r Ψ(r) + ∇ Ψ(r) + V (r)Ψ(r) = Etot Ψ(r) . (4.6)
2M 2m1 2m2 r
P2
Subtracting the energy of the center-of-mass motion with E = Etot − 2M and introducing the
abbreviation m−1 = m−1 −1
1 + m2 , we finally get,
−~2 2
∇ + V (r) Ψ(r) = EΨ(r) . (4.7)
2m r
∂2 ∂2 ∂2
∇2r = 2
+ 2+ 2 . (4.8)
∂x ∂y ∂z
However, in some situations, the symmetry of the system allows to reduce dimensionality sim-
ilarly to the cases of the box potential and the three-dimensional harmonic oscillator. Let us
now discuss the cases of cylindrical and spherical symmetry.
Electrons in magnetic fields are subject to the Lorentz force, which keeps them in a rotating
motion. We can rewrite the momentum operator in cylindrical coordinates,
as
∂2 1 ∂ 1 ∂2 ∂2
∇2r = + + + . (4.10)
∂ρ2 ρ ∂ρ ρ2 ∂ϕ2 ∂z 2
Now, with the assumption that the potential only depends on ρ, we can try ansatz Ψ(r) =
R(ρ)ξ(ϕ)ζ(z),
2
1 ~2 ∂ 1 ∂ ~2 1 ∂ 2 ~2 1 ∂2
− + + V (ρ) R(ρ)− ζ(z)− ξ(ϕ) = E . (4.11)
R(ρ) 2m ∂ρ2 ρ ∂ρ 2m ζ(z) ∂z 2 2mρ2 ξ(ϕ) ∂ϕ2
First, we separate the axial motion,
ζ 00 2mEz
− = const ≡ , (4.12)
ζ ~2
the solution of this equation being a superposition of two plane waves counterpropagating along
the axis z, ζ(z) = Aeikz z + Be−ikz z . Now, we separate the azimuthal motion,
The solution of the right-hand part of the equation is ξ(ϕ) = Ceimϕ ϕ + De−imϕ ϕ . Finally, we
have the radial equation,
1 ∂R(ρ)2 1 ∂R(ρ) 2m 2
m2ϕ
+ − [E − V (ρ)] − k z − =0, (4.14)
R(ρ) ∂ρ2 ρR(ρ) ∂ρ ~2 ρ2
~2 m2
with the effective potential Vef f = V (ρ) + 2mρϕ2 . For a homogeneous potential, V (ρ) = V0 , the
solution will be a superposition of Bessel functions.
Example 20 (Rigid rotor in cylindrical coordinates): To give an example, we disregard
the potential, V (ρ) = 0, and we consider for the particle an orbit with constant radius,
ρ = const such that R0 (ρ) = 0. In this case, we only need to treat the orbital motion
described by the right part of Eq. (4.13). For the solution of this equation, ξ(ϕ) = Aeimϕ ϕ ,
to be well-defined, we need ξ(ϕ) = ξ(ϕ + 2π). This implies,
mϕ = 0, ±1, ±2, ..
and
~2 m2ϕ
Eϕ = .
2mρ2
The allowed energies Emϕ = Eϕ can be obtained by letting the Hamiltonian
~2 ∂ 2
Ĥ = − ,
2I ∂ϕ2
with the moment of inertia I = mρ2 actuate on the azimuthal wavefunction ξ(ϕ). We now
define the operator,
ˆlz = ~ ∂ .
i ∂ϕ
This operator acts on the wavefunction ξ as follows,
ˆlz ξ(ϕ) = ~mϕ ξ(ϕ) .
as
1 ∂ 2 ∂ 1 L̂2 L̂2 1 ∂ ∂ 1 ∂2
∇2r = 2 r + 2 2 where ≡ sin ϑ + , (4.16)
r ∂r ∂r r ~ ~ 2 sin ϑ ∂ϑ ∂ϑ sin2 ϑ ∂ϕ2
is an abbreviation called Legendre operator. For an isotropic potential, V (r) = V (r), we can try
the ansatz,
Ψ(r) = R(r)Y (ϑ, ϕ) (4.17)
to solve the Schrödinger equation (2.54),
r2 ~2 1 ∂ 2 ∂ −1 L̂2 Y (ϑ, ϕ) ~2
− r + V (r) − E R(r) = = const ≡ − `(`+1) , (4.18)
R(r) 2m r2 ∂r ∂r 2m Y (ϑ, ϕ) 2m
104 CHAPTER 4. ROTATIONS / CENTRAL POTENTIALS
where we choose a separation constant, `(` + 1), the significance of which we shall soon learn.
Considering only the angular part,
we obtain,
2 1 1 ∂ ∂ 1 ∂2
sin ϑ sin ϑ Θ(ϑ) + `(` + 1) = − Φ(ϕ) = const ≡ m2 , (4.21)
Θ(ϑ) sin ϑ ∂ϑ ∂ϑ Φ(ϕ) ∂ϕ2
As in the case of the cylindrical potential, the solution of the azimuthal equation is, using the
normalization,
Φ(ϕ) = √12π eimϕ , (4.24)
with the magnetic quantum number m = 0, ±1, ±2, ...
The polar equation,
1 1 ∂ ∂ m2
sin ϑ Θ(ϑ) + `(` + 1) = , (4.25)
Θ(ϑ) sin ϑ ∂ϑ ∂ϑ sin2 ϑ
is called Legendre’s differential equation and can be solved by a power series in cosk ϑ. For
m = 0, the solutions are the Legendre polynomials, P` (cos ϑ) with
1 d`
P` (z) = [(z 2 − 1)` ] . (4.26)
2` `! dz `
The first polynomials are,
dm (−1)m 2 m/2 d
`+m
P`m (z) = (−1)m (1 − z 2 )m/2 P ` (z) = (1 − z ) [(z 2 − 1)` ] (4.28)
dz m 2` `! dz `+m
(` − m)! m
P`−m (z) = (−1)m P (z) .
(` + m)! `
The polar function must still be normalized,
s
2` + 1 (` − m)!
Θm m
` (ϑ) = P` (cos ϑ) . (4.29)
2 (` + m)!
4.1. PARTICLE IN A CENTRAL POTENTIAL 105
l=0 l=1
90 1 90 1
120 60 120 60
180 0 180 0
l=2 l=3
90 1 90 1
120 60 120 60
180 0 180 0
Figure 4.1: (Code: QM Rotation Legendre.m) Angular wavefunctions. Shown are the Legendre
polynomials Plm (cos ϑ) for ` = 0, 1, 2, 3 and m = 0, .., `. Red: m = 0, green: |m| = 1, blue:
|m| = 2, and magenta: |m| = 3.
The functions Y`m (ϑ, ϕ) are the spherical harmonics. They form an orthonormal system,
Z π Z 2π
Y`∗0 m0 (ϑ, ϕ)Y`m (ϑ, ϕ) sin ϑdϑdϕ = δ`0 ` δm0 m . (4.30)
0 0
Finite solutions only exist when the angular momentum quantum number is ` = 0, 1, .. and for
|m| ≤ `.
The solutions of the angular part of the Schrödinger equation for the hydrogen atom are
finally,
s
1 m 2` + 1 (` − m)! imϕ
Y`m (ϑ, ϕ) = √ P` (cos ϑ) e . (4.31)
2π 2 (` + m)!
The spherical harmonics are simultaneously eigenfunctions of the operators L2 , as can be seen
from Eq. (4.19), and of the operator Lz according to Eq. (4.23). The quantities represented by
the quantum operators Ĥ, L̂2 , L̂z are conserved in the hydrogen system. The conservation of
the angular momentum is due to the spherical symmetry of the Coulomb potential.
We will verify the parity of the spherical harmonics in Exc. 4.5.1.1.
Now, we make the substitution R(r) = u(r)/r and the radial equation becomes,
~2 ∂ 2 L2
− + + V (r) u(r) = Eu(r) . (4.33)
2m ∂r2 2mr2
This equation is very similar to a one-dimensional Schrödinger equation, but there is an addi-
tional potential term called centrifugal potential,
L̂2
V` (r) ≡ . (4.34)
2mr2
For example, for the potential of an electron orbiting a proton, we have,
~2 ∂ 2 Ze2 ~2 `(` + 1)
− − + − E uE` (r) = 0 . (4.35)
2m ∂r2 4π0 r 2mr2
We will discuss this equation intensely in the context of the hydrogen atom.
V(r)
0 5 10
r/aB
Figure 4.2: (Code: QM Rotation Centrif ugal.m) Sum of a Coulomb potential and centrifugal
potential for ` = 0 (lower curve), ` = 1 (center curve), and ` = 2 (upper curve).
In Exc. 4.5.1.2 we derive the radial Gross-Pitaevskii equation for a Bose-Einstein condensate
trapped in a spherical potential. In the Exc. 4.5.1.3 we will study particles inside a central
potential of zero depth, in the Excs. 4.5.1.4 and 4.5.1.5 we consider 3D spherical box potentials
and in Exc. 4.5.1.6 a spherical harmonic potential.
~2 `(` + 1)
E` = ,
2I
with the momentum of inertia I = mR2 .
cloud, we treat it as an entity with mass M and charge Ze, where Z is the number of protons
and corresponds to the order of the element in the periodic system.
The canonical procedure for calculating all properties of an atom is to establish its Hamil-
tonian, that is, to determine the kinetic energies of all components and all interaction energies
between them, and to solve the Schrödinger equation. For each component we write the kinetic
energy,
XZ
P2 p2i
Tncl = and Tele = . (4.36)
2M 2m
i=1
Here, (R, P) are the nuclear coordinates and (ri , pi ) those of the electrons. The energies that
corresponds to the interactions, that is, Coulombian attraction or repulsion, between the com-
ponents of the atom are,
Z
X Z
X
Ze2 e2
Vncl−ele = − and Vele−ele = . (4.37)
4πε0 |R − ri | 4πε0 |ri − rj |
i=1 i6=j=1
There are also interactions due to the spin of the particles, which we will deal with later.
Obviously, the solution to this many-body problem is very complicated. For this reason, we
will in this chapter, based on the Schrödinger equation, calculate the complete spectrum of the
simplest possible atom, hydrogen. This atom consists of a proton and an electron, only.
e-
Ze -
(Z-1)e-
Figure 4.3: The hydrogen model applies to other atoms having a single valence electron occupying
a sufficiently large space, that it sees the nucleus together with rest of the electrons shielding
the nucleus as a single positive charge.
1 Ze2 1 Z 2 ~2 1 Z 2 e2 1 Z2
En = − =− = − = − 13.6 eV , (4.38)
2 4πε0 rn 2ma2B n2 4πε0 2aB n2 n2
Paschen
Brackett
Ballmer
Lyman
abgd
hydrogen spectrum (En − Em )/~ are the Lyman (m = 1), the Balmer (m = 2), the Paschen
(m = 3) an the Brackett series (m = 4).
The discussion of the hydrogen atom within quantum mechanics can start from the radial
Schrödinger equation (4.35) with the Coulomb attraction potential,
~2 ∂ 2 Ze2 ~2 `(` + 1)
− − + − E uE` (r) = 0 . (4.40)
2m ∂r2 4πε0 r 2mr2
In order to facilitate comparison with Bohr’s classical model, let us express the energy in terms
of Bohr’s energy, E ≡ En = E1 /n2 , and write the radius in units of aB , that is, r̃ ≡ Zr/aB .
This yields,
00 `(` + 1) 2 1
un,` (r̃) + − + − 2 un,` (r̃) = 0 . (4.41)
r̃2 r̃ n
To ensure that for large radii, r → ∞, the solution is finite, we need an asymptotic behavior
like un,` (r̃ → ∞) = e−r̃/n . To ensure that for small radii, r → 0, the solution is finite, we
need un,` (r̃ → 0) = r̃`+1 . We derive the asymptotic solutions in Exc. 4.5.2.1. The resulting
differential equation only has solutions for an integer and positive main quantum number n
and when ` = 0, 1, .., n − 1. That is, in the relation E = E1 /n2 the parameter n is integer
and positive, such that energy levels remain degenerate in ` and m. This means that Bohr’s
postulate of discrete (i.e. quantized) energy levels is valid (uff!)
Substituting the ansatz,
un` (r̃) = Dn` r̃`+1 e−r̃/n L(r̃) , (4.42)
it’s easy to show (see Exc. 4.5.2.2), that the differential equation (4.41) reduces to,
r̃L00 (r̃) + 2 (` + 1) − n1 r̃ L0 (r̃) + 2 1 − n1 (` + 1) L(r̃) = 0 . (4.43)
(2`+1)
The solutions of this differential equation, Ln−`−1 (r̃), are the Laguerre polynomials. These
polynomials are listed in mathematical tables. Using the properties of these polynomials it is
possible to show that the radial functions are orthogonal and can be normalized (see Exc. 4.5.2.3).
Fig. 4.6 shows the curves for the lowest orbitals.
Finally, we can write the total solutions,
un,` (r) ~2 Z 2
ψn,`,m (r, θ, φ) = Y`,m (θ, φ) with En = − , (4.46)
r 2ma2B n2
n−1
X
(2` + 1) = n2 (4.47)
`=0
times degenerate.
1
Laguerre’s associated differential equation is,
eρ ρ−α dν
Lν(α) (ρ) = e−ρ ρν+α .
α! dρν
110 CHAPTER 4. ROTATIONS / CENTRAL POTENTIALS
0.6 0.05
0.5 〈 r10 〉
0.04
0.4
2
aB3/2 R(r)
0.03
a [rR(r)]
0.3
0.2
B
0.02 〈 r20 〉
0.1
〈 r30 〉
0.01
0
−0.1 0
0 5 10 15 20 25 0 5 10 15 20 25
r/aB r/aB
0.12 0.025
〈 r32 〉
0.1
0.02
0.08
2
aB3/2 R(r)
0.015
aB [rR(r)]
0.06
〈 r31 〉
0.04 0.01
〈 r30 〉
0.02
0.005
0
−0.02 0
0 5 10 15 20 25 0 5 10 15 20 25
r/aB r/aB
Figure 4.6: (Code: QM Rotation Laguerre.m) Radial wavefunctions, left R; right u. Top for
(n, `) = (1..3, 0); bottom (n, `) = (3, 0..2).
3/2
ψ100 = √1
π
Z
aB e−r̃ (4.48)
3/2
ψ200 = √1
4 2π
Z
aB (2 − r̃)e−r̃/2
3/2
ψ210 = √1
4 2π
Z
aB r̃e−r̃/2 cos θ
3/2
ψ21±1 = √1
8 2π
Z
aB r̃e−r̃/2 sin θe±iϕ
3/2
ψ300 = √1
81 3π
Z
aB (27 − 18r̃ + 2r̃2 )e−r̃
√ 3/2
ψ31±1 = √2
81 3π
Z
aB (6 − r̃)r̃e−r̃/3 sin θe±iϕ
3/2
ψ320 = 1
√
81 6π
Z
aB r̃2 e−r̃/3 (3 cos2 θ − 1) ,
where we use the abbreviation r̃ ≡ Zr/aB . Using these wavefunctions we can now calculate
4.2. QUANTUM TREATMENT OF HYDROGEN 111
In Exc. 4.5.2.5 we calculate the eigenvalues hr−1 i and hp2 i and we verify the virial theorem.
Finally, in Exc. 4.5.2.6 we calculate transition matrix elements between different orbitals.
To better understand the properties of the angular momentum operator in quantum mechanics
we will derive in the Excs. 4.5.3.1 and 4.5.3.2 some of its properties.
We also found the common eigenvalues and eigenfunctions of operators l̂2 and ˆlz [see Eqs. (4.19)
and (4.23)]. We now use the notation |`, mi ≡ Y`m (θ, φ) for the eigenfunctions,
[Ĥ, ˆlz ]|ψi = Ĥ~m|ψi − ˆlz E|ψi = 0 and [Ĥ, l̂2 ]|ψi = Ĥ~2 `(` + 1)|ψi − l̂2 E|ψi = 0 . (4.57)
Exc. 4.5.3.3 asks to show explicitly, at the example of an isotropic three-dimensional harmonic
oscillator, that l̂2 and ˆlz are constants of motion.
4.3. ANGULAR MOMENTUM 113
or [ĵm , ĵn ] = i~kmn jk using the Levi-Civita symbol, we obtain a consistent algebra.
Since ĵ2 and ĵz commute (we show this from Eq. (4.59) in Exc. 4.5.3.4), they have common
eigenfunctions |j, mi. We can write the eigenvalues as,
where, for now, we only know that m is real and j ≥ 0. But since hj, m|ĵ2 |j, mi ≥ hj, m|ĵz2 |j, mi,
it is clear that j(j + 1) ≥ m2 .
[ĵz , ĵ± ] = ±~ĵ± and [ĵ2 , ĵ± ] = 0 and ĵ∓ ĵ± = ĵ2 − ĵz2 ∓ ~ĵz . (4.61)
ĵz ĵ± |j, mi = ([ĵz , ĵ± ] + ĵ± jz )|j, mi = ~(m ± 1)ĵ± |j, mi (4.62)
ĵ2 j± |j, mi = ĵ± ĵ2 |j, mi = ~2 j(j + 1)ĵ± |j, mi .
That is, ĵ± |j, mi is a eigenstate of ĵ2 and ĵz with the eigenvalues j and m ± 1, respectively, if
j± |j, mi =
6 0. Hence,
ĵ+ |j, mi ∝ |j, m + 1i . (4.63)
In order not to violate the condition m2 ≤ j(j + 1), we need to fix ĵ± |j, ±ji = 0. Therefore, for
a specified j, the m can have only one of the 2j + 1 possible values m = −j, −j + 1, .., j. Since
2j + 1 is an integer, j can only have values j = 0, 12 , 1, 32 , ... Thus, the eigenvalue equation of the
observables ĵ2 , ĵ is solved since we could have chosen instead of ĵz any one of the components of
ĵ, knowing that the others do not commute with the chosen one.
All spin components ĵz and the scalar ĵ2 can only have discrete eigenvalues. The smallest
unit is ~/2. With the normalization hj, m|j 0 , m0 i = δj,j 0 δm,m0 we have,
hj, m|ĵ∓ ĵ± |j, mi = hj, m|(ĵ2 − ĵz2 ∓ ~ĵz )|j, mi = ~2 [j(j + 1) − m(m ± 1)] , (4.64)
and p
ĵ± |j, mi = ~ j(j + 1) − m(m ± 1)|j, m ± 1i . (4.65)
In Exc. 4.5.3.5 we calculate the uncertainty of the angular momentum components, in Exc. 4.5.3.6
we write the operator ĵx in a matrix form, and in Exc. 4.5.3.7 we calculate projections of the
spin of the electron in different directions of the quantization axis.
114 CHAPTER 4. ROTATIONS / CENTRAL POTENTIALS
and
The Pauli matrices act on the spin of the individual electrons a and b. They can be extended
4.4. COUPLING OF ANGULAR MOMENTA 115
With these operators we can now build other operators. We first consider the three components
of the total angular momentum,
0 1 1 0 0 −1 −1 0 1 0 0 0
1 0 0 1 1 0 0 −1 0 0 0 0
Ŝx = ~2
1
, Ŝy = i~ , Ŝz = ~ .
0 0 1 2 1 0 0 −1 0 0 0 0
0 1 1 0 0 1 1 0 0 0 0 −1
The operator for the square of the absolute value of the total angular momentum is calculated
as follows:
2 0 0 0
0 1 1 0
Ŝ2 = Ŝx2 + Ŝy2 + Ŝz2 = ~2
0 1 1
. (4.71)
0
0 0 0 2
Now we look for the eigenvalues of the total angular momentum. The equation for the eigenvalues
of Sz ,
Ŝz |γk i = MS |γk i , (4.72)
is already diagonal in the introduced basis {γk } with the eigenvalues,
MS = ~, 0, 0, −~ . (4.73)
For Ŝ2 the situation is more interesting: The states |γ1 i and |γ4 i are eigenstates of S2 for the
eigenvalue 2~2 , but the states |γ2 i and |γ3 i are not eigenstates. On the other hand, we know
that the linear combination of two eigenstates with the same eigenvalue is also a eigenstate.
Therefore, the states
are eigenstates of Ŝz , but they also are eigenstates of Ŝ2 , since we can easily verify,
Ŝ2 |γs i = 2~2 |γs i and Ŝ2 |γa i = 0~2 |γa i , (4.75)
using the matrices (4.70). In summary, for the eigenvalue hŜ2 i = 2~2 there exist the following
three states:
|γ1 i ms = 1
|γ4 i ms = −1 triplet , S = 1 (4.76)
|γs i ms = 0
116 CHAPTER 4. ROTATIONS / CENTRAL POTENTIALS
By exchanging the two electrons, the vectors |γ1 i and |γ4 i retain their shape, while the mixed
vectors change their shape: γ2 ←→ γ3 . Under particle exchange |γa i reverses its sign, that is, it
is antisymmetric, while |γ1 i, |γ4 i and |γc i conserve their signs, thas is, they are symmetrical.
In summary, the triplet states have the quantum number of the total angular momentum
(with the expected value for Ŝ2 of ~2 S(S + 1) = 2~2 ), and they are symmetrical about the
exchange of particles. The singlet state has the quantum number of the total angular momentum
S = 0, and it is antisymmetric about the exchange of particles.
A similar treatment can be done with bosons, as will be discussed in Sec. 9.1.
In this case, the perturbation Hamiltonian does not commute with the total angular momentum,
Another type of perturbation affects both spin symmetrically (e.g., superradiance Dicke with
two atoms in the same radiative mode or two counterpropagating modes in a ring cavity). The
interaction Hamiltonian is now the sum of the individual perturbations,
Ω Ω
Ω∗ Ω
Ĥ = Ĥ1 ⊗ I + I ⊗ Ĥ1 =
Ω∗
. (4.81)
Ω
Ω∗ Ω∗
S now it’s a good quantum number. Singlet states do not couple with triplets. This is the idea
behind Dicke’s superradiance. The absolute value of the total angular momentum is conserved.
The quantum number S is called Dicke cooperatividade [74].
In Sec. ?? we will discuss the coupling of two counterpropagating modes in a ring cavity. In
Sec. ?? we will discuss Dicke states.
4.4. COUPLING OF ANGULAR MOMENTA 117
Ĥint = −A(r1 ) · (êx σ̂1x ⊗ I + êy σ̂1y ⊗ I) − A(r2 ) · (êx I ⊗ σ̂2x + êy I ⊗ σ̂2y ) , (4.84)
with r1 ' r2 . By the rules (4.69) we find in matrix notation a Hamiltonian equivalent to (4.81),
0 Ax − iAy Ax − iAy 0
Ax + iAy 0 0 Ax − iAy
Ĥ =
Ax + iAy
. (4.85)
0 0 Ax − iAy
0 Ax + iAy Ax + iAy 0
±
Note, that the Hamiltonian is Hermitian, although it consists of sums of operators σ̂1,2 . Thus,
the coupling is inherently different from a beamsplitter-type couplings generated by terms ex-
pressions like σ̂1+ σ̂2− .
Using MAPLE we find the eigenvector and eigenvalue matrices,
q
Ax −iAy A −iA Ax −iAy
Ax +iAy − Axx +iAyy 0 −2 A2x + A2y 0 0 0
q qAx +iAy
− Ax −iAy 0 −1
Ax −iAy
0 0 0 0
U= q Ax +iAy q Ax +iAy and Ê =
.
− Ax −iAy Ax −iAy
0 0 0 q 0
Ax +iAy 0 1 Ax +iAy
1 1 0 1 0 0 0 2 A2x + A2y
(4.86)
Now we get to the evolution of the various states,
q q
−A +iA
− Axx+iAyy sin 2t A2x + A2y
0 q 0 0
2 + A2
cos 2t A 1 1
iÊt −1 1 x y
Ue U =
q
and U eiÊt U −1
−1 = −1 .
1 cos 2t A2x + A2y
0 q q 0 0
Ax +iAy 2 + A2
−Ax +iAy sin 2t A x y
(4.87)
For 2J atoms,
p
hJ, M − 1|σ − |J, M i = J(J + 1) − M (M + 1) . (4.88)
For 2J = 2 atoms,
I = |h1, M − 1|σ − |1, M i|2 = 2 − M (M + 1) . (4.89)
118 CHAPTER 4. ROTATIONS / CENTRAL POTENTIALS
Obviously, for M = 1 the transition is not possible, whereas for M = −1, 0 the I rate is doubled.
J is called cooperation number,
| + +i = |1, 1i (4.90)
| + −i = |1, 0i
| − +i = |1, 0i
| − −i = |1, −1i .
The angular momenta of two particles or two angular momenta of different origins in a single
particle represent independent degrees of freedom, [j1 , j2 ] = 0. Without interaction between
angular momenta the Hilbert spaces are orthogonal:
H1 0
H1 ⊗ H 2 = . (4.92)
0 H2
That is, there is a complete set of commuting operators {ĵ21 , ĵ1z , ĵ22 , ĵ2z }. Therefore, we can
specify quantum numbers j1 , j2 , mj1 , and mj2 simultaneously. On the other hand, the group
4.4. COUPLING OF ANGULAR MOMENTA 119
{ĵ21 , ĵ22 , ĵ2 , ĵz } also represents a complete set of commuting operators, as shown in Exc. 4.5.4.2.
It has the basis
|(j1 , j2 )j, mj i . (4.94)
In Exc. 4.5.4.3 we derive the matrix representation of two spins in the decoupled and the coupled
base.
To describe the two angular momenta simultaneously we must opt between the decoupled
picture |j1 , mj1 ; j2 , mj1 i and the coupled picture |(j1 , j2 )j, mj i. For now, the choice of the
picture makes no difference, but we will see later that there may be an energy associated with
the coupling 2 . In this case, as we will show, that the choice of the coupled base is more natural,
because the energy commutes like [Ĥ, ĵ2 ] = 0 = [Ĥ, ĵz ], but [Ĥ, ĵ21 ] 6= 0 6= [Ĥ, ĵ22 ].
Now we need to find the possible values of j and mj in the coupled picture. The values of
mj follow immediately from ĵ1 + ĵ2 = ĵ,
|mj | ≤ j1 + j2 . (4.97)
We often know the two angular momenta j1 and j2 and all their projections in the decoupled
base,
|mj1 | ≤ j1 and |mj2 | ≤ j2 . (4.98)
To find the quantum numbers in the coupled base, we arrange the states ordering them by their
total magnetic quantum number mj . We can, without loosing generality, concentrate on the
situation j1 ≥ j2 . The following table reproduces the possible combinations. The x-symbols
represent Clebsch-Gordan coefficients.
2
That is, the Hamiltonian of the system does not contain terms of type ĵ1 · ĵ2 , but may have terms proportional
to ĵ1 + ĵ2 .
120 CHAPTER 4. ROTATIONS / CENTRAL POTENTIALS
|j1 − j2 | ≤ j ≤ j1 + j2 . (4.99)
Each value of j has the degeneracy 2j + 1. Therefore, as will be verified in Exc. 4.5.4.4, the total
degeneracy is,
jX
1 +j2
there exist common eigenstates |j1 , j2 , j, mi for the set of observables ĵ21 , ĵ22 , ĵ2 and ĵz . These
eigenstates are linear combinations of the individual states,
X
|(j1 , j2 )j, mi = |j1 , j2 , m1 , m2 ihj1 , j2 , m1 , m2 |(j1 , j2 )j, mi (4.103)
m1 ,m2
X j1 j2 j
= |j1 , j2 , m1 , m2 i .
m1 m2 m
m1 ,m2
Example 24 (Clebsch-Gordans for the coupling of two spins 21 ): For example, for
the system consisting of two 12 spins we have,
1 1 1 q0 0 0 1 1 1 1
|( 2 , 2 )1, +1i q
1 1 |2, +2; 2, +2i
|( 1 , 1 )1, 0i 0 2 2 0 | 1 , − 1 ; 1 , + 1 i
12 21 q q 21 21 12 12 .
|( , )0, 0i = 1 1 | , + ; , − i
2 2
1 1
0 2 − 2 0 21 21 12 12
|( 2 , 2 )1, −1i |2, −2; 2, −2i
0 0 0 1
In the Excs. 4.5.4.5 and 4.5.4.6 we write all possible states of two angular momenta in
decoupled and coupled bases. In Excs. 4.5.4.7, 4.5.4.8, 4.5.4.9, 4.5.4.10, and 4.5.4.11 we practice
the transformation between decoupled and coupled bases, and in Exc. 4.5.4.12 we verify a rule
guaranteeing the unitarity of the Clebsch-Gordan transformation.
j1 + j2 = j12
+ +
j3 j3 (4.106)
= =
j13 + j2 = J
2S+1
LJ . (4.108)
4.4.4.3 jj-coupling
There is also the case that for each electron its spin couples to its own orbital angular P
momentum,
jk = lk + sk , before coupling to the total angular momenta of other electrons, J = k jk . This
is called jj-coupling. In the case of two electrons the recoupling of the four involved spins
l1 + l2 = L
+ + +
s1 + s2 = S (4.109)
= = =
j1 + j2 = J
l1 l2 L
is described by {9j} = s1 s2 S -symbols,
j1 j2 J
X
|[(l1 , s2 )j1 , (l2 , s2 )j2 ]Ji = {9j}|[(l1 , l2 )L, (s1 , s2 )S]Ji . (4.110)
L,S
4.5 Exercises
4.5.1 Particle in a central potential
4.5.1.1 Ex: Parity of the spherical harmonic functions
P
We consider the parity transformation P with (x, y, z) −→ (−x, −y, −z). Use spherical coordi-
P
nates to show that Y`m −→ (−1)` Y`,m , and therefore that a spherical surface function has even
parity when ` is even, and odd parity, when ` is odd.
where the factor g depends on the force of the interatomic interaction and Vtrp is the potential
trapping the atoms. For V (r) = V (r) the wave function will have radial symmetry, ψ(r) = φ(r)
r .
Rewrite the Gross-Pitaevskii equation for the function φ.
4.5. EXERCISES 123
V = 12 mω 2 (x2 + y 2 + z 2 ) .
a. Obtain the energy levels of this particle. That is, determine the eigenvalues of
~2 2
− ∇ ψ + V ψ = Eψ .
2m
b. Consider the fundamental level and the first two excited levels. Set up a table showing for
each of these three levels the energy value, the degeneracy, and the respective states in terms of
the quantum numbers.
c. Using
2 1 ∂ 2 ∂ L2
∇ ψ= 2 r − 2 2 ψ
r ∂r ∂r ~ r
and remembering L̂2 Y`m (θ, φ) = ~2 `(` + 1)Y`m , write down the differential equation of item (a)
for the radial part of the wavefunction (it is not necessary to solve it). Identify in this equation
the effective potential Vef f (r).
d. Solve the differential equation of the previous item for the case where ` = 0 and determine
2
the corresponding eigenvalue. To do this, allow for a solution of the type e−αr and determine
α.
a. calculate the normalization constant Dn,l for a hydrogen-like atom with atomic number Z;
b. calculate the mean value
n2 aB 1 `(` + 1)
hrinlm = 1+ 1− ;
Z 2 n2
4.5.3.4 Ex: Commutation of the absolute value and the components of the orbital
angular momentum
Show [ĵ2 , ĵ] = 0.
Initially (t = 0) the particle is in the state ψ(t = 0)i = |+i and is subject to a uniform magnetic
field B = Bêy , so that:
Ĥ = −~µ · B = −γB Ŝy .
a. What are the possible measurements of the spin projection on the y-axis?
b. Find the eigenvectors of Ŝy .
c. Get |ψ(t)i at t > 0 in terms of the eigenvectors |+i and |−i defined above.
d. Obtain the mean expectation values of the observables Sx , Sy and Sz as a function of time.
126 CHAPTER 4. ROTATIONS / CENTRAL POTENTIALS
1. The rubidium atom 85 Rb has one valence electron. In the first excited state this electron has
the orbital angular momentum, L = 1. What are the possible states?
2. In the fundamental state of this atom the total electronic angular momentum J couples with
the spin of the nucleus, I = 5/2, to form the total angular momentum F = J + I. Determine
the possible values for the angular momentum F and the magnetic quantum number mF .
Determine for the states S1/2 and P3/2 of an atom with nuclear spin I = 3/2 and hyperfine
coupling Ĵ · Î how the eigenstates of the coupled base expand into the decoupled base. Do not
consider external magnetic fields.
b. What are the possible hyperfine states F 0 resulting from a coupling of I with the total elec-
tronic angular momentum of the excited state 2 P3/2 , F 0 = 2? What are the possible Zeeman
sub-states of F 0 ?
c. A transition between a ground hyperfine state and an excited hyperfine state can be described
by a coupling of the total angular momentum F with the angular momentum of the photon κ
forming the angular momentum of the excited state F 0 . To see this, we now consider the levels
F = 1 and F 0 = 2. Expand the coupled angular momentum |(F, κ)F 0 , mF 0 i = |(1, 1)2, mF 0 i
on a decoupled basis for every possible value m0F . Use the table in Fig. 4.10 to determine the
Clebsch-Gordan coefficients.
Note: The Clebsch-Gordans only compare the oscillator strengths of transitions between Zee-
man sub-states of a given set (F, F 0 ). In order to compare the oscillator strengths to other
transitions (F, F 0 ) it is necessary to calculate 6j-coefficients.
4
See Cohen-Tannoudji, Vol.2, Complement A X.
36. Clebsch-Gordan coefficients 1
3/2 1 + cos θ θ
d 3/2,3/2 = cos
2 2
√ 1 + cos θ ³ 1 + cos θ ´2
3/2 θ d 22,2 =
d 3/2,1/2 = − 3 sin 2
2 2
√ 1 − cos θ θ 1 + cos θ
3/2
d 3/2,−1/2 = 3 cos d 22,1 =− sin θ
2 2 2
√ 1 + cos θ
3/2 1 − cos θ θ 6 d 21,1 = (2 cos θ − 1)
d 3/2,−3/2 = − sin d 22,0 = sin2 θ 2
2 2 4 r
3/2 3 cos θ − 1 θ 1 − cos θ 3
d 1/2,1/2 = cos d 22,−1 = − sin θ d 21,0 = − sin θ cos θ
2 2 2 2
3 cos θ + 1 θ ³ 1 − cos θ ´2 1 − cos θ ³3 1´
3/2
d 1/2,−1/2 = − sin d 22,−2 = d 21,−1 = (2 cos θ + 1) d 20,0 = cos2 θ −
2 2 2 2 2 2
Approximation methods
Virtually every problem going beyond the potential well, the harmonic oscillator, or the hydrogen
atom without spin and external fields is impossible to solve analytically. In this chapter we will
talk about techniques to solve approximately problems in more realistic situations. There are
a number of methods of which we will discuss the only following: 1. The stationary or time-
dependent perturbation method is useful for evaluating small perturbations of the system, for
example, caused by external electric or magnetic fields; 2. the variational method, which serves
to find and improve trial wavefunctions, the initial shapes of which are generally motivated by
the symmetries of the system; 3. the semi-classical WKB method; 4. and finally the method of
self-consistent fields, which is an iterative method of solving the Schrödinger equation.
By inserting all the expansions above and segregating all orders of λk , we find the following
system of equations,
129
130 CHAPTER 5. APPROXIMATION METHODS
We insert this expansion into the second equation (5.6) and multiply with hn|,
X
hn|(Ĥ (0) − En(0) ) |mihm|ψn(1) i = 0 = hn|En(1) − Ĥ (1) |ni . (5.8)
m
We obtain for the first order correction of the energy of unperturbed states,
As a first example we will calculate in Exc. 5.5.1.1 the first order correction for the energy
of a slightly deformed one-dimensional box potential.
Now let us have a look at the first-order correction for the wavefunction again considering the
second equation (5.6),
(1)
When n = m, the left side of this equation disappears. Therefore, En − hn|Ĥ (1) |ni = 0, and
(1)
we can restrict to the terms n 6= m discarding the terms in En ,
(1)
En δmn − hm|Ĥ (1) |ni hm|Ĥ (1) |ni
hm|ψn(1) i = (0) (0)
= (0) (0)
. (5.11)
Êm − En En − Em
We obtain for the first-order correction for the energy of the states,
This procedure simulates the distortion of the state by blending it with other states. The
perturbation induces virtual transitions to other states. The perturbation is large when the
blended levels are close.
See Exc. 5.5.1.2. In Exc. 5.5.1.3 we calculate the first order correction due to the finite
extension of the hydrogen nucleus. In Exc. 5.5.1.4 we treat the coupling of the energy levels of
a two-level system as a first order perturbation, and compare the result with the exact solution.
The Stark effect for an electron confined in a box can be discussed (see Exc. 5.5.1.5) in first
order TIPT.
5.1. STATIONARY PERTURBATIONS 131
import it into the third equation (5.6) and multiply with hn|,
X X
hn|(Ĥ (0) −En(0) ) |mihm|ψn(2) i = hn|(En(2) − Ĥ (2) )|ni+hn|(En(1) − Ĥ (1) ) |mihm|ψn(1) i . (5.14)
m m
Now,
X X
(2) (0) (0) (2) (2) (1) (1) (1)
hm|ψn i(En − Em )δnm = 0 = En − hn|Ĥ |ni + hm|ψn i En δnm − hn|Ĥ |mi .
m m
(5.15)
The left-hand side of this equation disappears. Also, on the right-hand side, for n 6= m, the term
(1)
En δnm disappears, and for n = m the whole parenthesis disappears. Therefore, we can discard
(1) (1)
the term En and restrict the sum to terms with n 6= m. Inserting the coefficients hm|ψn i
calculated in (5.11), we finally obtain,
The first term is similar to the first order correction; the eigenvalue of the second order pertur-
bation calculated in the base of the unperturbed states. The second term describes the shift of
the energies through possible temporary transitions to other states.
In Exc. 5.5.1.6 we treat a system of three coupled levels up to the second perturbative order.
The Stark effect discussed in Exc. 5.5.1.7 needs the TIPT calculation up to the second order.
(0)
When the perturbation Ĥ (1) is applied, we assume that the state |ψnµ i is distorted towards the
(0)
similar state |ψnµ i, and the energy changes from En to Enµ . We now need the index µ to label
the energy, since the degeneracy can be removed by the perturbation. As before, we write now,
The replacement of these expansions in Ĥ|ψnµ i = Enµ |ψnµ i, and a collection of the terms in λ
up to first order gives,
Ĥ (0) |ψnµ
(0)
i = En(0) |ψnµ
(0)
i (5.20)
(En(0) − Ĥ (0) (1)
)|ψnµ i = (1)
(Enµ − Ĥ (1) (0)
)|ψnµ i .
As before, we try to express the first-order corrections for the wavefunctions through degen-
(0) (0)
erate unperturbed wavefunctions, |ψnµ i, and non-degenerate wavefunctions 2 , |ψm i:
X X
(1) (0) (0)
|ψnµ i= bµν |ψnν i+ anm |ψm i. (5.21)
ν m
and multiplying the two sides with hn, µ|, we get zero on the left-hand side, since we can choose
the non-degenerate states to be orthogonal hn, ν|mi = δm,n . Hence,
X h i
(1)
cµν Enµ hn, µ|n, νi − hn, µ|H (1) |n, νi = 0 . (5.24)
ν
This secular equation (one for each µ) represents, in fact, a set of r linear equations for the
coefficients cµν . The condition for having non-trivial solutions is,
det hn, ν|H (1) |n, µi − Enµ
(1)
δµ,ν =0. (5.25)
µ,ν
(1)
The solution of this secular determinant yields the solicited energies Eµ . Now, the solution of
the secular equation (5.24) for each energy value produces those coefficients, which represent
2 (1)
Note that we label all states which are not degenerate with the state under investigation |ψnµ i with the index
m, even if there are degeneracies between them.
5.1. STATIONARY PERTURBATIONS 133
the best linear combinations adapted to the perturbation. Unlike in previous calculations with
degenerate states, here we consider linear combinations of vectors of the degenerate subspace
prior to switching on the perturbation.
In practice, we apply perturbation theory only to the lowest relevant order. That is, we only
calculate the second order correction if first order corrections vanish. One famous example is
the quadratic Stark effect discussed in Sec. 8.3. In the case of eigenvalues, which are degenerate
in the absence of perturbation, the first order will always produce a remarkable correction, as
in the example of the linear Stark effect, also discussed in Sec. 8.3. For this reason, we need not
discuss higher perturbation orders in the case of degenerate eigenvalues.
That is, the correction of the energy vanishes in first order, while the correction of the
wavefunction diverges. Obviously, the |νi obtained by the diagonalization of the matrix
Ĥ (0) is not adapted to the calculation of the matrix elements Ĥ (1) .
Now, applying degenerate perturbation theory, we obtain by the secular determinant,
!
h i −Eµ
(1)
Ω
(1) (1)
0 = det hν|Ĥ )|µi − Eµ δµ,ν = det (1) = (Eµ(1) )2 − Ω2 ,
Ω −Eµ
(1) (1)
eigenvalues are E1 = Ω and E2 = −Ω allowing the establishment of the secular equation,
h i
(1)
c11 E1 − h1|Ĥ (1) |1i − c12 h1|Ĥ (1) |2i = c11 [Ω − 0] − c12 Ω = 0
h i
(1)
−c21 h2|Ĥ (1) |1i + c22 E2 − h2|Ĥ (1) |2i = −c21 Ω + c22 [−Ω − 0] = 0 .
In Exc. 5.5.1.8 we study a partially degenerate three-level system and the breakdown of the
degeneracy due to a perturbation. And in Exc. 5.5.1.9 we will treat a perturbation in a box
potential with degenerate energy levels.
where we assume real coefficients and eigenfunctions. For this, the derivatives with respect to
all coefficients must vanish:
P P P P P
∂E k ck hk|Ĥ|qi + m cm hq|Ĥ|mi k,m ck cm hk|Ĥ|mi ( k ck hk|qi + m cm hq|mi)
= P − P 2
∂cq k,m ck cm hk|mi
k,m ck cm hk|mi
(5.32)
P P
k ck (hk|Ĥ|qi − Ehk|qi) + m cm (hq|Ĥ|mi − Ehq|mi)
= P =0,
k,m ck cm hk|mi
using the definition of E (5.31). The equation is satisfied when the numerator disappears:
X
0= cm (hq|Ĥ|mi − Ehq|mi) . (5.33)
m
The condition for the existence of solutions is that the secular determinant disappears,
The solution of this equation leads to a set of values E, and the lowest value, Emin , is the best
approximation for the ground state energy. The coefficients of the wavefunction are obtained
by solving the eigenvalue equation (5.33) with Emin .
In Exc. 5.5.2.3 we will use the Rayleigh-Ritz method to estimate the effect of a finite nuclear
mass of the hydrogen atom on the energy levels. In Exc. 5.5.2.4 we will use the Rayleigh-Ritz
method to find the maximum number of atoms allowing for a stable Bose-Einstein condensate
made of atoms subject to an attractive interatomic force.
~2 d2
− ψ(x) + V (x)ψ(x) = Eψ(x) , (5.35)
2m dx2
136 CHAPTER 5. APPROXIMATION METHODS
Substituting this function into the Schrödinger equation we obtain a system of coupled equations
in terms of A(x) and φ(x),
h i
A00 = A (φ0 )2 − k 2 and (A2 φ0 )0 = 0 . (5.38)
The equations (5.39) and (5.37) are completely equivalent to the Schrödinger equation. The
second Eq. (5.38) is easy to solve,
C
A= √ 0 , (5.39)
φ
being C a real constant. We can not say the same thing about the solution of the first Eq. (5.38).
In order to solve it we are going to use the WKB approach, assuming that A varies slowly, so
the term A00 → 0. By doing this approximation we can rewrite Eq. (5.39) in this way:
(φ0 )2 = k 2 . (5.40)
Solving this last expression we obtain two linearly independent solutions, φ0 = ±k. So we get
the expression for the phase: Z
φ(x) = ± k(x)dx . (5.41)
We write this indefinite integral, because the constant term can be absorbed by the constant C.
Finally, we obtain the expression for the wavefunction in the WKB approximation:
C R
ψ(x) = p e± |k(x)|dx . (5.42)
|k(x)|
Here, taking the absolute value of the wavevector, we have already generalized for the case that
the energy E of the particle is lower than the potential V (x) (classically forbidden region).
and insert it in the above equation. Collecting the orders in ~, we find in the first orders,
1 0 2
0
2m S0 (x) + V (x) − E ~ = 0
i~ 00 1 0
− 2m S0 (x) + m S0 (x)S10 (x) ~1 = 0
1 0 2 00
2
2m S1 (x) − 2m S1 (x) ~ = 0 .
i
Rxp
The solution of the zeroth order equation, S0 (x) = ± 2m[E − V (x0 )dx0 , gives
Rx√
i
2m[E−V (x0 )dx0
ψ(x) = Ae± ~ .
The WKB approximation can be used to describe continuous potentials (or barriers) by
stepwise constant potentials. The transmission |T |2 through these parts can be obtained by
multiplying the individual tunneling probabilities,
Z
2
ln |T | ' −2 κ(x)dx ,
barrier
p
with κ(x) = ~1 2m[V (x) − E].
Figure 5.1: (a) Potential for which we want to obtain the connection formula. (b) Turning point.
Let us start with the right turning point Fig. 5.1(a)]. First, we shift the coordinate system
so that the turning point coincides with zero, as shown in Fig. 5.1(b). As seen above, the WKB
solutions will be given by the following equations:
h R0 0 0
R0 0 0
i
√1 Bei x k(x )dx + Ce−i x k(x )dx if x≤0
k(x)
ψ(x) ≈ Rx . (5.43)
√ 1 De− 0 |k(x0 )|dx0 if x≥0
k(x)
In the vicinity of the turning point we approximate the potential by a straight line (Taylor series
expansion up to first order) with the following functional dependence,
The Schrödinger equation for this potential acquires the following format,
d2 ψt
= α3 xψt , (5.45)
dx2
with α = [ 2m
~2
V 0 (0)]1/3 . Through a change of variables, z = αx, we fall back on Airy’s equation,
d2 ψt
= zψt , (5.46)
dz 2
having as solution a linear combination of the two solutions of the Airy equation,
Now let’s have a look at the WKB solutions in the two regions √ in the vicinityRof the turning
x
point. In the classically forbidden region we have k(x) = α3/2 −x, thus being 0 |k(x0 )|dx0 =
2 3/2 . Thus, the WKB solution in the classically forbidden region near the turning point
3 (αx)
will be given by:
D 2 3/2
ψ(x) ≈ √ e− 3 (αx) . (5.48)
~α x
3/4 1/4
Using the asymptotic forms of Airy functions in the solution (5.47) we obtain the following
expression for ψt (x),
a 2 3/2 b 2 3/2
ψp (x) ≈ √ 1/4
e− 3 (αx) + √ 1/4
e 3 (αx) , (5.49)
2 π(αx) π(αx)
q
4π
which when compared to equation (5.48) shows us that a = α~ D and b = 0. Repeating the
previous steps in the negative region we see that the WKB solution in the asymptotic forms of
the Airy solutions for approximately linear potentials takes the following format (with b = 0):
1 h 2 3/2 2 3/2
i
ψ(x) ≈ √ Bei 3 (−αx) + Ce−i 3 (−αx) , (5.50)
~α3/4 (−x)1/4
and
a 1 h iπ/4 i 2 (−αx)3/2 −iπ/4 −i 23 (−αx)3/2
i
ψp (x) ≈ √ e e 3 − e e . (5.51)
π(−αx)1/4 2i
a
When compared, 2i√ π
eiπ/4 = √B~α and − 2i√ a
π
e−iπ/4 = √C~α . Having all this information we can
rewrite the WKB solutions for all positions in the potential, including the turning points 3 :
hR i
√2D sin xx2 k(x0 )dx0 + π4 , if x ≤ x2
k(x)
ψ(x) ≈ − x |k(x0 )|dx0
R . (5.52)
√ D
e x2 , if x ≥ x2
k(x)
Repeating the process for a decreasing turning point [left turning point of the potential of
Fig. 5.1(a)], we obtain the following expression:
0 Rx
1 0 0
√D e− x |k(x )|dx , se x ≤ x1
k(x) hR i
ψ(x) ≈ . (5.53)
√2D sin x k(x0 )dx0 + π , if x ≥ x1
0
k(x) x1 4
3
Note that we shifted the turning point to an arbitrary position x2 .
5.3. WKB APPROXIMATION 139
and hence the zeros of these functions, so the arguments of those sines must be equal (except
for a multiple of π),
Z x2 Z x1
π π
kdx0 + = − kdx0 − + nπ (5.54)
x 4 x1 4
Z x Z x2 1
+ kdx0 = n − π
x1 x 2
Z x2 1
kdx0 = n − π,
x1 2
with n = 1, 2... 4 . With this information we take a harmonic potentialq of the type
q V (x) =
1 2 2E 2E
2 κx . In this case, the turning points for a given energy E will be at − κ and κ . For
q
this potential we will have that k(x) = 2m 1 2
~2 (E − 2 κx ). Calculating the integral of k(x)
between these two turning points we get,
Z x2 Z √2E/k r
2m 1 m 1
k(x)dx = ~ √ E − κx2 dx = πE = n − π~ ,
x1 − 2E/k 2 κ 2
p
isolating E and taking ω = κ/m we have,
1
E = n− ~ω ,
2
with n = 1, 2..., the exact spectrum of the harmonic oscillator, but this is just a coincidence.
Eigenstates: Now we will calculate the eigenstates of the harmonic oscillator. The eigen-
functions were calculated on a computer. The first graph (Fig. 5.2) compares the first exact
excited state with that obtained using the WKB method.
Figure 5.2: First excited state calculated accurately and through the WKB approximation.
Note that the WKB approach is very good when x → 0 and x → ∞, regions where the
4
Note that n 6= 0, because the integral (5.54) has to be greater than zero.
140 CHAPTER 5. APPROXIMATION METHODS
difference between the oscillator energy and the potential are large (E V (x → 0) and
E V (x → ∞)), because in these regions the wavelength λ(x) acquires the lowest values,
p
since it is proportional to |1/ E − V (x)|. Hence, the spatial region in which the potential
needs to be practically constant is smaller, which explains why the approximation is closer
to the exact solution. In the intermediate regions the difference between E and V (x) begins
to decrease, and the WKB approximation delivers its worst results.
As we increase the energy of the harmonic oscillator, the approximation becomes better (for
the same reason as discussed in the previous paragraph). The following graph illustrates this
effect for n = 10.
Figure 5.3: Wavefunction of the vibrational state n = 10 calculated exactly and using the WKB
approximation.
Example 28 (Hydrogen atom): Eigenenergies: For the hydrogen atom the effective
potential is given by,
e2 1 ~2 l(l + 1)
V (x) = − + .
4π0 r 2m r2
Note that the WKB method for this case obeys the relation (5.54), hence we get,
Z r2 √ Z r2 s
e2 1 ~2 l(l + 1)
p(r)dr = 2m E+ − dr
r1 r1 4π0 r 2m r2
Z r2 s
√ 1 e2 ~2
= −2mE −r2 − r+ l(l + 1)dr .
r1 r 4π0 E 2mE
Notice that E < 0. Let us make the following substitution to facilitate algebraic manipula-
tions,
e2 ~2
B=− and C=− l(l + 1) .
4π0 E 2mE
The turning points r1 and r2 are given by the following expressions,
√ √
B − B 2 − 4C B + B 2 − 4C
r1 = and r2 = .
2 2
Thus, returning to the integral we will have the following:
Z r2 Z r2 p
√ 1 √ π √ √
p(r)dr = −2mE (r − r1 )(r2 − r)dr = −2mE ( r2 − r1 )2
r1 r1 r 2
√ π √ π √
= −2mE (r1 + r2 − 2r1 r2 ) = −2mE (B − 2 C)
2√ 2
π e2 2m p 1
= − √ − 2~ l(l + 1) = n − π~ .
2 4π0 −E 2
5.4. TIME-DEPENDENT PERTURBATIONS 141
Isolating E we obtain the energy spectrum of the hydrogen atom in the WKB approximation:
m e2 2 1 13.6
E=− h i2 = − h i2 eV .
2~ 4π0
n − 1/2 + l(l + 1) n − 1/2 + l(l + 1)
be good. Note that not only do eigenfunctions oscillate, but the coefficients also depend on time,
because the composition of the states can change. The instantaneous probability of finding the
system in state n is |an (t)|2 . Importing the above linear combination into the Schrödinger
equation,
h i ∂
Ĥ (0) + Ĥ (1) (t) |ψ (1) (t)i = i~ |ψ (1) (t)i , (5.59)
∂t
we find,
(0) (0) (0) (0)
a1 Ĥ (0) |ψ1 i + a2 Ĥ (0) |ψ2 i + a1 Ĥ (1) |ψ1 i + a2 Ĥ (1) |ψ2 i
" #
(0) (0)
∂a1 (0) ∂a2 (0) ∂|ψ1 i ∂|ψ2 i
= i~ |ψ i + |ψ i + a1 + a2 (5.60)
∂t 1 ∂t 2 ∂t ∂t
(0) (0) (0) (0)
=⇒ a1 Ĥ (1) |ψ1 i + a2 Ĥ (1) |ψ2 i = i~ȧ1 |ψ1 i + i~ȧ2 |ψ2 i ,
because the other terms satisfy the Schrödinger equation of zero order. Replacing the stationary
eigenfunctions,
a1 e−iE1 t/~ Ĥ (1) |1i + a2 e−iE2 t/~ Ĥ (1) |2i = i~ȧ1 e−iE1 t/~ |1i + i~ȧ2 e−iE2 t/~ |2i , (5.61)
142 CHAPTER 5. APPROXIMATION METHODS
and multiplying this equation with h1|× and h2|×, we find with the abbreviation ~ω0 ≡ E2 − E1 ,
i~ȧ1 = a1 h1|Ĥ (1) |1i + a2 e−iω0 t h1|Ĥ (1) |2i e i~ȧ2 = a1 eiω0 t h2|Ĥ (1) |1i + a2 h2|Ĥ (1) |2i . (5.62)
Frequently, the perturbation induces only a coupling, but does not directly influence the energies,
hn|Ĥ (1) |ni = 0,
e−iω0 t eiω0 t
ȧ1 = a2 h1|Ĥ (1) |2i and ȧ2 = a1 h2|Ĥ (1) |1i . (5.63)
i~ i~
Without perturbation, hm|Ĥ (1) |ni = 0, no dynamics develops; the eigenfunctions evolve
independently.
Example 29 (The Rabi formula): Now, a perturbation be suddenly switched on at time
t = 0, h1|Ĥ (1) |2i ≡ ~Ω · Θ(t). With the equations of motion we can, starting from the initial
situation a1 (0) and a2 (0), calculate the temporal evolution,
ȧ1 = −iΩa2 e−iω0 t and ȧ2 = −iΩ∗ a1 eiω0 t .
We solve this system of differential equations by differentiating one and substituting the
other,
ä2 = −iȧ1 Ω∗ eiω0 t + ω0 a1 Ω∗ eiω0 t = −a2 |Ω|2 + iω0 ȧ2 .
We find solutions via the ansatz a2 = eiω0 t/2 (AeiGt + Be−iGt ). The equation for a2 yields,
(iG + 2i ω0 )2 AeiGt+iω0 t/2 + (−iG + 2i ω0 )2 Be−iGt+iω0 t/2
= −|Ω|2 (AeiGt+iω0 t/2 + Be−iGt+iω0 t/2 )
h i
+ iω0 (iG + 2i ω0 )AeiGt+iω0 t/2 + (−iG + 2i ω0 )Be−iGt+iω0 t/2 .
Separating the parts in A and in B we obtain two equations with the same result,
G2 = |Ω|2 + 14 ω02 .
Ω is called Rabi frequency and G is the generalized Rabi frequency. Using the initial con-
ditions, a1 (0) = 1 and a2 (0) = 0, we can fix one of the coefficients A and B, since
a2 (0) = A + B = 0,
a2 = 2iAeiω0 t/2 sin Gt .
We now import this solution into the differential equation for a1 ,
ȧ1 = 2ΩAe−iω0 t/2 sin Gt .
The integral is,
Z t
0 2A −iω0 t/2
a1 (t) = 2ΩAe−iω0 t /2 sin Gt0 dt0 = − ∗
e G cos Gt + 12 iω0 sin Gt .
0 Ω
Using the normalization condition,
2A 2 2
1 = |a1 |2 + |a2 |2 = − ∗ e−iω0 t/2 G cos Gt + 12 iω0 sin Gt + A2ieiω0 t/2 sin Gt
Ω
4A2 G2
= 2
G2 cos2 Gt + 14 ω02 sin2 Gt + 4A2 sin2 Gt = 4A2 2 .
|Ω| |Ω|
Hence, A = |Ω|/2G. In general, we can choose Ω real, and the final solution is,
iω0 iΩ iω0 t/2
a1 (t) = −e−iω0 t/2 cos Gt + sin Gt and a2 (t) = e sin Gt . (5.64)
2G G
5.4. TIME-DEPENDENT PERTURBATIONS 143
The Rabi formula (5.64) can be derived exactly without perturbative calculus. In Exc. 5.5.4.1
we show a derivation using the Laplace transformation method.
When the energies En are degenerate, under the influence of the perturbation, the popula-
tions of the system oscillate with the Rabi frequency Ω. When the energies are different, the
oscillation frequency is higher, but the amplitude decreases as well. The initially empty state
never reaches unitary population. In Exc. 5.5.4.2 we calculate the time required to allow the
perturbation to invert the population of a two-level system.
This is called the Dyson series. For N = 1, we get the first order of the perturbation series 7 ,
Z
λ t
|ψI (t)i = 1 + Ŵ (τ )dτ |ψI (0)i . (5.70)
i~ 0
5
See Becker-Sauter II, p.118ff e [241], p.104ff. An alternative treatment is found in [164], p.191ff or in
Blochinzew, p.332ff.
6
Note that by substituting W by Ĥ (1) , the equation (5.67), i~∂t |ψI (t)i = Ĥ (1) (t)|ψI (t)i, corresponds to a first-
order perturbative approximation, i.e., the perturbation eigenvalues Ĥ(1) are calculated with the eigenvectors of
the unperturbed system. Thus, in first order TDPT we can substitute W for Ĥ (1) .
7
For higher orders,
X n Z t n Z t
N λ λ
|ψI (t)i ≈ Ŵ (τ )dτ |ψI (0)i = T exp Ŵ (τ )dτ |ψI (0)i .
n=1 i~ 0 i~ 0
144 CHAPTER 5. APPROXIMATION METHODS
(0)
The stationary states of the unperturbed Hamiltonian are P given by Ĥ |f i = Ef |f i. Now,
the perturbed states are expanded on this basis, |ψI (t)i = f |f iaf (t). The expansion coeffi-
cients are 8 ,
Z t
λ
af (t) = hf |ψI (t)i = hf |ψI (0)i + hf | S −1 (τ )Ĥ (1) (τ )S(τ )|ψI (0)idτ . (5.71)
i~ 0
Now, we assume that the system be initially in the eigenstate |ψI (0)i = |ii. The amplitudes
then are,
Z
λ t iEf τ /~
ai→f (t) = hf |ii + e hf |Ĥ (1) (τ )|iie−iEi τ /~ dτ (5.72)
i~ 0
Z
λ t
= δif + hf |Ĥ (1) (τ )|iieiωif τ dτ .
i~ 0
The varying potential is considered a perturbation, and a variation of the system’s state is
observed. As the energy is not conserved, [∂t , Ĥ(t)] 6= 0, the time-dependence is not separable
and the system exchanges energy with the potential. In first-order perturbation theory we only
consider weak perturbations, i.e. the initial state is gradually emptied, ai→i (dt) ' ai→i (0) = 1.
For an initially empty state the growth is obviously considerable. For i 6= f we have,
λ
dai→f (t) = ai→f (t + dt) − ai→f (t) = hf |Ĥ (1) (t)|iieiωif t dt . (5.73)
i~
This formula is nothing more than a generalization of the formula (5.63) obtained for a two-level
system assuming that the initial state does not deplete considerably. In Exc. 5.5.4.3 we calculate
the dynamics of a harmonic oscillator perturbed by a decaying force.
can be calculated in a single command line using the Matlab ’expm’ function.
When the system varies temporally, Ĥ(t), we may divide time into small units dt and prop-
agate the wavefunction as,
|ψ(t + dt)i = e−iĤ(t)dt/~ |ψ(t)i ' |ψ(t)i 1 − i Ĥ~ dt , (5.75)
continuously reinserting the solution into the equation. This Newton method does not converge
quickly (dt should be chosen small enough, when Ĥ(t) varies rapidly), but there are other more
sophisticated methods like the Runge-Kutta method.
A variation of this method is called steepest descent method. This method is similar to
the Newton Eq. (5.75), but replaces the time dt with an imaginary time. Thus, the coherent
8
We could define the coefficients in Schrödinger’s picture, af ≡ hf |ψi, but this would only introduce a phase
factor, ai→f −→ ai→f ei(Ef −Ei )t/~ , which is unimportant for absolute values |ai→f |2 . This corresponds to a
transformation to a rotating system, which will be discussed in Sec. 10.2.1.
5.4. TIME-DEPENDENT PERTURBATIONS 145
temporal evolution of the Schrödinger equation is replaced by a dissipative evolution. The loss
of energy automatically takes the system to the ground state. The method also applies to
more complicated equations than the Schrödinger equation, for example, the Gross-Pitaevskii
equation.
Another numerical method often used in quantum mechanics is the method called the quan-
tum Monte Carlo simulation of the wavefunction [184]. This method simulates trajectories of
quantum systems treating intrinsic quantum noise as random processes disrupting the uniformity
of the trajectory. The advantage of this method is that it also applies to dissipative systems.
We obtain for i 6= f ,
2
λ2 2 sin (ωif t/2)
|ai→f (t)|2 = |hf | Ĥ (1)
|ii| . (5.77)
~2 (ωf i /2)2
The generalization of this transition rate for all perturbation orders is,
1 2πλ2 X (1) λ X hf |Ĥ (1) |lihl|Ĥ (1) |ii
= hf |Ĥ |ii + + ... (5.80)
τ ~2 f ~ l ωi − ωl
2
λn X hf |Ĥ (1) |l1 ihl1 |...|ln−1 ihln−1 |Ĥ (1) |ii
+ n δ(ωif ) ,
~ l1 ,...,ln−1 (ωi − ωl1 ) ... ωi − ωln−1
The first term being small, we neglect it in the rotating wave approximation (RWA). We obtain,
This result coincides with the Rabi formula (5.64), except that the energy difference between
the states ωf i is shifted by the frequency of the perturbation ω. The quantity ∆f i ≡ ω − ωf i
is called a detuning. The transition probability is maximal, when we are at resonance, that is
∆f i = 0. In this case,
|ai→f (t)|2 −→ |Ωf i |2 t2 . (5.84)
This can be seen by expanding the numerator in a Taylor series for small (ωf i − ω)t.
Note, that the probability exceeds 1 for long times, which can not be. In fact, the restriction
to the first order in the Taylor expansion used in the derivation of the last equation is no longer
valid for long times, when (ωf i − ω)t > 1, and we need to take into account higher orders.
corresponds to the probability of the initial state |ii to be depleted. When the final state lies
within a continuum, the sum in (5.85) must be replaced by an integral. With the density-of-
states written in the form ρ(E), where ρ(E)dE is the number of states found in the energy range
5.4. TIME-DEPENDENT PERTURBATIONS 147
Z Emax
Pi→F (t) = |ai→f (t)|2 ρ(Ef )dEf , (5.86)
Emin
where E ∈ [Emin , Emax ] is the regime of energies within reach of the perturbation. Now,
Z 1
Emax sin2 − E)t
2~ (Ef i
Pi→F (t) = |Ωf i |2 1 ρ(Ef )dEf . (5.87)
Emin 4~2
(Ef i − E)2
δ(x) = lim t
sinc2 xt , (5.88)
t→∞ π
with the substitution x ≡ (Ef i − E)/2~, we obtain after sufficiently long times 12 ,
Z Emax
π E i −E
Pi→F (t) = |Ωf i |2 t2 δ( f2~ )ρ(Ef )dEf = 2π~t|Ωf i |2 ρ(Ei + E) . (5.89)
Emin t
dPi→F 2π
= |hf |Ĥ (1) |ii|2 δ(Ef i − E) , (5.91)
dt ~
where we went back to the definition of the perturbation Hamiltonian (5.81). This expression
is called Fermi’s Golden Rule. In Exc. 5.5.4.5 we will calculate the photoelectric effect.
To derive Eq. (5.83), we considered perturbations with fixed oscillation frequencies. To handle
frequency distributions %(ω), we must generalize this equation by calculating the integral,
Z
2 2 sin2 1 (ωf i − ω)t
|ai→f (t)| = |Ωf i | %(ω) 1 2 2
dω (5.92)
4 (ωf i − ω)
Z ∞
2
' |Ωf i | t%(ωf i ) sinc2 xdx = πt|Ωf i |2 %(ωf i − ω) ,
−∞
again using the representation (5.88) of the Dirac function. The approximation %(ω) = %(ωf i )
can be used if the width of the sinc function is much narrower than the frequency distribution,
which is the case for sufficiently long times, t π/2∆f i .
11 P
With ρ(E) ≡ f ∈F δ(Ef − E) the integral is converted back into a sum.
12
Remember δ(ax) = a1 δ(x).
148 CHAPTER 5. APPROXIMATION METHODS
Rf →i = Bf i Nf J (ωf i ) (5.95)
and
~ωf3i ~ωf3i Ri→f ωf3i
Sf →i = Af i Nf = 2 3 Bif Nf = 2 3 = |df i |2 , (5.96)
π c π c J (ωf i ) 3π0 ~c3
exploiting the relation (1.100). We shall return to the subject of transitions in chapter ??.
5.5 Exercises
5.5.1 Stationary perturbations
5.5.1.1 Ex: One-dimensional well with a deformation in the centre
Consider a one-dimensional potential well between −L/2 and L/2 with infinitely high walls. In
the center of the well is a small deformation,
(1) ε for −a
2 ≤x≤ 2
a
H =
0 outside that region .
Calculate the correction for the eigenenergies in first order and discuss the limits a L and
a → L.
Periodic systems
Many physical systems treat quantum particles in periodic potential. Examples are electrons in
crystals or cold atoms in optical lattices. The periodicity adds a wealth of new phenomena.
~2 2
Ĥ = − ∇ + V (r) (6.2)
2m
we can write the Schrödinger equation,
Since V and ∇ are invariant under translations Utr (R)ψ(r) ≡ ψ(r + R) by a fixed distance R,
where the translation operator has been defined in Eq. (2.168), we have,
This relation holds for all vectors R of the lattice, such that,
The relationship f (R1 + R2 ) = f (R1 )f (R2 ) is satisfied by the ansatz f (R) ≡ eik·R , where k is
an arbitrary vector of reciprocal space. We get the famous Bloch theorem,
which represents a necessary condition for any eigenfunction ψk of the Schrödinger equation
with periodic potential. Bloch’s theorem simply postulates that, apart from a phase factor, the
wavefunction has the same periodicity as the potential.
1
This also holds true for degenerate eigenvalues if we choose suitable basis of eigenvectors.
153
154 CHAPTER 6. PERIODIC SYSTEMS
automatically satisfies Bloch’s theorem. That is, the wave function of the electron ψ is a plane
wave eik·r modulated by a function uk having the same periodicity as the lattice [151]. Although
the vector of the electronic wave is arbitrary, it is possible (and useful) to restrict its value to the
first Brillouin zone defined by k ∈ [−π/a, π/a], where a is an elementary vector of the lattice.
The reason is that we can reduce a wavevector k in a wavefunction trespassing the first Brillouin
zone by an appropriate vector G of reciprocal lattice,
k0 = k + G , (6.9)
yielding,
0
ψk (r) = uk (r)eik·r = uk (r)e−iG·r eik ·r . (6.10)
We now define another function uk0 (r) ≡ uk (r)e−iG·r , which also satisfies the requirement (6.8),
knowing that G · R = n2π, we see,
Hence,
0
ψk (r) = uk0 (r)eik ·r = ψk0 (r) . (6.12)
That is, we neglect superposition states of the electron at various sites of the lattice. The atom
is subject to a potential Uat (r − Ri ) located near the atom at the position Ri , and it is described
by the eigenfunction φ(r − Ri ) (only defined for the site i) with energy E0 ,
~2 2
− ∇ + Uat (r − Ri ) φ(r − Ri ) = E0 φ(r − Ri ) . (6.14)
2m
Even so, the function ψk (r) must satisfy Bloch’s theorem. This is the case when ci (k) = eik·Ri
and therefore,
X
ψk (r) = eik·Ri φ(r − Ri ) . (6.15)
i∈lattice
We now calculate the energy E(k) of an electron with the wavevector k inserting the function
ψk (r) of (6.15) in the Schrödinger equation and obtain,
X X
~2 2
− ∇ + U (r) eik·Ri φ(r − Ri ) = E(k) eik·Ri φ(r − Ri ) . (6.16)
2m
i i
U (r) is the potential energy of the electron illustrated in Fig. 6.1 together with the energy
Uat (r − Ri ) of a free electron.
Substituting the kinetic energy term of (6.16) by the kinetic energy of (6.14), we calculate,
X
eik·Ri [−Uat (r − Ri ) + E0 + U (r) − E(k)]φ(r − Ri ) = 0 . (6.17)
i
P ik·Rj ∗
Now, multiplying this equation with ψk∗ (r) = je φ (r − Rj ) and integrating over the
volume of the crystal, we obtain,
X Z
ik·(Ri −Rj )
[E(k) − E0 ] e φ∗ (r − Rj )φ(r − Ri )dV (6.18)
i,j
X Z
= eik·(Ri −Rj ) φ∗ (r − Rj )[U (r) − Uat (r − Ri )]φ(r − Ri )dV .
i,j
The functions φ∗ (r − Rj ) and φ(r − Ri ) overlap only a little, even for adjacent atoms, such
that we can neglect the terms i 6= j on the left side. The sum then corresponds to the number
N of sites in the lattice. On the right side we can not neglect the terms involving other sites,
because even if the wavefunctions of adjacent sites overlap little, the contribution of the potential
difference |U (r) − Uat (r − Ri )| is much lower for r = Ri than for r = Rj . On the other hand,
as the wavefunctions φ(r − Ri ) disappear quickly when |r − Ri | > |Rm − Ri |, we can focus on
adjacent sites (called Rm ),
Z
N [E(k) − E0 ] =N φ∗ (r − Ri )[U (r) − Uat (r − Ri )]φ(r − Ri )dV (6.19)
X Z
+N eik·(Ri −Rm ) φ∗ (r − Rm )[U (r) − Uat (r − Ri )]φ(r − Ri )dV .
m=adjacent
Figure 6.1: Potential energy U (r) of a crystal electron (red) and potential energy UA (r − Ri ) of
the electron of a free atom (blue).
Now we further suppose that the eigenfunction φ exhibits radial symmetry corresponding to
156 CHAPTER 6. PERIODIC SYSTEMS
The interpretation is as follows: The combination of the atoms in a lattice produces an energy
displacement α. In addition, it generates a spitting into a continuous band of energies as a
function of reduced wavevector k...
Consequently, we can make for Bloch functions (6.10) the following periodic ansatz,
X
ψk (r) = uk (r)eik·r with uk (r) = √1V uG (k)eiG·r , (6.22)
c
G
that is,
Xh 2 i 0
X 00 ·r
X 0
~ 0 2
√1
V
− 2m (k + G ) − E(k) uG0 (k)ei(k+G )·r + √1
Vc
UG00 eiG uG0 (k)ei(k+G )·r = 0 .
c
G0 G00 G0
(6.26)
Now multiplying with √1V ei(k+G)·r and integrating over the volume of the crystal (knowing
1
R iG·r c
Vc Vc e dV = δG,0 ), we obtain,
h i X
~2
2m (k + G)2 − E(k) uG (k) + UG−G0 uG0 (k) = 0 , (6.27)
G0
6.1. THE BLOCH MODEL 157
~2
[(k + G)2 − k 2 ]uG (k) − U0 uG (k) + U0 uG (k) + UG u0 (k) = 0 (6.28)
2m
UG u0 (k)
uG (k) = ~2 . (6.29)
2 2
2m [k − (k + G) ]
Since the Fourier coefficients UG have, for G 6= 0, small values, the function uG (k) is not
negligible only for k 2 ' (k + G)2 that is,
P
In the Fourier expansion, V (z) = K UK eiKz , this potential corresponds to the Fourier coeffi-
cients U0 = −V0 and U±K = − V20 . We also expand the wavefunction into plane waves, ψ(z) =
P iqz
q cq e , and we insert these expansions into Schrödinger’s stationary equation Ĥψ = εψ,
yielding, " #
−~2 ∂ 2 X X X
iKz
2
+ UK e cq eiqz = ε cq eiqz . (6.35)
2m ∂z q q
K
In matrix notation,
M̂ c = εc . (6.37)
where the matrix is around n = .., −1, 0, +1, ..:
.
.. ..
.
~2
2
2m (k − K) − V0 − 12 V0
ck−K
M̂ =
− 21 V0 ~2 2
2m k − V0 − 12 V0
with c= ck .
~2
ck+K
1 2
− 2 V0 2m (k + K) − V0
.. .
..
.
(6.38)
For shallow potentials, V0 ~ K /2m, we can neglect the coefficients V0 in the Eq. (6.36)
2 2
and we find,
ε ≈ ~2 q 2 /2m , (6.39)
which corresponds to the dispersion relation for free particles. On the other hand, looking at
the bottom of deep potentials, V0 ~2 K 2 /2m,pwe can harmonically
√ approximate the cosine
m 2 2 −1
potential by V (z) ≈ −2V0 + 2 ω z with ω = K V0 /m = ~ 2V0 Er . For this case we expect,
ε ≈ −2V0 + ~ω n + 21 . (6.40)
The exact spectrum of eigenvalues ε can be calculated by numerically determining the eigenvalues
of the matrix (6.38) for the first Brillouin zone, k ∈ [−K/2, K/2], and the above limits are
confirmed.
To estimate the width of the forbidden band, we cut out a 2 × 2 matrix within the matrix
M̂ and neglect its coupling with the others submatrices,
!
~2 2−V 1
(k − K) 0 − V
2 0
M̂s = 2m ~2 2 . (6.41)
− 12 V0 2m k − V0
At the edges of the Brillouin zone, k = 12 K, we get the eigenvalues ε = ~ mK − V0 ± V20 , that is,
2 2
the band gap is ∆ε = V0 2 . Bloch’s theorem says that Schrödinger’s equation can be solved for
any Bloch states. These are superpositions of plane wave momentum states [10],
Figure 6.2: (Code: QM Lattices BlochBands.m) (Continuous red line:) Bloch Bands. (Dot-
ted black line:) Without potential, V0 = 0. The parameters are ωr = (2π) 20 kHz,
ωho = (2π) 12 kHz, λL = 689 nm, and V0 = 0.2~ωr .
E = A cos ak , (6.47)
where a is the lattice parameter and A a constant. Then, v(k) is given by,
Aa
v(k) = − sin ak , (6.48)
~
and the position of the electron by,
Z
A aeE
x(t) = v(k(t))dt = − cos t . (6.49)
eE ~
This shows, that the electron is oscillating in real space. The oscillation frequency is given by,
ae|E|
ωB = . (6.50)
~
∂ ψ̂ ~2 ∂ 2 ψ̂ ~W0
i~ =− + sin(2kl x)ψ̂ − mgxψ̂ . (6.52)
∂t 2m ∂x2 2
The additional term, which contains the frequency of the Bloch oscillation νb , increases
linearly over time. As time goes by, a resonance is crossed when t = −nτb , and the crossing
is periodically repeated at every n = −1, −2, 0, ... Tracing the matter wave evolution in the
laboratory system, we see that whenever the resonance is crossed, the momentum undergoes
a change of sign corresponding to a reflection of its motion. We expand the population of the
momentum states into plane (Bloch) waves with |cn (t)|2 [219, 220],
∞
X
ψ̂(x, t) = cn (t)e2inkl x · eimgxt/~ (6.53)
n=−∞
6.1. THE BLOCH MODEL 161
transform into the moving frame, and obtain the time dependent solution with the usual defini-
~k2
tion of the recoil frequency ωr = 2ml and
g
νb = . (6.54)
ωr
Here we give another pictorial representation of the Bragg reflection process. The require-
ment for commensurability of the Broglie wavelength and wavelength of the standing light wave
is equivalent to saying that the matter wave momentum is equal to the recoil of a single photon.
In other words, the matter wave is on the edge of a Brillouin zone. In fact, the dispersion relation
of a free particle is distorted due to the periodicity of the potential generated by the standing
light wave such as to open a forbidden band. As a consequence, instead of being accelerated
without limits, the atom enters the second Brillouin zone, which is to say that it is reflected to
the other side of the first Brillouin zone.
An important question concerns the origin of the interaction between the matter wave and
the standing light wave? An interaction is only possible if the atom has an internal transition
capable of scattering photons from the light beams. As any absorption and emission process
transfers a recoil momentum to the atom, we can understand the process as a Raman scattering
process: a photon of the laser beam coming from the left is absorbed and re-emitted to the left.
This transfers twice the photonic recoil to the atom, which explains the Bragg reflection of the
atom by the optical lattice.
Of course there are some conditions that need to be met to observe Bloch oscillations. The
transfer of momentum is efficient only in the rapid adiabatic passage (ARP) regime characterized
by the conditions 2(νb /ωr ) (W0 /4ωr )2 16. The first condition requires that the force that
drives the atoms to perform the Bloch oscillations must be weak enough to avoid transitions
between Bloch bands, which guarantees the adiabaticity of the process. The other condition
requires that the optical lattice be weak enough so that the dynamics involves only two adjacent
momentum states at the same time and the transfer between the two is successful.
162 CHAPTER 6. PERIODIC SYSTEMS
1.8
1.6
1.4
1.2
En /h̄ωr
0.8
0.6
0.4
0.2
0 W0
-0.2
-0.5 0 0.5
q/2ki
Inserting into the Schrödinger equation the plane wave ansatz ψ = AeiKx + Be−iKx for the
wavefunction in the valley, 0 < x < a, and ψ = CeQx + De−Qx in the peak, −b < x < 0,
we obtain ε = ~2 K 2 /2m and U0 − ε = ~2 Q2 /2m. Choosing the constants A, B, C, D such
that ψ and ψ̇ are continuous in x = 0, a, we derive, using the periodicity of the Bloch wave
ψ(a < x < a + b) = ψ(−b < x < 0)eik(a+b) ,
1 1 −1 −1 A
iK −iK −Q Q B
= 0 . (6.58)
eiKa e−iKa −e−Qb+ik(a+b) −eQb+ik(a+b) C
iKeiKa −iKe−iKa −Qe−Qb+ik(a+b) QeQb+ik(a+b) D
The determinant of the matrix must be zero, or,
Q2 − K 2
sinh Qb sin Ka + cosh Qb cos Ka = cos k(a + b) . (6.59)
2QK
For δ-shaped peaks, we let b = 0 and U0 = ∞ such that Q2 ba/2 = P , this simplifies to,
P
sin Ka + cos Ka = cos ka . (6.60)
Ka
The dispersion relation for light is different. According to [172],
√ √ −1
ε+ ε √ √
− sin( εωa/c) sin(ωa/c) + cos( εωa/c) cos(ωa/c) = cos ka . (6.61)
2
For ε = 1 the equation simplifies to ka = 2ωa/c.
6.3. EXERCISES 163
ω0
Re θ ρ(ω) ω0
Figure 6.5: Dispersion ratio and state density for a one-dimensional optical lattice.
6.3 Exercises
6.3.1 The Bloch model
6.3.1.1 Ex: Gaussian approximation for Wannier function
Consider a stationary light wave producing a dipolar potential of the V (x) = W0 sin2 (kx) =
W0 W0 2
2 − 2 cos 2kx com W0 Er = (~k) /2m.
a. Approximate the potential of a single site around x = 0 for a harmonic potential.
b. Calculate for which depth of the dipolar potential the approximate potential supports at
least one connected state. c. Calculate the spacing of the levels and the length of the harmonic
oscillator aho .
d. Use the appropriately normalized ground state of this harmonic oscillator as an approximation
for the Wannier function ω0,0 (x). It assumes the same depth of the network in all directions of
3/4
a 3D cubic network. Derive the formula U 3d = π8 kad W Er
0
starting from,
Z
3D 3D 4
U =g ω(0,0) (x, y, z)d3 r
4π~ad
com g 3D = m .
164 CHAPTER 6. PERIODIC SYSTEMS
(a) (b)
hw
(c)
E0
W0
Figure 6.6: (a) Approximation of a periodic potential by a potential harmonic. (b-c) An infinite
periodic potential should be considered as free or confining?
165
Chapter 7
167
168 CHAPTER 7. THE ELECTRON SPIN AND THE ATOMIC SUBSTRUCTURE
We obtain,
" 2 #
1 ∂2 me c2
− ∇2 + ψ=0 . (7.6)
c2 ∂t2 ~
This is the Klein-Gordon equation. The stationary solution of this equation is a spherical
wave,
1
ψ = ψ0 e−2πr/λC , (7.7)
r
where λC = h/me c is the Compton wavelength. We show this in Exc. 7.5.1.1. For example, in
the case of heavy bosonic particles, such as a field of π-mesons, ψ is the Yukawa potencial.
In the framework of the standard model, it is believed that matter is composed of two
fundamental types of particles, bosons and fermions. Bosons are exchanged between fermions
conveying the interaction between them. A typical example is the one of two electrons whose
Coulomb interaction is mediated by the exchange of photons. Bosons obey the Klein-Gordon
equation, fermions the Dirac equation derived in the following section.
We must now ensure that the relativistic energy-momentum condition (7.5) be satisfied.
Example 31 (Derivation of the Dirac equation): Taking the square on the right-hand
side of the equation (7.9),
The matrices αi must satisfy the condition (7.10). It is possible to show that this requires
at least four-dimensional matrices of the following form:
I 0 0 σj
α0 = and αj = , (7.11)
0 −I σj 0
where j = x, y, z = 1, 2, 3. In this notation the components of the matrices are also matrices:
1 0 0 1 0 −i 1 0
I= , σx = , σy = , σz = , (7.12)
0 1 1 0 i 0 0 −1
called Pauli spin matrices. The state vector must also have four dimensions,
~ t)
~~
Φ(r, t) =
φ(r,
with ~ t) = φ1 (r, t)
φ(r, and χ
~ (r, t) =
χ1 (r, t)
. (7.13)
χ
~ (r, t) φ2 (r, t) χ2 (r, t)
Φj are called large components, χj are called small components. This designation is explained
later. Combining the matrices αj to a three-dimensional vector α~ , we can now write the Dirac
equation (7.9) like,
~ ~
~ t) = me c2 α0 + c~
i~∂t Φ(r, ~ t) .
α · p Φ(r, (7.14)
Or, using the notation (7.11) and combining the Pauli matrices σj to a three-dimensional vector
~σ , we can write the Dirac equation as:
∂ φ ~ 2 I 0 0 ~σ · p ~
φ
i~ = me c +c . (7.15)
∂t χ
~ 0 −I ~σ · p 0 χ
~
γ 0 ≡ α0 and γ k = γ 0 αk . (7.17)
We obtain,
I 0 0 σk
γ0 = and γk = . (7.18)
0 −I −σk 0
We also define another important matrix by,
0 I
γ5 ≡ iγ0 γ1 γ2 γ3 = . (7.19)
I 0
With this, using Einstein’s notation 3 , the Dirac equation (7.13) adopts the form,
The complete system is summarized in the Minkowski metrics of time-space in the form,
[γ µ , γ ν ]+ = 2η µν , (7.21)
A fundamental theorem states that, if two distinct sets of matrices are given, which both
satisfy Clifford’s relations, then they are connected to each other by a similarity transforma-
tion:
γ 0µ = S −1 γ µ S . (7.24)
If, in addition, the matrices are all unitary, as is the case of Dirac’s set, then S is unitary,
γ 0µ = U † γ µ U . (7.25)
7.1.2.1 Anti-particles
Disregarding for a moment the non-diagonal matrix, the Dirac equation separates into two
independent equations,
~
∂φ ∂~
χ
i~ ~
= me c2 φ and i~ = −me c2 χ
~ . (7.26)
∂t ∂t
These are eigenenergy equations with the eigenvalues me c2 and −me c2 . The state with negative
energy is interpreted as anti-particle. Therefore, the non-diagonal matrix mixes particles and
anti-particles.
3
∂0 ≡ 1c ∂t
7.1. THE DIRAC EQUATION 171
Example 34 (Vanishing rest mass): Let us note that for the case of vanishing rest mass,
me = 0, the Dirac equation (7.15) dramatically simplifies. Taking the time derivative of the
upper equation (7.15) and inserting the lower equation (7.15), we find,
ŝ ≡ ~2 ~σ , (7.37)
and we calculate the commutation relations between its components. From the definitions (7.12)
we obtain the rule,
~2 0 1 0 −i ~2 0 −i 0 1 ~2 2i 0
[ŝx , ŝy ] = 4 − 4 = = i~Ŝz . (7.38)
1 0 i 0 i 0 1 0 4 0 −2i
In general terms the following holds true: [si , sj ] = ijk i~sk . It is interesting to compare this with
the commutation relation for the orbital angular momentum [ˆli , ˆlj ] = ijk i~ˆlk . The coincidence
suggests a generalization of the concept of angular momentum: We now call angular momentum
operator every three-dimensional vector operator satisfying this commutation relation 4 . We
consider the eigenvalue equation for Ŝz ,
~ 1 0 φ1 φ
~
Ŝz φ = 2 = ms ~ 1 . (7.39)
0 −1 φ2 φ2
The eigenvalues are obviously ms = ± 12 . The angular momentum related to the matrices S is
obviously half-integer. We are dealing here with a new type of angular momentum, which is
not included in the usual definition of orbital angular momentum L = r × p. The new angular
momentum is called intrinsic angular momentum or spin of the particle. The spin represents
a new structure or dimension additional to space comparable to the polarization of light. The
photons of a circularly polarized light beam also contribute to an intrinsic angular momentum,
which however in this case is integer.
In Exc. 7.5.1.3 we will see that neither ˆlz nor ŝz are constants of motion of the Hamiltonian
(7.15), but the sum ĵz ≡ ˆlz + ŝz ,
[ĵz , Ĥ] = 0 . (7.40)
we obtain coupled stationary equations for the large and small components,
~
(E − me c2 )φ(r) = cσ · p~
χ(r) and (E + me c2 )~ ~
χ(r) = cσ · pφ(r) . (7.42)
4
This concept can be derived from the requirement of symmetry under rotation of space as discussed in Sec. 2.5.
7.1. THE DIRAC EQUATION 173
p̂ −→ p̂ − qA ≡ ~π and Ĥ −→ Ĥ + qU . (7.44)
called the minimal coupling. These rules will be derived in Sec. 8.1. In addition to the substitu-
tion of the momentum, we must add the scalar potential qU , and we obtain the Dirac equation
for a particle inside an applied electromagnetic field,
˙
~ ~
~ = me c2 α0 + c~
i~Φ ~ .
α · ~π + qU Φ (7.45)
with ` = 0, 1, ...
The energy now depends on two quantum numbers. The degeneracy of the orbital angular
momentum j is lifted, and the new quantum number besides the main quantum number n is that
of the total angular momentum j. The intransparent expression for the energy can be expanded
by α,
2 α2 α4 1 3
En,j ' me c 1 − 2 − 3 − . (7.49)
2n 2n j + 1/2 4n
5 E/c U/c
In quadrivetorial notation: πµ = pµ − qAµ with pµ = and Aµ = .
p A
6
See also http://einstein.drexel.edu/∼bob/Term Reports/Whitehead 3.pdf
174 CHAPTER 7. THE ELECTRON SPIN AND THE ATOMIC SUBSTRUCTURE
The second term reproduces the energy of Bohr’s model, but there are correction terms pro-
portional to α4 . We will show in Secs. 7.1.4, that the energy levels, called fine structure, result
from several relativistic corrections of different origins. We will transform the Dirac equation
into spherical coordinates in Exc. 7.5.1.4 and derive its solution (7.48) in 7.5.1.5 and 7.5.1.6.
In the expression (7.49) for the electron energy in the Coulomb potential, the last term is
positive and proportional to 1/n4 . It describes relaxation of the binding due to the contribution
of weak components. The term containing the quantum number j is called the spin-orbit cou-
pling. To better understand this contribution we must first analyze more deeply the matrices
~σ .
which holds for [~σ , B] = 0 = [~σ , C] and has been demonstrated in Exc. 7.5.1.2. The relationship
yields,
(~σ · p)2 = p2 and [~σ · pU (r)](~σ · p) = pU (r) · p + i~σ · [pU (r) × p] , (7.56)
so that,
~ ' qU (r)φ
~+ E 0 −qU (r) ~− ~2 q ~+ ~ ~ . (7.57)
E0φ 1
2me 1− 2me c2
p2 φ 4m2e c2
∇U (r) · ∇φ 4m2e c2
~σ · [∇U (r) × p]φ
p2 1 e2 p4 e 1 ∂U ~2 e ∂U ∂ ~
Ĥ ' − − − s · l − φ . (7.60)
2me 4πε0 r 8m3e c2 2m2e c2 r ∂r 4m2e c2 ∂r ∂r
The first two terms are those arising from Bohr’s atom model, the third one is a correction due to
the relativistic velocity of the electron, the forth comes from the electron’s spin-orbit coupling,
and the fifth is called the Darwin term. All contributions represent perturbations to the non-
relativistic Schrödinger theory of Bohr’s atom and will be discussed extensively in Secs. 7.2. We
will show in 7.5.1.7 that l̂2 , ŝ2 , and ĵ2 are constants of motion of the above Hamiltonian.
p2 1 e2 i~e e
Ĥ ' − − (∇ · A + A · ∇) + s·B , (7.64)
2me 4πε0 r 2me me
neglecting terms in A2 .
The Pauli equation serves for a classical (non-relativistic) approach to the electron’s spin-
orbit coupling, as we will see below and in the discussion of the fine structure in Sec. 7.2.2. In
also serves to calculate the coupling of the nuclear spin to the electron, as will be shown in the
discussion of the hyperfine structure in Secs. 7.3.
For the g-factor of the electron, we obtain ge = 2 8 . Neutron and proton are also fermions
with spin 21 , but they do not obey the Dirac equation! Their g-factors are gproton = 5.5858
and gneutron = −3.8261. The large deviation from g = 2 points to the existence of an internal
structure.
The rapid motion of the electron within the electrostatic field E of the nucleus produces,
following the theory of relativity, in the electron’s reference frame a magnetic field B0 with which
the electronic spin can interact. As we will show in 7.5.2.1, the field seen by the electron can be
approximated in first order in v/c by,
v
B0 ' ×E . (7.71)
c2
With this the interaction energy (7.68) becomes,
e e
µs · B0 =
Ĥ`s = −~ 2
s · (v × E) = − 2 2 s · (p × ∇U ) (7.72)
me c me c
e r dU 1 dV (r)
= − 2 2s · p × =− 2 2 s·l ,
me c r dr me c r dr
We will discuss the various terms in the following sections. Note that the expression for the
Darwin term differs from that of (7.60). We will see in Exc. 7.5.1.8, that they are, in fact,
equivalent.
p p2 p4
Ekin = p2 c2 + m2e c4 ' me c2 + − + ... . (7.74)
2me 8m3e c2
p4
Hrl 8m3e c2 v2 α2
= p2
= ' ≈ 0.01% . (7.75)
H0 4c2 4
2me
Due to the degeneracy of these states, it would be appropriate to use perturbation theory
with degenerate states. However, as Ĥrl only depends on spatial coordinates commuting with l
and s, the degeneracy is not very important, since Ĥrl is already diagonal in the base |n, `, mi,
that is, hn, `, m|n0 , `0 , m0 i = δ``0 δmm0 . Starting from,
2 2 2
p4 1 p 1 Ze2
Ĥrl = − 3 2 = − =− Ĥ0 + (7.76)
8me c 2me c2 2me 2me c2 4πε0 r
2
1 2En n2
=− Ĥ0 − ,
2me c2 r̃
Zr
with r̃ ≡ aB and using as an abbreviation the energies of hydrogen following Bohr’s model,
Z 2 e2 1 me c2 Z 2 α2
En = hn, `|Ĥ0 |n, `i = − = − . (7.77)
4πε0 2aB n2 2 n2
We have
" 2 #
1 4E n n 2E n n
∆Erl = hn, `|Ĥrl |n, `i = − hn, `|Ĥ02 |n, `i − hn, `| Ĥ0 |n, `i + hn, `| |n, `i
2me c2 r̃ r̃
" #
Z 2 α2 2 2 2 1 2 4 1
= En − 4En n 2 + 4En n 3 , (7.78)
4En n2 n n (` + 21 )
using the eigenvalues calculated in (4.49). Finally, we obtain the following relativistic correction,
" #
2 1 3
∆Erl = En (Zα) 1 − 4n2 . (7.79)
n(` + 2 )
Obviously, the degeneracy with respect to the angular momentum ` is lifted by this correction.
7.2. FINE STRUCTURE OF HYDROGEN-LIKE ATOMS VIA TDPT 179
In the following we will derive the expression from classical arguments borrowed from electro-
dynamic theory.
respectively,
Z
µ0 (x − r0 ) × j(r0 ) 3 0
B(x) = ∇x × A(x) = d r (7.82)
4π V |x − r0 |3
Z Z Z
µ0 ∞ 0 ∞ 0 0 2π (x − r0 ) × v Ze Zeµ0 (x − r) × v
=− dz r dr dφ 0 |3 2πr
δ(r − r0 )δ(z 0 ) = ,
4π −∞ 0 0 |x − r 4π |x − r|3
where we replaced the expression for the current density (7.65). With the expression for the
Coulomb potential between the electron and the proton and its radial derivative,
To diagonalize the angular part of the Hamiltonian, we need the common wavefunctions of l2
and l · s. We can rewrite the coupling term as:
l · s = 12 (j2 − l2 − s) . (7.92)
In the common eigensystem of j2 , l2 , and s2 the Hamiltonian, therefore, is diagonal. We know
the basis of this system from the theory of the addition of angular momenta. The states of the
basis are linear combinations of the functions |n, `, m` , ms i. Since the spins precess around each
other, `z and sz are not good observables, the non-coupled base is not appropriate. But s2 , l2 ,
and j2 are good observables. In the coupled basis {n, (`, s)j, mj },
~2
hn, (`, s)j, mj |s · l|n, (`, s)j, mj i = 2 [j(j + 1) − `(` + 1) − s(s + 1)] . (7.93)
Since j = ` ± 1/2, we find that every level splits into two levels, one with the energy En` + `ζn`
and the degeneracy 2`+2 and the other with the degeneracy En` −(`+1)ζn` with the degeneracy
2`, where we introduced the abbreviation,
~2
ζn` ≡ 2 hξ(r)i . (7.94)
All in all, we get an energy correction due to the spin-orbit interaction of,
j(j + 1) − `(` + 1) − 43
∆E`s = −En (Zα)2 . (7.95)
2n`(` + 1/2)(` + 1)
Note, that the coupling l · s lifts the degeneracy with respect to l, but not with respect to `z
(see Fig. 7.1). As we have already seen in Exc. 7.5.1.7, in the presence of an energy associated
with the coupling l · s, only the total angular momentum l + s is a constant of motion.
7.2. FINE STRUCTURE OF HYDROGEN-LIKE ATOMS VIA TDPT 181
where f (r0 ) is a radially symmetric and normalized density-type function that takes significant
values only in the vicinity of r within a volume (λC /2π)3 centered at r0 = 0. Expanding the
potential V (r + r0 ) near the origin,
The second term is null due to the parity of f (r0 ) and the third produces the Darwin correction
using V (r) = V (r). Letting the function be constant within the volume, f (r) ' f0 , and with
the normalization,
Z ~/2me c Z ~/2me c Z ~/2me c 3
~
1= f (r)dxdydz = f0 , (7.99)
−~/2me c −~/2me c −~/2me c me c
Also,
Ze %(r) Ze2 δ 3 (r)
∇2 V (r) = −e∇2 = −e =− . (7.101)
4πε0 r ε0 ε0
Hence,
Z
Ze2 π~2 Ze2 3
f (r0 )V (r + r0 )d3 r0 = − + δ (r) + ... , (7.102)
4πε0 r 2m2e c2 4πε0
which is precisely the electrostatic energy with the Darwin correction in the expressions (7.60)
and (7.73).
10
The smearing out of the electron’s position is also known as Zitterbewegung.
182 CHAPTER 7. THE ELECTRON SPIN AND THE ATOMIC SUBSTRUCTURE
To estimate the importance of this effect we consider the ground state, inserting its wave-
function (4.48),
Z
∗ π~2 Ze2 3 π~2 Ze2 π~2 Ze2 1
hĤdw i = d3 rψ100 (r) 2 2 δ (r)ψ100 (r) = |ψ(0)|2
= .
2me c 4πε0 2m2e c2 4πε0 2m2e c2 4πε0 πa3B
(7.103)
We obtain,
π~2 Ze2 1
Hdw 2m2e c2 4πε0 πa3B ~2 Z α2
= = = ≈ 0.01% . (7.104)
H0 e2 2m2e c2 a2B 2
4πε0 aB
Darwin’s correction does not depend on the angular momentum l nor on nthe spin s, such
that,
∆Edw = hĤdw i = −En (Zα)2 . (7.105)
That is, the levels are now degenerate in j (see Fig.. 7.1)11 . Obviously the levels which are most
affected by relativistic corrections are those with low values of n and `.
The levels are labeled by n`j . For example, the state 3d5/2 has the main quantum number
n = 3, the orbital angular momentum ` = 2, and the total angular momentum j = 5/2.
11
It is interesting that the quantum treatment presented here, including relativistic corrections, coincidentally
agrees with the corrections of Arnold Johannes Wilhelm Sommerfeld,
α2
1 3
En,j = En 1 + − .
n j + 1/2 4n
7.3. HYPERFINE STRUCTURE 183
States with equal j are degenerate. For large n or j the fine structure disappears. The new
energy scheme is shown in Fig. 7.1. We note that, taking into account all relativistic corrections
(but without the Lamb shift), we still have a partial degeneracy of the quantum number j. For
example, the states 2 s1/2 and 2 p1/2 have the same energy. This is a particularity of the hydrogen
atom.
where gp ≡ µI /I is once again a factor taking into account possible corrections between the
classical derivation and quantum mechanics 13 .
12
See [48] p. 1229 and [246] p. 23 for further reading.
13
In fact, the proton factor g is anomalous, gp = 5.58, which reduces the faction µl /µI . For the neutron we
have: gp = −3.83
184 CHAPTER 7. THE ELECTRON SPIN AND THE ATOMIC SUBSTRUCTURE
with µS = − mee s.
The nuclear spin produces, at the position of the electrons, a magnetic vector potential,
~I × r
µ0 µ
Adp (r) = , (7.109)
4π r3
interacting with the angular momentum of the electron L in the form,
e e µ0
ĤLI = A · p̂ = µI × r) · p
(~ (7.110)
me me 4πr3
e µ 0 µN µ0 µ B µN
= gp (I × r) · p = gp L · I ,
me 4πr3 ~ 2πr3 ~ ~
using the definition of Bohr’s magneton.
In addition, the potential vector generated by the nuclear spin produces a magnetic field,
µ0
B=∇×A= µI · r̂)r̂ − µ
[3(~ ~I] , (7.111)
4πr3
as will be shown in Exc. 7.5.3.1 14 . This field interacts with the spin of the electron S of the
form,
µ0
ĤSI = −~
µS · B = − µI · r̂)(~
[3(~ µS · r̂) − (~
µS · µ
~ I )] (7.112)
4πr3
µ 0 µB µN
= gs gp [3(I · r̂)(S · r̂) − (S · I)] ,
4πr3 ~ ~
using equation (7.70).
Combining the two terms (7.110) and (7.112), we obtain,
µ 0 µB µN
ĤJI = ĤLI + ĤSI = gs gp [3(I · r̂)(S · r̂) + L · I − S · I] (7.113)
4πr3 ~ ~
µ0 µB µN
= gs gp N · I ,
4πr3 ~ ~
introducing
N ≡ 3(S · r̂)r̂ + L − S . (7.114)
Defining the complete total angular momentum of the atom
F≡I+J , (7.115)
is useful for calculating the coupling I · J = 21 (F2 − I2 − J2 ). Now, as the coupling L · S is strong,
we project the two angular momenta on the total electronic angular momentum J,
N·J J I·J J
N→ , I→ . (7.116)
|J| |J| |J| |J|
We consider two cases:
14
We present here a simplified calculation short-circuiting the Fermi contact term.
7.3. HYPERFINE STRUCTURE 185
N · S ' S2 , (7.117)
and replace the J with S in the projections (7.116). With this the coupling between the spins
of the electronic layer and the nucleus becomes,
(N · S)(I · S) S2 (F2 − I2 − S2 )
N·I= = . (7.118)
~ |J|
2 2 2~2 |S|2
We calculate
`=0
∆Ehf s = h(S, I)F, mF |ĤJI |(S, I)F, mF i (7.119)
µ0 µ B µ N N·I
= gs gp
4π ~ ~ r3
3
µ0 µB µN
3 8π Z 1
= gs gp F (F + 1) − I(I + 1) − 4 .
8π ~ ~ 3 aB πn3
As an example consider the hyperfine structure of the state 1s1/2 of the hydrogen atom.
With J = I = 12 and Z = n = 1 we obtain,
we can write
L6=0 AJ
∆Ehf s = 2 [F (F + 1) − J(J + 1) − I(I + 1)] . (7.124)
186 CHAPTER 7. THE ELECTRON SPIN AND THE ATOMIC SUBSTRUCTURE
Note, that the J · I-coupling breaks the degeneracy of J in the hydrogen atom, but not of Jz .
We can derive the following interval rule,
Besides the magnetic interaction between the angular momenta of the nucleus and the elec-
tronic shell there is an interaction between the nucleus, when it is not spherically symmetric, and
the shell. This interaction causes deviations from the interval rule and an additional splitting
of the hyperfine states.
1 e2 1 e2
Ĥqud = − − , (7.126)
4π0 |re − rN | 4π0 re
where re is the electronic coordinate and rN is the nuclear coordinate, both having their origin
in the center mass of the nucleus. For re > rN this interaction can be obtained after several
mathematical steps as,
3(I · J)(2I · J + 1) − 2I2 J2
Ĥqud = BJ , (7.127)
2I(I − 1)2J(J − 1)
where BJ is called the constant of the quadrupolar electron-nucleus interaction. With this
expression we can calculate,
3
2 K(K + 1) − 2I(I + 1)J(J + 1)
∆Equd = hIJKmK |Ĥqud |IJKmK i = BJ , (7.128)
2I(2I − 1)2J(2J − 1)
where the constants AJ and BJ depend on the atom and the total electronic angular momentum.
7.4. EXOTIC ATOMS 187
In Excs. 7.5.3.2 and 7.5.3.3 we determine the hyperfine structures of sodium and rubidium
atoms.
ium, except that here the positron is replaced by a muon whose mass is mµ+ = 207me . Leptons
are, according to the present understanding, particles without internal structure. Both systems
are unstable: the two particles annihilate each other producing γ-photons. The energy levels
and orbits of the two particles are similar to that of the hydrogen atom. However, because of
the reduced mass, the frequencies of the spectral lines are less than half of the corresponding
hydrogen lines.
The fundamental state of positronium, like that of hydrogen, has two possible configurations
depending on the relative orientation of the electron and positron spins. The singlet state with
antiparallel spins (S = 0, Ms = 0) is known as para-positronium (p-Ps) and denoted by 1 S0 . It
has an average lifetime of
2~
τ= = 124.4 ps (7.130)
me c2 α5
and decays preferably in two gamma rays with energy of 511 keV each (in the center-of-mass).
The triplet state with parallel spins (S = 1, Ms = −1, 0, 1) is known as ortho-positronium (o-
Ps) and denoted as 3 S1 . It has an average life of 138.6 ns, and the most common form of
decay produces three photons. Other forms of decay are negligible. For example, the decay
channel producing five photons is 10−6 times less likely. Measurements of these lifetimes and
the positronium energy levels have been used in precision tests of quantum electrodynamics.
While the precise calculation of the positronium energy levels is based on the Bethe-Salpeter
equation, the similarity between positronium and hydrogen allows for an approximate estimate.
In this approach, the energy levels are suppodsed to be different from those of hydrogen because
of the difference in the value of the reduced mass µ, used in the energy equation. Since µ = me /2
for positronium, we have
A di-positronium molecule, that is, a system of two bound positronium atoms, has already
been observed. Positronium in high energy states has been conjectured to become the dominant
form of atomic matter in the universe in the very distant future if the proton decay becomes
tangible.
e2
Ucou = − . (7.137)
4πε0 r
The similarity of the effective potential ’seen’ by the outer electron and the authentic hydrogen
potential suggests a classical treatment within the planetary model. There are three notable
exceptions:
• An atom can have two (or more) electrons in highly excited states with comparable orbital
radii. In this case, the electron-electron interaction gives rise to a significant deviation from
the hydrogen potential. For an atom in a multiple Rydberg state the additional term Uee
includes a sum over each pair of highly excited electrons:
e2 X 1
Uee = . (7.138)
4πε0 |ri − rj |
i<j
• If the valence electron has very low angular momentum (interpreted classically as an
extremely eccentric elliptical orbit), it can pass close enough to the nucleus to polarize it,
giving rise to an additional term,
e 2 αd
Upol = − . (7.139)
(4πε0 )2 r4
7.5. EXERCISES 191
• If the outer electron penetrates the inner electronic shells, it sees more of the charge of
the nucleus and therefore feels a larger force. In general, the modification of the potential
energy is not simple to calculate and should be based on some knowledge of the nucleus’
geometry.
E1
EB = − . (7.140)
n2
The binding energy is weak at high values of n, which explains the fragility of the Rydberg
states that can easily be ionized, e.g. by collisions.
Additional terms modifying the potential energy of a Rydberg state require the introduction
of a quantum defect, δ`, in the expression for the binding energy:
E1
EB = − . (7.141)
(n − δ` )2
The long lifetimes of Rydberg states with high orbital angular momentum can be explained
in terms of overlapping wavefunctions. The wavefunction of an electron in a state with high
` (large angular momentum, ’circular orbit’) has little overlap with the wavefunctions of the
internal electrons and therefore stays relatively unperturbed. Also, the small energy difference
between adjacent Rydberg states decreased the decay rate according to the result (??).
ES = −d · E . (7.142)
Depending on the sign of the projection of the dipole moment onto the vector of the local electric
field, the energy of a state increases or decreases with the intensity of the field. The narrow
spacing between adjacent levels n in the Rydberg series means that the states can approach
degeneracy even for relatively weak fields. Theoretically, the force of the field in which a level
crossing would occur (assuming no coupling between the states) is given by the Inglis-Teller
limit,
e
FIT = . (7.143)
12πε0 a20 n5
In hydrogen the pure Coulomb potential does not couple the Stark states of an n level, which
results in a real crossover. In other elements, deviations from the ideal 1/r-potential allow for
avoided crossings.
7.5 Exercises
7.5.1 Fine structure and the Dirac equation
7.5.1.1 Ex: Yukawa potential
Show, that Yukawa’s potential satisfies the Klein-Gordon equation.
192 CHAPTER 7. THE ELECTRON SPIN AND THE ATOMIC SUBSTRUCTURE
Show that the expressions for the Darwin correction (7.60) and (7.73) are equivalent.
Calculate the magnetic field generated by the orbiting proton as it is perceived by the electron.
a. Calculate the vector potential A(r) and the magnetic dipole moment µ ~ produced by an
0 −1
orbiting electron by Biot-Savart’s law using the expansion of |r − r | in Legendre polynomials.
b. Calculate the magnetic field B(r).
Determine the hyperfine structure of the 2 S and 2 P states of the sodium atom including energy
shifts. See Tab. 7.2 for the hyperfine constants AJ e BJ .
Given the following energy distances νF,F 0 of the hyperfine levels of the rubidium isotopes 87 Rb
and 85 Rb [21],
87
Rb, S1/2 splits into ν1,2 = 6834.7 MHz
87
Rb, P3/2 splits into ν0,1 = 72.3 MHz, ν1,2 = 157.1 MHz, ν2,3 = 267.2 MHz
85
Rb, S1/2 splits into ν1,2 = 3035.7 MHz
85
Rb, P3/2 splits into ν1,2 = 29.4 MHz, ν2,3 = 63.4 MHz, ν3,4 = 120.7 MHz ,
F = qE + q ṙ × B , (8.1)
where
∂A
E = −∇Φ − and B=∇×A , (8.2)
∂t
where Φ and A are called scalar and vector potential, respectively.
As we learned in electrodynamics it is possible to derive this force from a Lagrangian for the
electronic motion,
m
L(ri , ṙi ) = ṙ2 − qΦ(r) + q ṙ · A(r) . (8.3)
2
With this aim we first determine the momentum by
∂L
pi = = mṙi + qAi , (8.4)
∂ ṙi
and the Hamiltonian by,
X m m
H= pi ṙi − L(ri , ṙi ) = (mv + qA) · ṙ − ṙ2 + qΦ − q ṙ · A = v2 + qΦ . (8.5)
i 2 2
That is,
1
H(ri , pi ) = (p − qA)2 + qΦ . (8.6)
2m
The following equations hold,
∂H ∂H
ṙi = and ṗi = − . (8.7)
∂pi ∂ri
The first equation is easily verified by inserting the Hamiltonian (8.5). The second leads to the
Lorentz force,
∂H
Fi = mv̇i = ṗi − q Ȧi = − − q Ȧi = qEi + q(v × B)i , (8.8)
∂ri
where the last step of the derivation will be shown in the Exc. 8.4.1.1 using the Coulomb gauge
∇ · A = 0.
195
196 CHAPTER 8. ATOMS WITH SPIN IN EXTERNAL FIELDS
mv −→ p − qA and H −→ H + qΦ . (8.9)
This substitution rule, called minimal coupling, can be applied in quantum mechanics,
The fourth term called diamagnetic term is quadratic in A and usually so small that it can
be neglected. The second and third terms describe the interaction of the electron through its
momentum p̂ with the potential vector A produced by magnetic moments inside the atom or
outer magnetic fields. Within the Coulomb gauge we have (∇ · A)ψ = (A · ∇)ψ + ψ(∇ · A) =
(A · ∇)ψ, such that,
e
Ĥint = mA · p̂ . (8.12)
Thus the interaction energy between the electron and the field is given by the Hamiltonian
(8.12),
i~e i~e i~e
Vzee (B) = − A·∇=− (B × r) · ∇ = − B · (r × ∇) (8.14)
m 2m 2m
e µB µB
= − 2m B · L̂ = − gL L̂ · B = −~ µL · B = − L̂z B ,
~ ~
e
with gL = 1 using the relation µ ~ L = 2m L between the angular momentum of the electron and
the resulting magnetic moment. This relationship holds for an atom without spin (two electrons
can couple their spins to a singlet state) and no hyperfine structure (or an unresolved hyperfine
structure). The energies are therefore,
µB
∆Ezee (B) = − Bhn, L, mL |L̂z |n, L, mL i = −µB mL B . (8.15)
~
In the Excs. 8.4.2.1 and 8.4.2.2 we represent the interaction between an atomic angular momen-
tum and a magnetic field in different bases characterized by different quantization axes.
8.2. INTERACTION WITH MAGNETIC FIELDS 197
before projecting the result over the B field. The potential is,
µB µB h J
J
J
J
i
Vzee (B) = −~µJ · B = − (L + 2S) · B −→ − L · |J| · B + 2 S · · B
~ ~ |J| |J| |J|
µB µB 1 2
=− [L · J + 2S · J] J · B = − J + L2 − S2 + 2(J2 + S2 − L2 ) J · B
~|J| 2 ~|J| 2
2
2
µB 1 3J − L + S 2 2
=− J·B . (8.18)
~ 2 |J|2
we can write
µB
∆Ezee (B) = − gJ hJˆz Bi = −µB gJ mJ B . (8.21)
~
This expression describes the anomalous Zeeman effect, for which S 6= 0. For the normal Zeeman
effect, for which the spin is zero, we find again gJ = 1.
Therefore,
µB µB h B
B
B
B
i
Vpb (B) = − (L̂ + 2Ŝ) · B −→ − L̂ · + 2 Ŝ · ·B , (8.23)
~ ~ |B| |B| |B| |B|
198 CHAPTER 8. ATOMS WITH SPIN IN EXTERNAL FIELDS
Figure 8.1: Coupling angular moments for the effect (a) Normal Zeeman effect, (b) anoma-
lous Zeeman effect, (c) Paschen-Back effect, (d) Zeeman effect of the hyperfine structure, and
(e) Paschen-Goudsmith effect.
such that
∆Epb (B) = −µB (mL + 2mS )B . (8.24)
This is the Paschen-Back effect.
The derivations we have made so far have focused on simple situations well described by
CSCOs in various coupling schemes. The projections on the different quantization axes [the
total spin (8.17) in the Zeeman case or the applied magnetic field (8.22) in the Paschen-Back
case] ensure that the Hamiltonians Vls and Vzee (B) in these CSCOs are described by diagonal
matrices. However, in regimes intermediate between Zeeman and Paschen-Back, Vls ' Vzee (B),
it is generally not possible to find a diagonal representation.
In order to calculate the energy spectrum in intermediate regimes we must, therefore, deter-
mine all the components of the matrix,
µB
Vls + Vzee (B) = ξ(r)L̂ · Ŝ + (L̂ + 2Ŝ) . (8.25)
~
Using L̂± ≡ L̂x ± iL̂y and Ŝ± ≡ Ŝx ± iŜy , we can easily rewrite the energy in the following way,
µ
B
Vls + Vzee (B) = ξ(r) L̂z Ŝz + 12 L̂+ Ŝ− + 21 L̂− Ŝ+ + (L̂ + 2Ŝ) · B . (8.26)
~
This operator acts on the uncoupled states,
∆Els + ∆Ezee (B) = hL0 m0L ; S 0 m0S |ξnl (L̂z Ŝz + 12 L̂+ Ŝ− + 12 L̂− Ŝ+ ) + µB (L̂z + 2Ŝz )B|LmL ; SmS i
= ~2 ξnl mL mS δmL ,m0L δmS ,m0S + 21 L+ S− δmL ,m0L −1 δmS −1,m0S + 21 L− S+ δmL −1,m0L δmS ,m0S −1
+ ~µB (mL + 2mS )BδmL ,m0L δmS ,m0S , (8.27)
p
with the abbreviations L± ≡ L(L + 1) − mL (mL ± 1). The energies are now the eigenvalues of
this matrix. The factor ξnl is usually determined experimentally by letting B = 0. In Exc. 8.4.2.3
we calculate the re-coupling of the spins of two electrons in an external magnetic field.
Figure 8.2: Transition between the Zeeman regime and the Paschen-Back regime for the case
L = 1 and S = 1/2.
we project the nuclear spin and the total electronic momentum separately in the direction of F,
F F F F
J −→ J · |F| |F| and I −→ I · |F| |F| . (8.29)
µB F2 + J2 − I2 µN F2 − J2 + I2
Vzee (B) = − gJ BFz + gp BFz , (8.32)
~ 2|F| 2 ~ 2|F|2
such that
∆Ehf s + ∆Ezee (B) ' ∆Ehf s + µB gF mF B , (8.33)
The Zeeman effect of the hyperfine structure becomes a hyperfine structure of the Zeeman effect,
also called Paschen-Back effect of the hyperfine structure or Paschen-Goudsmith effect. We
can diagonalize the potential on a basis, where I, mI , J, and mJ are good quantum numbers.
Using the expression (7.129) but disregarding the quadrupolar contribution to the hyperfine
interaction, BJ ' 0, we obtain,
AJ
Vhf s + Vzee (B) = Vhf s − (~ ~I) · B ' 2 J · I + µ
µJ + µ ~J · B (8.36)
~
AJ B B B B AJ
−→ 2 J · · I· + µJz B = 2 Jz Iz + µJz B ,
~ B B B B ~
where we neglect the interaction of the dipole moment of the nucleus with the external magnetic
~ I ' 0. We obtain,
field, µ
Here, we project the angular momentum J and the spin I separately onto the direction of the
magnetic field. The re-coupling of the state |F mF i to |mI mJ i in strong magnetic fields is
described by Clebsch-Gordan coefficients,
X
|F mF i = |mI mJ ihmI mJ |F mF i . (8.38)
mI +mJ =mF
We develop the complete matrix representation of this Hamiltonian within the puncoupled
base, where mp J , mI are good quantum numbers, introducing the abbreviations I± ≡ I(I + 1) − mI (mI ± 1)
and I±± ≡ I(I + 1) − (mI ± 1)(mI ± 2). The SU(2) algebra provides useful expressions,
I · J = Iˆz Jˆz + 12 (Iˆ+ Jˆ− + Iˆ− Jˆ+ ). The elements of the matrix are,
where
I·J 2
hm0I m0J | ~ |mI mJ i = (mI mJ )2 + 14 I−
2 2 2 2
J+ + 41 I+ J− δm0I mI δm0J mJ + (8.41)
+ 12 m0I m0J + mI mJ I+ J− δm0I mI +1 δm0J mJ −1 +
+ 12 m0I m0J + mI mJ I− J+ δm0I mI −1 δm0J mJ +1 +
+ 14 I+ J− I++ J−− δm0I mI +2 δm0J mJ −2 +
In this form the matrix can be programmed, e.g. using computational software such as
MATLAB, and all eigenvalues of the Hamiltonian for any state 2S+1 XJ and nuclear spin I can
be calculated numerically. Obviously, the eigenvalues follow from a diagonalization of the matrix
and do not depend on the chosen base. Fig. 8.3 shows the result obtained for 6 Li (I = 32 ) in the
state 2P3/2 knowing that AJ /h = −1.17 MHz and BJ = 0.
20
10
∆ν (MHz)
-10
-20
0 2 4
B (G)
Figure 8.3: (Code: AM Staticf ields HiperLithium.m) Hyperfine and Zeeman structure of the
state 2 P3/2 of 6 Li.
It is interesting to analyze the so-called fully stretched states. In this states, the interval
factor is A = Vhf s /2. By inserting F = I + J into the first formula, we obtain,
AJ 3(I + J)
Vhf s +Vzee (B) = I·J−BJ +gJ µB B·J−gI µN B·I ≈ AJ mJ mI +µB gJ mJ B .
~2 2(2I − 1)(2J − 1)
(8.42)
The energy displacement of the fully stretched√states is linear in the magnetic field. We can also
look at the matrix elements I+ = 0 and I− = 2I and note that all non-diagonal terms vanish.
When one of the spins, J or I, is equal to 1/2 only two possible hyperfine states exist:
F = I ± J, which are necessarily totally stretched. For this case there is an approximate
analytic formula called the Breit-Rabi formula [17], which will be derived in Exc. 8.4.2.4,
∆Ehf s + ∆Ezee (B) = h A~2J I · J + gJ µB B · J − gI µN B · Ii (8.43)
r
AJ AJ (I + 12 ) 4mF
=− + µN gN mF B ± 1+ x + x2 ,
4 2 2I + 1
202 CHAPTER 8. ATOMS WITH SPIN IN EXTERNAL FIELDS
B = Bêz (8.44)
reproduces the field (8.44). However, we have the freedom in choosing the potential vector,
given by the gauge transformation, to add the gradient of a scalar field, for example,
By
1
1
χ ≡ 2 Bxy =⇒ ∇χ = Bx =⇒ A0 ≡ A + ∇χ = Bxêy . (8.47)
2
0
The potential vector A0 gives the same magnetic field and only changes the general phase of the
wave function, but the physical properties do not change. In this gauge, which is called Landau
gauge, the Hamiltonian is,
p̂2x 1 qB x̂ 2
Ĥ = + p̂y − . (8.48)
2m 2m c
The operator p̂y commutes with this Hamiltonian, since the y operator is absent due to the
choice of the gauge. Thus, the operator p̂y can be replaced by its eigenvalue ~ky . Hence, by
introducing the cyclotron frequency, ωc = qB/mc, we obtain,
2
p̂2 mωc2 ~ky
Ĥ = x + x̂ − . (8.49)
2m 2 mωc
This is exactly the Hamiltonian of the quantum harmonic oscillator, except that the minimum of
the potential is displaced in position space by the value x0 = ~ky /mωc . To find the energies, we
note that the translation of the potential of the harmonic oscillator does not affect the energies.
The energies of this system are therefore identical to those of the standard quantum harmonic
oscillator,
En = ~ωc n + 12 , (8.50)
for n ≥ 0. Since the energy does not depend on the quantum number ky , we will have degen-
eracy. For the wavefunctions, we remember that p̂y commutes with the Hamiltonian. Then the
8.3. INTERACTION WITH ELECTRIC FIELDS 203
wavefunction splits into a product of eigenstates of the momentum in y-direction and eigenstates
of the harmonic oscillator |φn i shifted by a value x0 in x-direction:
That is, the state of the electron is characterized by two quantum numbers, n and ky .
Each set of wavefunctions with the same n is called Landau level. Effects due to Landau levels
are only observed when average thermal energy is lower than the separation of energy levels,
which means that low temperatures and strong magnetic fields are required. Each Landau level
is degenerate because of the second quantum number ky , which can adopt the values,
2πN
ky = , (8.52)
Ly
with N ∈ N. The allowed values of N are more restricted by the condition that the center of
force of the oscillator, x0 , must be physically inside the system, 0 ≤ x0 < Lx . This gives the
following range for N ,
mωc Lx Ly
0≤N < . (8.53)
2π~
For particles with charge q = Ze, the upper limit in N can simply be written as a ratio of
fluxes,
ZBLx Ly Φ
=Z , (8.54)
(hc/e) Φ0
where Φ0 = h/2e is the fundamental flux quantum and Φ = BA the flux through the system
(with area A = Lx Ly ). Thus, for particles with spin S, the maximum number of particles per
Landau level D is,
Φ
D = Z(2S + 1) . (8.55)
Φ0
In general, Landau levels are observed in electronic systems with Z = 1 and S = 1/2. As the
magnetic field increases, more and more electrons can fit a certain Landau level. The occupation
of the highest Landau level ranges from entirely full to entirely empty, leading to oscillations in
various electronic properties (see Haas-van Alphen effect, Shubnikov-de Haas effect and quantum
Hall effect).
This is the Stark effect. This effect is usually weak, and its observation requires strong fields or
high spectral resolution. Stationary perturbation theory TIPT gives,
Z
(1) (0) (0)
En = hψn | − d · E|ψn i = eEz · z|ψn(0) |2 d3 r = 0 . (8.57)
R3
This only applies when the states have well-defined parity and are NOT degenerate in `. When
they ARE degenerate in `, which is the case of hydrogen, the states have no defined parity. For
204 CHAPTER 8. ATOMS WITH SPIN IN EXTERNAL FIELDS
example, the states s and p contributing to the same state |ψn,j i have different parities. In this
case, the condition (8.57) does not have to be satisfied, and the first perturbation order yields a
value. This is the case of the linear Stark effect.
Other atoms do not have this degeneracy, and we must calculate the quadratic Stark effect
in second order TIPT,
X (0) hψ (0) (0)
0 |ẑ|ψn i
|ψn(1) i = eEz |ψn0 i n . (8.58)
0
En − En0
n 6=n
and
X |hψ (0) (0) 2
n0 |ẑ|ψn i|
En(2) = e2 Ez2 . (8.59)
0
En − En0
n 6=n
To simplify the matrix elements, we separate the radial part from the angular part,
Z ∞ Z
(0) 0 0 0
(0)
hψn0 |ẑ|ψn i = hn J mJ |ẑ|nJmJ i = r Rn0 J 0 RnJ dr YJ∗0 m0 zr YJmJ dΩ .
3
(8.60)
J
0
and called the irreducible matrix element, no longer depends on the magnetic quantum number.
On the other hand, the angular part may be expressed by Clebsch-Gordan coefficients, as will
be discussed more extensively in Sec. ??. The result is called Wigner-Eckart theorem,
|hn0J J 0 m0J |ẑ|nJ JmJ i|2 1 J 1 J0
= . (8.62)
hn0 J 0 ||ẑ||nJ Ji2 2J 0 + 1 mJ 0 −m0J
J
8.4 Exercises
8.4.1 Charged particles in electromagnetic fields
8.4.1.1 Ex: Lagrangian of an electron in the electromagnetic field
a. Show that the Lagrangian (8.3) reproduces the Lorentz force (8.1).
b. Show that the Hamiltonian (8.5) reproduces the Lorentz force (8.1).
between Zeeman and Paschen-Back? We start by calculating the Zeeman shifts for both levels
(s denotes the structure S1/2 , p the structure P3/2 )
s s
B hmJ mI |Hhf s + HB |msJ msI iB = E s (B)
p p
B hmJ mI |Hhf s + HB |mpJ mpI iB = E p (B) .
For the level P3/2 the interval factor is less. In particular for 6 Li it is so small that we are
immediately in the Paschen-Back scheme. This means that the matrix ∞ hm̃pJ m̃pI |mpJ mpI iB =
δmp ,m̃p δmp ,m̃p is diagonal. The element of the transition matrix is then,
J J I I
p p (Eκ)
X X p p p p p p (Eκ)
B hmJ mI |Tq |msJ msI iB = s s s s
∞ hm̃J m̃I |mJ mI iB ∞ hm̃J m̃I |mJ mI iB ∞ hm̃J m̃I |Tq |m̃sJ m̃sI i∞
m̃sJ m̃sI m̃pJ m̃pI
X p p (Eκ)
s s s s
= ∞ hm̃J m̃I |mJ mI iB ∞ hmJ mI |Tq |m̃sJ m̃sI i∞ .
m̃sJ m̃sI
The matrix elements in the pure Zeeman regime can be expressed by [Deh07, unpublished],
hF p mpF |Tq(Eκ) |F s msF i =0 mpJ mpI Tq(Eκ) |msJ msI i0
s 2 p 2
J κ Jp J J s κ (2F s + 1)(2J p + 1)(2κ + 1)
= .
msJ sign(mp − ms ) −mpJ Fs Fp I 2I + 1
p p (Eκ)
Discuss the pure Paschen-Back regime via ∞ hmJ mI |Tq |msJ msI i∞ .
a. Calculate the quadratic Zeeman effect for the ground state of the hydrogen atom caused by
the (usually neglected) diamagnetic term of the Hamiltonian in first order TPIT. Write down
the energy shift as ∆E = − χ2 B 2 in order to obtain the diamagnetic susceptibility χ.
b. Estimate the energy shift due to diamagnetism of the ground state of helium using the orbital
3
Zef f −Zef f (r1 +r2 )/aB 5
ψ(r1 , r2 ) = πa3B
e with Zef f = 2 − 16 obtained by the variational method. The
measured value is χ = 1.88 × 10−6 cm3 /mol.
Using the Hamiltonian for an atomic electron in a
magnetic field determine, for a state with angular orbital momentum 0, the energy shift in B 2
when the system is in a uniform magnetic field represented by the potential vector A = 12 B × r.
What is diamagnetic susceptibility χ for a helium atom in the ground state? Compare with the
measured value.
Consider the hydrogen atom immersed in a uniform electric field E applied along the êz -direction.
The term corresponding to this interaction in the total Hamiltonian is H (1) = −eEz. For typical
electric fields produced in laboratory, the condition H (1) H0 , which allows the use of TIPT, is
satisfied. The effect of the perturbation H (1) , called Stark effect, is the removal of the degeneracy
of some of the hydrogen atom states. Calculate the Stark effect for the state n = 2 of the
hydrogen atom.
8.4. EXERCISES 207
but this is not valid for na 6= nb . If na = nb , we have ψa = ψb . That is, the particles stay in
the same state and we do not need to worry about indistinguishability:
However, the fact that this state is never observed with two electrons shows, that theory
must be corrected to allow a true description of reality.
We need to construct the total wavefunction in another way. Let us consider linear combi-
nations of Ψ(1,2) ,
ΨS,A ≡ √1 (Ψ(1,2)
2
± Ψ(2,1) ) = √1
2
[ψa (x1 )ψb (x2 ) ± ψa (x2 )ψb (x1 )] . (9.1)
1
We note that states are orthogonal because
Z Z Z Z
Ψ∗(1,2) Ψ(2,1) dx1 dx2 = ψa∗ (x1 )ψb∗ (x2 )ψa (x2 )ψb (x1 )dx1 dx2 = ψa∗ (x1 )ψb (x1 )dx1 ψb∗ (x2 )ψa (x2 )dx2 = δna ,nb .
209
210 CHAPTER 9. ATOMS WITH MANY ELECTRONS
This symmetrized wavefunction (or anti-symmetrized) represents a trick to eradicate the label
sticking to the particles. For, under particle exchange described by the operator Px ψa (x1 )ψb (x2 ) ≡
ψa (x2 )ψb (x1 ), the (anti-)symmetrized functions behave like 2 ,
That is, for a symmetric system, the probability of finding two particles at the same location is
doubled, whereas for an anti-symmetric system, this probability is zero.
Wolfgang Pauli showed that the (anti-)symmetric character is related to the spin of the
particles. Particles with integer spin called bosons must be symmetric. Particles with semi-
integer spin called fermions must be anti-symmetric. Electrons are fermions. Therefore, in an
atom, they can not be in the same state (location), but must be distributed over a complicated
shell of orbitals. We note, that this applies not only to elementary particles, but also to composed
particles such as, for example, atoms. We will determine in Exc. 15.3.1.1 the bosonic or fermionic
character of several atomic species.
Px,s Ψ(1,2) ≡ Px,s ψa (x1 , s1 )ψb (x2 , s2 ) = ψa (x2 , s2 )ψb (x1 , s1 ) = Ψ(2,1) . (9.5)
We now assume that the electrons not only do not interact with each other, but there is
also no interaction between the position and the spin of each electron. That is, we will discard
L · S-coupling.3 We can then write the total wavefunction of an electron as the product of a
spatial function, ψ(x), and a spin function, χ(s) = α ↑ +β ↓, where α and β are probability
amplitudes of finding the electron in the respective spin state, such that,
Ψ = {ψa (x1 )ψa (x2 ), ψa (x1 )ψb (x2 ), ψb (x1 )ψa (x2 ), ψb (x1 )ψb (x2 )} (9.14)
with the same probability of p = 1/4. Bosonic indistinguishable particles can stay in one of the
following three states,
Ψ = {ψa (x1 )ψa (x2 ), √12 [ψa (x1 )ψb (x2 ) + ψb (x1 )ψa (x2 )], ψb (x1 )ψb (x2 )} (9.15)
with the same probability of p = 1/3. Finally, fermionic indistinguishable particles can only be
in a state,
Ψ = { √12 [ψa (x1 )ψb (x2 ) − ψb (x1 )ψa (x2 )]} (9.16)
with the same probability of p = 1. We see that a simple two-particle system already exhibits
qualitative modifications of its statistical behavior. These differences generate different physics
as we deal with systems of large numbers of particles, as we can see in the cases of the free
electron gas and the Bose-Einstein condensate. For a broader discussion, see Chap. ??.
We finally note a result of the standard model of particle physics assigning a fermionic
character to all fundamental constituent particles of matter while the mediators of fundamental
forces are always bosons.
9.2 Helium
The simplest atom to discuss Pauli’s principle is helium. The helium atom has a charged nucleus
Z = +2e and mass mHe ≈ 4mH .
and we get two Schrödinger equations, the Hamiltonian being equal to the one of hydrogen-like
atoms: 2
~ 2 e2 Z
− ∇i − Ψi (ri ) = En(i) Ψi (ri ) , (9.18)
2µ 4πε0 ri
with i = 1, 2. For hydrogen-like atoms we have,
1 1
E = En(1) + En(2) = EB Z 2 + , (9.19)
n21 n22
with EB = −13.6 eV. With this, we get the energy for the ground state:
The value predicted by Bohr’s model is far from experimental reality: The ionization energy
measured for the first electron is 24.6 eV, for the second 54.4 eV, totalizing a binding energy for
two electrons of −78.983 eV . This corresponds to an error of about 38%. The lower energy of
the first electron is due to the shielding of the nucleus by the second.
Let us now investigate the excited states of helium, in particular those, where only one electron
is excited, the other one being in the ground state, (1s)1 (2s)1 and (1s)1 (2p)1 . The energy of
the electron in the shell n = 2 is smaller than predicted by Bohr’s model with Z = 2 because of
the interaction with the other electron. Also, the (2s) and (2p) levels are no longer degenerate,
because the electrostatic potential is no longer Coulombian (see Fig. 9.1).
Figure 9.1: Helium levels and allowed singlet and triplet transitions. Note that the state (1s)↑↑
does not exist.
As we have seen in the discussion of the fine structure, the energy of the l · s-coupling
given by (??) is ∝ En (Zα)2 ∝ Z 4 . For helium which still has a small Z, the energy of the
coupling is weak, so that we can count on a direct coupling of the two spins. Since the orbits of
the electrons are now different, we can construct combinations of symmetric or antisymmetric
spatial wavefunctions ΨS,A , are therefore combinations X A,S of anti-parallel or parallel spins.
When the spins are parallel (S = 1), the spatial wavefunction is antisymmetric, when they are
antiparallel (S = 0), it is symmetric. From the symmetry of the wavefunction depends the
energy of the coulombian interelectronic interaction, because in the symmetric state the average
distance of the electrons is much smaller than in the antisymmetric state, where the total spatial
function disappears for zero distance. Consequently, the configuration (1s)1 (2s)1 has two states
with S = 0.1, with energy ES=0 > ES=1 . Likewise, all configurations are split, as shown in
Fig. 9.1. The energy difference (∼ 1 eV) is considerable and well above the energy of fine
structure interaction (∼ 10−4 eV). This explains why the two spins first couple to a total spin,
s1 + s2 = S, before this spin couples to the total orbital angular momentum, S + L = J. This
is the L · S-coupling.
The energy difference between the two states S = 0.1 is called exchange energy. It comes out of a
first-order perturbation calculation. For example, for the (1s)1 (2s)1 we write the anti-symmetric
wavefunctions,
ΘA
± =
√1
2
[ψ100 (r1 )ψ200 (r2 ) ± ψ100 (r2 )ψ200 (r1 )] · χA,S , (9.26)
9.2. HELIUM 215
where the (+) sign holds for χA (S = 0) and the (−) sign for χS (S = 1). The energies are,
Z Z
e2
∆E S,A = 21 dr13 dr23 Θ∗A± ΘA,± (9.27)
4π0 |r1 − r2 |
Z Z
e2
= 2 dr1 dr23
1 3
[|ψ100 (r1 )|2 |ψ200 (r2 )|2 + |ψ100 (r2 )|2 |ψ200 (r1 )|2 ]
4π0 |r1 − r2 |
Z Z
e2
± 2 dr1 dr23
1 3
2ψ ∗ (r1 )ψ200 ∗
(r2 )ψ100 (r2 )ψ200 (r1 )
4π0 |r1 − r2 | 100
≡ ∆Ecoulomb ± ∆Eexchange .
yielding,
E (1) = J ± K .
That is, the states that were previously degenerate with energy E = Ea + Eb are now split
into two states with energies E = Ea + Eb + J ± K. And the corresponding eigenfunctions
are:
1
ΨS,A (1, 2) = √ [ψa (1)ψb (2) ± ψb (1)ψa (2)] .
2
This result shows that the repulsion between the two electrons breaks the degeneracy (of
separable functions written in product form) into states with an energy difference 2K. Note
that the eigenfunctions are symmetric, which is discussed in the next section.
So far we have seen that, if the electrons are in the same orbital, we have an energy term
E = 2Ea + J and, when they are in different orbitals, we have E = Ea + Eb + J ± K, with a
separation between levels of 2K.
In practice, we consider only the excitation of one electron, because the energy to excite the
two electrons exceeds the ionization energy of the helium atom. To find the selection rules for
transition between symmetric and antisymmetric states, we calculate the dipole moment of the
transition. For a two-electron system the dipole moment is d = −er1 − er2 , which is symmetric
with respect to a permutation of the two electrons. The dipole moment for transition is:
Z
d± = −e Ψ∗A (r1 , r2 )(r1 + r2 )ΨS (r1 , r2 )dV1 dV2 . (9.31)
If we exchange the electrons, the above integral changes sign, but the integral can not depend
on the nomenclature of the integration variables, so it must be zero. The transition between
a symmetric and an antisymmetric state can not occur. Considering the spin wavefunction,
Θ = ΨS χA or ΨA χS , we find that transitions are only allowed between singlet states or between
triplet states. That is, there is a selection rule for the spin postulating ∆S = 0 6,7 .
Because of the differences observed in the singlet and the triplet spectrum of helium, il-
lustrated in Fig. 9.1, it was first believed that they belong to different atomic species, called
para-helium e ortho-helium. A chemical analysis showed later that it was the same element.
6
Moreover, transitions between the states 1 S0 and 3 S1 are impossible, because they violate the selection rule
for the angular momentum, ∆L = ±1.
7
We can understand the selection rules as follows: As long as the wavefunction can be written as a product,
Θ = Ψ(x)χ(s), the symmetry character is preserved for the two functions separately. The eigenvalues of the
operators Px and Ps are then good quantum numbers. But this only holds for weak L · S-coupling. The electric
dipole operator for the transition does not act on the spin (which prevents the recoupling S = 1 ↔ S = 0
via E1-radiation) and also does not act on the symmetry character of the orbitals (which prevents transitions
ΨS ↔ ΨA ).
In principle, this holds for any species of atoms with two valence electrons. In reality however, the influence of
the L · S-coupling grows with Z, which weakens the interdiction of the intercombination transition. In this case,
only the operator Px,s yields good eigenvalues.
9.3. ELECTRONIC SHELL STRUCTURE 217
is extremely complicated to solve, even for the simplest case (Z = 2) we must use approximation
methods.
Note that, if we assume independent electrons (Vele−ele = 0), that is, each electron moves
independently of the others within the electrostatic potential generated by the nucleus and the
other Z − 1 electrons, the problem would be easily solvable: We would solve the Schrödinger
equation for a product state of all the electronic wavefunctions, and we would know the eigen-
functions and individual eigenenergies of each electron (as for the hydrogen atom). In principle,
we should use antisymmetric wavefunctions, but as a first approach we can choose to only respect
Pauli’s weak principle, that is, assign an individual and unique set of quantum numbers to each
electron. The total energy would be the sum of the energy of every electron, and the associated
physical eigentates would be obtained by means of an antisymmetrization of the tensor product
of the multielectronic state.
Example 37 (Fermi gas): We consider an infinite potential well, that we gradually fill up
with electrons. The Pauli principle allows us to place at most two electrons in each orbital,
This total wavefunction satisfies the weak Pauli principle, but is obviously not antisymmet-
ric. The approximation is good, when the interaction between the electrons is negligible.
Otherwise, we need to consider the exchange energy terms.
This model, called Fermi gas model, is often used to describe the behavior of electrons that
can freely move within the conductance band of a metal. Let N be the number of electrons
placed in the potential successively filling all the orbitals with quantum number n up to
the maximum number nmax = N/2, provided each orbital can withstand two electrons with
antiparallel spins. For a one-dimensional well of size L, the maximum energy, called Fermi
energy is,
N2
EF ∝ 2 .
L
For a three-dimensional well of volume V ,
2/3
N
EF ∝ .
V
potential which, when subdivided into small volumes filled with electrons, produces the same
charge distribution. This principle is called self-consistency.
The wavefunctions (spatial distributions of electrons) determined by this method often serve
as a starting point for the Hartree method discussed below. One of the important predictions of
the Thomas-Fermi model is that the average radius of an atom depends on the nuclear charge
as R̄ ∝ Z −1/3 .
The Thomas-Fermi model allows us to understand the electronic configuration of the fun-
damental states and provides the basis for the periodic system of elements. In this model, the
electrons are treated as independent particles, on one side forming an effective radial electric
potential, on the other side being subjected to this potential. Instead of requiring anti-symmetry
of the wavefunction, it is only necessary to ensure that all electrons are distinguished by at least
one quantum number. The functions of complex atoms are similar to the functions of hydrogen.
We can use the same quantum numbers n, `, m` , and ms for every electron.
However, the effective radial potential depends very much on the species and is quite different
from the Coulomb potential. So the degeneracy in ` is lifted. In general, electrons with small
` are more strongly bound, because they have a higher probability of being near the nucleus,
where the potential is deeper (see Fig. 9.7). The same argument explains why electrons with
small n are more strongly bound. We will discuss these effects in more depth in Sec. 9.4.2 by
comparing the excitation levels of the valence electron in different alkalis.
Z 2 e2 Z(Z − 1)e2
Vncl−ele ' and Vele−ele ' , (9.33)
aB 2aB
we realize that Vele−ele /Vncl−ele varies between 14 for Z = 2 and 21 for Z 1/2. For this reason
the use of alternative methods to describe multielectronic atoms is necessary.
The Thomas-Fermi model is a semi-classical model that aims to roughly describe the total
energy of the electrons as a density functional of atomic/molecular electrons. Such a model
is the basis for more sophisticated methods to obtain the electronic structure, such as density
functional theory (DFT).
π 2 ~2
Enx ,ny ,nz = (n2 + n2y + n2z ) . (9.34)
2me L2 x
For N electrons in the ground state, each particle occupies the energetically lowest state available,
obeying the Pauli exclusion principle and considering the spin. The total energy is given by the
sum of the energies of the N less energetic states, and the final physical state is given by the
9.3. ELECTRONIC SHELL STRUCTURE 219
antisymmetrization of the corresponding state. The energy of the N -th electron (the most
energetic one) is called the Fermi energy (EF ) 8 .
It is interesting to calculate the number of existing states n(E) whose energy is less than E.
For this we write the energy as follows,
~2 2
Enx ,ny ,nz = k , (9.35)
2me nx ,ny ,nz
n π
where knx ,ny ,nz = kx2 + ky2 + kz2 = ( nLx π )2 + ( Ly )2 + ( nLz π )2 . Each set of values k = (kx , ky , kz )
corresponds to a possible state which is accessible to the system, and each state is associated
with a volume element (π/L)3 in k-space. Note that the p number of states n(E) with energy less
given by the number of states in which k ≤ 2me E/~2 . This corresponds to a sphere
than E is p
of radius 2me E/~2 centered on the origin, thus
Fig. 9.2 illustrates the possible states in k-space and the sphere with energy E.
With the above expression we are able to calculate the Fermi energy, since the condition
n(EF ) = N implies,
2/3
~2 2N
EF = 3π 3 . (9.37)
2me L
It is interesting to note that for T = 0 K all states below EF are occupied and all those above
are unoccupied. In addition, we define the density of states ρ(E), where ρ(E)dE is the number
of states with energies between E and E + dE. Thus, the density of states becomes,
3/2
dn(E) L3 2me 3N E 1/2
ρ(E) = = 2 E 1/2 = . (9.38)
dE 2π ~2 2 E 3/2
F
8
Note, however, that for large N (e.g., 1 mol of electrons) this process turns out to be unfeasible.
220 CHAPTER 9. ATOMS WITH MANY ELECTRONS
Therefore, the total kinetic energy of the electrons in the electronic shell is,
Z
T [ρ] = C ρ5/3 d3 dr . (9.42)
The potentials associated with the interactions e− -e− and e− -p+ can also be written as func-
tionals of electronic density, such that,
Z Z
Ze2 ρ(r) 3 1 e2 ρ(r)ρ(r0 ) 3 3 0
Vep [ρ] = − d r and Vee [ρ] = d rd r . (9.43)
4π0 r 2 4π0 |r − r0 |
Thus, the total energy (Thomas-Fermi energy) can be written as a functional of the electronic
density of the atom,
HT H [ρ] = T [ρ] + Vncl−ele [ρ] + Vele−ele [ρ] . (9.44)
δHT H [ρ]
δHT H [ρ] − µδρ(r) = 0 =⇒ =µ.
δρ(r)
The above expression describes the electron density of the atom in its ground state. It is
interesting to note that, as δHT H [ρ]/δρ(r) = µ, we can identify the Lagrange multiplier µ as a
chemical potential. In particular, for non-interacting neutral atoms, we have µ = 0, so,
3/2
3
ρ(r) = ξef (r)3/2 . (9.48)
5C
In addition, since for an atom both the potential and the electronic density must have spherical
symmetry, we can write,
Zχ(r)
ξef (r) = . (9.49)
r
Now, it is common to perform the variable transformation x = αr (α = 27/3 Z 1/3 /(3π)2/3 =
1.1295Z 1/3 ) and use atomic units, where ~2 /me = 1 such that C = 35/3 π 4/3 /10 = 2.871 [11].
Thus, the ’electronic density’ is,
3/2
Z 2 25 χ(x)
ρ(x) = 2 3 . (9.50)
3 π x
In addition, the potential must satisfy the Poisson equation ∇2 ξef = 4πρ, resulting in the
Thomas-Fermi equation:
d2 χ χ3/2
= . (9.51)
dx2 x1/2
It is important to note that the previous equation does not depend on the parameter Z,
thus being a general result for any neutral atom. The function χ(x) is determined numerically,
but we can analyze its asymptotic values given the expected behavior of the effective potential
ξef (r): for r → 0 we expect that ξef (r) = Zr , hence χ(0) = 1; on the other hand for r → ∞ we
expect ξef (r) = 0, hence χ(∞) = 0.
With χ(x) known we get the charge density ρ(x), and hence we are able to calculate the
total energy of the atom under investigation. Thus, it is possible to show that [11],
It is important to highlight some points: (i) the result holds for neutral atoms; (ii) there is no
electronic shell structure assumed; (iii) quantum statistical effects of identical particles are not
taken into account. A more refined model which deals with the criticism (iii) and, in addition,
is closer to density functional theory (DFT) is the Thomas-Fermi-Dirac model.
approach afterward. Knowing the effect of the shielding of the nucleus by electronic charges, we
already know the asymptotes (see Fig. 9.7),
Ze2 e2
V0 = − para r→0 and V0 = − for r → ∞. (9.53)
4π0 r 4π0 r
An effective potential, V0 , constructed to satisfy these limits serves as a first trial to establish
and numerically solve the Schrödinger equation for each electron independently,
~2 2
Ĥi = − ∇ + V0 ψi (ri ) = ei ψi (ri ) . (9.54)
2m i
With this we calculate all energies and eigenunctions (only the radial parts are of interest)
minimizing the total energy and respecting the weak Pauli principle, that is, all states are
successively filled with electrons. For the total wavefunction we obtain,
N
! N
X X
Ĥi ΨN = En ΨN com ΨN = ψ1 · .. · ψN and En = ei . (9.55)
i=1 i=1
With the eigenfunctions we calculate the charge densities e|ψj (rj )|2 .
We integrate the field to
obtain a potential that represents an improved estimation for the electronic mean field,
Ze2 XZ e2
Vi ←− − + d3 rj |ψj (rj )|2 . (9.56)
4π0 ri 4π0 |ri − rj |
j6=i
We replace that potential in the Schrödinger equation, and repeat the whole process from the be-
ginning. This self-consistent method is called Hartree method. Fock improved these calculations
using antisymmetric wavefunctions for the valence electrons. This method is called Hartree-Fock
method. The idea of the Hartree method is shown in the following diagram,
choose a common potential V (r) for
all electrons, f.ex. Thomas-Fermi
↓
(0)
calculate
the eigenfunctions
ψi with
−~2 (0) (0) (0)
2m ∆i + V0 (ri ) ψi = ei ψi
↓
new entry
←− ψi ←−
↓ ↓ ↑
↓ classify according to increasing energies ei ↑
↓ satisfying Pauli’s principle ↑
↓ ↓ ↑
↓ calculate the mean potential Vi acting ↑
↓ on electron i for all i ↑
P R
↓ Vi (ri ) = j6=i 4πεZe 0 rij
|ψi (ri )|2 d3 rj ↑
↓ ↓ ↑
↓ calculate
new ψ i with these
Vi ↑
−~2 Ze
↓ 2m ∆i − 4πε0 ri + Vi (ri ) ψi = ei ψi ↑
↓ ↓ ↑
bad
−→ compare with initial ψi −→
↓ good
result
O Método de Hartree-Fock é um método utilizado para tratar sistemas atômicos/moleculares de
muitos corpos que visa a obtenção da função da onda eletrônica do sistema. Ele é um refinamento
do Método
9.3. ELECTRONIC (SCF:
SHELL
de Hartree Self-Consistent-Field ), pois leva em conta a antissimetrização da223
STRUCTURE
função de onda. De maneira geral o método se baseia no princípio variacional e na suposição
9.3.3 Hartree Fock method
de que podemos escrever a função de onda global como um determinante de Slater, sendo que
The Hartree-Fock method used to treat atomic or molecular many-body systems aims at obtain-
ingcada
the elétron ocupa um estado orbital específico (spin-orbital) e interage com um potencial efetivo
electronic wavefunction of the system. It is a refinement of the Hartree method (SCF:
oriundo dos elétrons que
Self-Consistent-Field), ocupam
since outros
it takes orbitais.
into accountDesta
themaneira, ao invés de resolvermos
antisymmetrization a equação
of the wavefunction.
In general, the method is based on the variational principle and on the assumption that we
de Schrödinger devemos resolver um sistema de equações denominada de Equações de Hartree-
can write the global wavefunction as a determinant of Slater, with each electron occupying a
specific do tipo state
Fock orbital (spin-orbital)
F̂ ψk (1) O método
= εk ψk (1). and é realizado
interacting de maneira
with an effectivecíclica até convergência
potential stemming from dos the
electrons which occupy other orbitals. In this way, instead of solving the Schrödinger equation
orbitais atômicos e suas respectivas energias, por este motivo é denominado como auto-consistente:
we must solve a set of equations called Hartree-Fock equations of the type F̂ ψk (1) = k ψk (1).
Thea partir
methodda suposição inicial
is performed da funçãountil
cyclically de onda global é calculado
convergence o potencial
of the atomic efetivo
orbitals andemtheir
cada respective
orbital
energies is reached in a procedure called self-consistent: Starting from an initial trial global
e é obtido um novo conjunto de funções de onda que, por sua vez, geram um novo potencial efetivo;
wavefunction we calculate the effective potential in each orbital and a new set of wavefunctions
assim,ineste
which, novo
turn, potencial
generate é utilizado
a new em um
effective novo sistema
potential. This de
newequações de Hartree-Fock.
potential is then used inA Figura
a new set
of 2.1
Hartree-Fock equations. Fig. 9.3 illustrates
ilustra a característica cíclica do método.
the iterative procedure.
Definição dos
spin-orbitais
Não
Sim
Fim
2.1: Método
Fig.Figure de Hartree-Fock.
9.3: Hartree method.
9.3.3.1 Hartree-Fock
2.2 Equações equations
de Hartree-Fock -
We have seen that the Hamiltonian of a multielectronic atom is given by,
Vimos que o Hamiltoniano de um átomo multieletrônico é dado por
XZ 2 X X Z X
pi Ze2 1 e2 1
Ĥ = − + = ĥ i + V̂ij , (9.57)
Z 0ri 2 2
24π0 |ri X
2
i=1
2mi 4π
X pi Ze
i6= j 1 − rj | 2
ei=1 i6=j
Ĥ = − +
i=1
2mi 4π0 ri 2 4π0 |rj − ri |
i6=j
where ĥi is the Hamiltonian onlyZ of the electron i, and V̂ij is the interaction term between the
X 1 Xwe must suppose that the multielectronic state can
electrons i and j. To implement
= theĥmethod
i+ V̂ji
be written as the product of thei=1 2 states of each electron:
individuali6=j
Now, we use the variational principle to minimize the expectation value of the ground state
energy by varying the functions ψk (n). In this way, the correct orbitals are those that minimize
the energy. The expectation value is written as,
Z
X X
E = hΨ|Ĥ|Ψi = hΨ| ĥi |Ψi + hΨ| 21 V̂ij |Ψi . (9.60)
i=1 i6=j
Hence,
Z
X Z
X
1
E= ĥi |Ψi + 2 [hψi ψj |V̂ij |ψi ψj i − hψj ψi |V̂ij |ψj ψi i] . (9.62)
i=1 i,j
The above expression can be minimized via Lagrange multipliers under the constraint that the
states are orthogonal hψi |ψj i = δij ,
X
δ hΨ|Ĥ|Ψi − ij [hψi ψj |V̂ij |ψi ψj i − hψj ψi |V̂ij |ψj ψi i] (9.63)
i,j
The operator K̂i , denominated exchange operator, is a consequence of the symmetrization process
and therefore a purely quantum effect, that is, without classical analogue:
Z
∗
K̂i ψk (1) = ψi (2)V12 ψk (2)dr2 ψk (1) . (9.66)
Once we known all the wavefunctions, the energies of the orbitals can be obtained in the following
way: ( )
Z X Z
∗
dr1 ψk (1) ĥ1 + ˆ
(2Ji − K̂i ) ψk (1) = k dr1 ψk∗ (1)ψk (1) = k ,
∗
(9.67)
i
9.3. ELECTRONIC SHELL STRUCTURE 225
that is,
Z X
k = dr1 ψk∗ (1)ĥ1 ψk∗ (1) + (2Jˆki − K̂ki ) , (9.68)
i
where,
Z
Jˆki = dr1 ψk∗ (1)Jˆi ψk (1) is the Coulomb integral (9.69)
Z
K̂ki = dr1 ψk∗ (1)K̂i ψk (1) is the exchange integral .
Furthermore, if we assume that, taking an electron away from the orbital ψk the electronic
distribution remains unchanged, it is possible to associate the energy k with the ionization
energy of the electron in this orbital. This equality is known as Koopman’s theorem,
Ik ' k . (9.71)
Finally, it is worth emphasizing that a refinement of the self-consistent procedure can be ob-
tained, if we expand each spin-orbital in a certain basis of functions (not necessarily orthogonal),
such that,
X
ψk (n) = Clk φl (n) , (9.72)
l
where N is the number of functions in the basis. Then, the Hartree-Fock equations can be
written as,
N
X N
X
F̂ ψk (1) = k ψk (1) =⇒ F̂ Clk φl (1) = k Clk φl (1) (9.73)
l l
N
X Z N
X Z
=⇒ Clk dr1 φ∗m (1)F̂ φl (1) = k Clk dr1 φ∗m (1)φl (1)
m l
N
X N
X
=⇒ Fml Clk = k Sml Clk .
m l
It is important to distinguish three different energetic sequences: 1. Tab. 9.6 shows, for a
given atom, the excited orbitals of the last electron. 2. The energy sequence shown in Tab. 9.4
tells us in which orbital the next electron will be placed, when we go to the next atom in the
periodic table 9.10. 3. The inner electrons are subject to different potentials and, hence, follow
a different sequence energetic sequence: While for the inner electrons we find,
the sequence is partially inverted for the outermost electron. Note that it is the outermost
electrons that determine the chemical reactivity of the atom. The sequence is illustrated in
Fig. 9.8.
Following Hund’s rule, the L · S-coupling is energetically favorable compared to the j · j-
coupling, which means that the spins of the outermost electrons, that is, the electrons outside
of filled subshells (n, `), prefer to orient their spins in parallel in order to anti-symmetrize the
spatial wavefunctions and thus maximize the distance between the electrons. Every sub-layer of
9.4. THE PERIODIC SYSTEM OF ELEMENTS 227
the series shown in Fig. 9.8 must be filled in the listed order before placing new electrons in the
next layer.
Noble gases have small radii, high excitation energies and high ionization energies. The
valence electron must fill the gap to higher major quantum numbers. Halogens have strong
electro-affinities, since the outer electron layer (nmax ) is incomplete and therefore malleable, such
that an electron approaching the halogen perceives the nucleus through a partially transparent
shield. Alkalis are similar to hydrogen and have excitation energies in the optical regime. Their
fundamental state 2 S1/2 is determined by a valence electron only in the ` = 0 orbital. Unlike
hydrogen, excitation energies are highly dependent on `, since orbits with small ` have higher
probabilities to be in the unshielded region −Z 2 e2 /r than orbits with large `, who spend more
time in the shielded region −e2 /r. For the same reason, energies corresponding to larger n
resemble more those of the hydrogen spectrum.
The interior shell structure of the atoms can be analyzed by X-ray scattering. Electrons
decelerated by atoms emit a continuous spectrum called Bremsstrahlung, but they can also
expel electrons from the inner layers leaving a hole behind. When a hole is filled by cascades of
electrons falling down from higher layers, the atom emits a specific X-ray spectrum (≈ 104 eV).
The selection rules ∆` = ±1 and ∆j = ±1 split the lines in two components. X-ray spectra
of neighboring elements in the periodic system of elements are very similar, because the inner
layers not being shielded, they see a potential close to ∝ Z 2 /r. Therefore, the Z-dependency of
the rows in the periodic table is more or less ω ∝ Z 2 , as predicted by Bohr’s atomic model.
consisting of kinetic energy, potential energy of the nucleus, and interaction between electrons
Ĥww . The index cites the electrons and runs from 1 to the nuclear charge number Z. We neglect
the electron-electron interaction Ĥww . Then, each electron perceives a hydrogen atom with the
solution α replaced by Zα, because the Coulomb potential of the nucleus is now,
Ze2 1 Zα~c
= . (9.77)
4π0 r r
For the energies we obtain then:
Z
X 2
me 2 Zα
E=− c . (9.78)
2 ni
i
Each of the energy levels in this sum contains several states. Each of these states can be
occupied by a single electron, according to the Pauli principle. In this way, we obtain the
electronic configuration for the atoms of the periodic system. In this picture, the energies of the
ground states of the elements, normalized by 1/Z 2 , can be arranged in the scheme 9.5:
This only works for atoms with up to 18 electrons. When the layer 3p is completely filled,
the next to be occupied is not 3d but the 4s. The new scheme is illustrated in 9.6.
The anomalies beginning at Z = 18 arise due to electron-electron interaction. To understand
this, we consider an electron in the mean field generated by all other electrons. The electron
228 CHAPTER 9. ATOMS WITH MANY ELECTRONS
N 4s 19
K 20
Ca 4p 4d 4f
M 3s 11
Na 12Mg 3p 13
Al 14
Si 15
P 16
S 17
Cl 18
Ar 3d
L 2s 3
Li 4
Be 2p 5
B 6
C 7
N 8
O 9
F 10
Ne
K 1s 1
H 2
He
is then located in an effective potential which evolves between two Coulomb potentials, one for
large and the other for small distances: for small distances, the electron sees only the positive
point charge of the nucleus and the potential is proportional to Zα,
Ze2 1
V (r) = − . (9.79)
4π0 r
For large distances, the nucleus and all other electrons together form a nearly spherical charged
source called the ’core’, and we can approximate the potential by,
e2 1
V (r) = − . (9.80)
4π0 r
The real potential evolves from one to the other coulombian potential, as distance from the
nucleus is increased:
Near the nucleus, the electrons shield the positive charge less than for large r. Thus, those
states that have a high probability near the nucleus are energetically lowered. That is,
E2s < E2p and E6s < E6p < E6d . (9.81)
The degeneracy of the orbital angular momentum in the Schrödinger model is thus lifted. The
shielding is, as can be seen in the example of the excited states of lithium, a large effect in the
range of some eV.
The shielding also accounts for the anomalies in the periodic system, such as in K or Ca.
Since E4s < E3d , the 4s state is filled before the 3d. Similar anomalies also occur in Rb (5s), Cs
(6s), and Fr (7s). In rare earths the shielding effect is even more pronounced. Here, the energy
of the state 6s is even below the energy of the 4f , which means that the shells 6s, 5s, 5p, and
5d protect the 4f shell very well 9 .
9.4.2 Alkalines
The electronic shell structure of alkalines consists of a completed noble gas shell and an additional
valence electron. Their spectrum is therefore very similar to hydrogen. An empirical approach
can be stressed to describe this feature,
c2
En,l = − 12 µEG c2 , (9.82)
n − ∆(n, l)
9
An example of this is Nd:YAG (Neodymium in Yttrium Aluminum Garnet). In this crystal, optical transitions
can be excited within the 4f shell of the Nd. However, these transitions are only allowed due to perturbations of
the crystalline field. The very strong shielding ensures a long life of the excited state. For this reason this crystal
is an excellent laser material.
9.4. THE PERIODIC SYSTEM OF ELEMENTS 229
Li Na K Rb Cs H
S P D F S P D F S P D F S P D F S P D F
0
7
6
5
5 5 5 5 5
5 5 6 5 7 6 8 7
4 4 4 4 4 4 4
-1 4 6 4 7 4
4 5 5 8
6
5 6 7
3
3 3 4 3
3 6 7
5 4
-2
3 4
5
5 6
4
-3
3
2
2
-4 6
5
4
-5
3
2
Figure 9.6: Comparison of the excitation energies of the valence electron for several alkaline
atoms.
where µEG is the reduced mass relative to the noble gas hull and ∆(n, l) is called quantum defect.
The quantum defect is tabulated for most alkaline states and is particularly important for low
energy states. For sodium, for example, the values are:
For states with a large angular momentum, the quantum defect disappears. In these states,
the electron is far from the nucleus and the potential is similar to that of hydrogen. Alkalines
are currently widely studied in quantum optics laboratories, for being comparatively simple,
but having a sufficiently rich structure to be interesting. The fundamental electronic transitions
typically lie in the visible and near-infrared spectral range and can be excited with compara-
tively simple laser sources. The lifetime of excited states is typically longer than 20 ns, which
corresponds natural linewidths of approximately (2π) 10 MHz.
230 CHAPTER 9. ATOMS WITH MANY ELECTRONS
−2
V(r) / (e2/aB)
−4
−6
−8
−10
0 1 2 3
r/aB
Q 7s 7p
P 6s 6p 6d
O 5s 5p 5d 5f actinides
M 3s 3p 3d
L 2s 2p
elementos de transição
K 1s
l= 0 1 2 3
Figure 9.8: Illustration of the sequence of filling the orbitals with electrons.
• li · lj -coupling
• si · sj -coupling
• relativistic corrections
9.4. THE PERIODIC SYSTEM OF ELEMENTS 231
1
P
1
4
S=0 D (3/2,5/2) 1,2,3,4 3
1 2
F
1
3
(3/2,3/2) 0,1,2,3 2
3
l =1, l =2
1 2 P 2 1
0
3
P 3
P 1
energia de 3
3
D P 0
1partícula
S=0
simetria e
(1/2,5/2) 2,3 3
energia de troca 3 2
F
interação residual
de Coulomb estrutura fina
(1/2,3/2) 1,2 2
1
(j , j )
1 2 J
• hyperfine structure
The above coupling scheme is called Russel-Saunders coupling or LS-coupling. It works when
the spin-orbit coupling is small. In this case, intercombination is forbidden, which means that
there can be no electromagnetic transition between states with different spins (see metastable
helium).
Since ELS ' Zα4 ' Z 4 , for heavy atoms, the coupling of an electronic spin to its own orbital
momentum grows strongly with Z, as well as the symmetrization and the exchange energy, which
mutually orient the spins, and the residual Coulomb interaction, which mutually couples the
angular orbital momenta. In this case, the orientation of Li relative to Si delivers more energy
than the exchange energy and the residual energy cost. Hence, the spin and the orbital angular
momentum of an individual electron couple first,
ji = li + si . (9.83)
HF S ∝ ji · jj . (9.84)
Pure jj-coupling only exists for very heavy nuclei. Normally, we have a so-called inter-
mediate coupling, which is a mixture of LS and jj-coupling. This can considerable relax the
intercombination prohibition. When the coupling is pure, we have the following dipolar selection
rules:
LS-coupling: ∆S = 0, ∆L = ±1, ∆` = ±1
jj-coupling: ∆j = 0, ±1 for one e− , ∆j = 0 for all others
In addition we have for the two couplings: ∆J = 0, ±1, but J, J 0 = 0 is forbidden, ∆mJ =
0, ±1 when ∆J = 0 but mJ , mJ 0 = 0 is forbidden.
Of course, with the presence of other atoms, other interactions may generate other relevant
contributions to the Hamiltonian.
The following interactions contribute to the potential V : The Coulomb interactions,
Z
X Z
X
Ze2 e2
Vncl−ele = − and Vele−ele = , (9.86)
4π0 |r − ri | 4π0 |ri − rj |
i=1 i<j=1
Vsym , (9.87)
interactions between the spin of the electrons and the nuclear spin and between the orbital
angular momentum of the electrons and the nuclear spin,
A
Vhf s = J·I , (9.91)
~2
and relativistic corrections.
Vrel , (9.92)
In addition, static external fields may displace energy levels and can influence the internal
coupling of angular momenta and spins,
Vext = −d · E , − µ
~ ·B . (9.93)
What quantum numbers are good depends on the relative amplitudes of intra-atomic inter-
actions:
r
Case 1: fine structure with L·S-coupling plus Zeeman splitting of hyperfine structure: Vncl−ele , Vele−ele
a
Vele−ele , Vsym Vls Vhf s VB the quantum number are |ni , li , L, S, J, F, mF i.
r
Case 2: fine structure with j·j-coupling plus Zeeman splitting of hyperfine structure: Vncl−ele , Vele−ele
a
Vls Vele−ele , Vsym Vhf s VB the quantum number are |ni , li , ji , J, F, mF i.
9.5. EXERCISES 233
r
Case 3: fine structure with L·S-coupling plus hyperfine structure of Zeeman splitting: Vncl−ele , Vele−ele
a
Vele−ele , Vsym Vls VB Vhf s the quantum number are |ni , li , L, S, J,mJ , mI i.
r
Case 4: fine structure with L·S-coupling plus Paschen-Back splitting of fine structure: Vncl−ele , Vele−ele
a
Vele−ele , Vsym VB Vls Vhf s the quantum number are |ni , li , L, S,mL , mS , mI i.
Vncl−ele splitting in n
coarse structure ↓
Vele−ele splitting in l
↓
Vsym splitting in S &
↓ Vl s splitting in ji
fine structure Vee splitting in L ↓
↓ Vsym , Vele−ele splitting in J
VLS splitting in J .
↓
hyperfine structure Vhf s splitting in F
↓
Zeeman effect VLS splitting in mF
9.5 Exercises
9.5.1 Symmetrization of bosons and fermions
9.5.1.1 Ex: Bosonic and fermionic isotopes
Consulting an isotope table determine the bosonic or fermionic character of the following atomic
species: 87 Sr, 86 Sr, 87 Rb, 39 K e 40 K.
9.5.2 Helium
9.5.2.1 Ex: Helium atom
Compare the measured binding energy with the prediction of Bohr’s model considering the
inter-electronic interaction up to first order TIPT.
Li Be !31 B C N O F Ne
2 electron mass me 9.1094 H 10 kg
Lithium Beryllium 2 Boron Carbon Nitrogen Oxygen Fluorine Neon
mec 0.5110 MeV
6.941 9.01218 !27 10.811 12.0107 14.00674 15.9994 18.99840 20.1797
2 2 2 proton mass mp 1.6726 H 10 kg 2 2 2 2 2 2 2 3 2 2 4 2 2 5 2 2 6
1s 2s 1s 2s 1s 2s 2p 1s 2s 2p 1s 2s 2p 1s 2s 2p 1s 2s 2p 1s 2s 2p
5.3917 9.3227 fine-structure constant a 1/137.036 8.2980 11.2603 14.5341 13.6181 17.4228 21.5646
!1
2 1 Rydberg constant R4 10 973 732 m 2 3 4 3 2 1
11 S1/2 12 S0 15 13 P°1/2 14 P0 15 S°3/2 16 P2 17 P°3/2 18 S0
R4c 3.289 84 H 10 Hz
Na Mg R4hc 13.6057 eV Al Si P S Cl Ar
!23 !1
3 Sodium Magnesium Boltzmann constant k 1.3807 H 10 JK Aluminum Silicon Phosphorus Sulfur Chlorine Argon
22.98977 24.3050 26.98154 28.0855 30.97376 32.066 35.4527 39.948
2 2 2 2 2 3 2 4 2 5 2 6
[Ne]3s [Ne]3s VIIIA [Ne]3s 3p [Ne]3s 3p [Ne]3s 3p [Ne]3s 3p [Ne]3s 3p [Ne]3s 3p
5.1391 7.6462 IIIA IVA VA VIA VIIA IB IIB 5.9858 8.1517 10.4867 10.3600 12.9676 15.7596
2 1 5 4 3 2 2 3 4 3 2 1
19 S1/2 20 S0 21 2D3/2 22 3F2 23 4F3/2 24 7S3 25 6S5/2 26 D4 27 F9/2 28 F4 29 S1/2 30 1S0 31 P°1/2 32 P0 33 S°3/2 34 P2 35 P°3/2 36 S0
4
K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr
Potassium Calcium Scandium Titanium Vanadium Chromium Manganese Iron Cobalt Nickel Copper Zinc Gallium Germanium Arsenic Selenium Bromine Krypton
39.0983 40.078 44.95591 47.867 50.9415 51.9961 54.93805 55.845 58.93320 58.6934 63.546 65.39 69.723 72.61 74.92160 78.96 79.904 83.80
Period
2 2 2 2 3 2 5 5 2 6 2 7 2 8 2 10 10 2 10 2 10 2 2 10 2 3 10 2 4 10 2 5 10 2 6
[Ar]4s [Ar]4s [Ar]3d4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s [Ar]3d 4s 4p [Ar]3d 4s 4p [Ar]3d 4s 4p [Ar]3d 4s 4p [Ar]3d 4s 4p [Ar]3d 4s 4p
4.3407 6.1132 6.5615 6.8281 6.7462 6.7665 7.4340 7.9024 7.8810 7.6398 7.7264 9.3942 5.9993 7.8994 9.7886 9.7524 11.8138 13.9996
2 1 2 3 6 7 6 5 4 1 2 1 2 3 4 3 2 1
37 S1/2 38 S0 39 D3/2 40 F2 41 D1/2 42 S3 43 S5/2 44 F5 45 F9/2 46 S0 47 S1/2 48 S0 49 P°1/2 50 P0 51 S°3/2 52 P2 53 P°3/2 54 S0
5
Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I Xe
Rubidium Strontium Yttrium Zirconium Niobium Molybdenum Technetium Ruthenium Rhodium Palladium Silver Cadmium Indium Tin Antimony Tellurium Iodine Xenon
85.4678 87.62 88.90585 91.224 92.90638 95.94 (98) 101.07 102.90550 106.42 107.8682 112.411 114.818 118.710 121.760 127.60 126.90447 131.29
2 2 2 2 4 5 5 2 7 8 10 10 10 2 10 2 10 2 2 10 2 3 10 2 4 10 2 5 10 2 6
[Kr]5s [Kr]5s [Kr]4d5s [Kr]4d 5s [Kr]4d 5s [Kr]4d 5s [Kr]4d 5s [Kr]4d 5s [Kr]4d 5s [Kr]4d [Kr]4d 5s [Kr]4d 5s [Kr]4d 5s 5p [Kr]4d 5s 5p [Kr]4d 5s 5p [Kr]4d 5s 5p [Kr]4d 5s 5p [Kr]4d 5s 5p
4.1771 5.6949 6.2171 6.6339 6.7589 7.0924 7.28 7.3605 7.4589 8.3369 7.5762 8.9938 5.7864 7.3439 8.6084 9.0096 10.4513 12.1298
2 1 3 4 5 6 5 4 3 2 1 2 3 4 3 2 1
55 S1/2 56 S0 72 F2 73 F3/2 74 D0 75 S5/2 76 D4 77 F9/2 78 D3 79 S1/2 80 S0 81 P°1/2 82 P0 83 S°3/2 84 P2 85 P°3/2 86 S0
6
Cs Ba Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Po At Rn
Cesium Barium Hafnium Tantalum Tungsten Rhenium Osmium Iridium Platinum Gold Mercury Thallium Lead Bismuth Polonium Astatine Radon
132.90545 137.327 178.49 180.9479 183.84 186.207 190.23 192.217 195.078 196.96655 200.59 204.3833 207.2 208.98038 (209) (210) (222)
Atomic Ground-state 2 1 4- 5- 6 7 8 9 6 5- 4- 3 2 1 2
Number Level
57 D3/2 58 G°4 59 -I °9/2 60 -I 4 61 H°5/2 62 F0 63 S°7/2 64 D°2 65 H°15/2 66 -I 8 67 -I °15/2 68 H6 69 F°7/2 70 S0 71 D3/2
1
La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Symbol 58 G°4 Lanthanum Cerium Praseodymium Neodymium Promethium Samarium Europium Gadolinium Terbium Dysprosium Holmium Erbium Thulium Ytterbium Lutetium
138.9055 140.116 140.90765 144.24 (145) 150.36 151.964 157.25 158.92534 162.50 164.93032 167.26 168.93421 173.04 174.967
2 2 3 2 4 2 5 2 6 2 7 2 7 2 9 2 10 2 11 2 12 2 13 2 14 2 14 2
Ce [Xe]5d6s [Xe]4f5d6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 5d6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 6s [Xe]4f 5d6s
Name 5.5769 5.5387 5.473 5.5250 5.582 5.6436 5.6704 6.1501 5.8638 5.9389 6.0215 6.1077 6.1843 6.2542 5.4259
Cerium 2 3 4 5 6 7 8 9 6 5- 4- 3 2 1 2
140.116 89 D3/2 90 F2 91 K11/2 92 L°6 93 L11/2 94 F0 95 S°7/2 96 D°2 97 H°15/2 98 -I 8 99 -I °15/2 100 H6 101 F°7/2 102 S0 103 P°1/2?
Atomic
†
Weight [Xe]4f5d6s 2 Ac Th Pa U Np Pu Am Cm Bk Cf Es Fm Md No Lr
5.5387 Actinium Thorium Protactinium Uranium Neptunium Plutonium Americium Curium Berkelium Californium Einsteinium Fermium Mendelevium Nobelium Lawrencium
(227) 232.0381 231.03588 238.0289 (237) (244) (243) (247) (247) (251) (252) (257) (258) (259) (262)
Ground-state Ionization 2 2 2 2 2 3 2 4 2 6 2 7 2 7 2 9 2 10 2 11 2 12 2 13 2 14 2 14 2
[Rn]6d7s [Rn]6d 7s [Rn]5f 6d7s [Rn]5f 6d7s [Rn]5f 6d7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 6d7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 7s [Rn]5f 7s 7p?
Configuration Energy (eV) 5.17 6.3067 5.89 6.1941 6.2657 6.0262 5.9738 5.9915 6.1979 6.2817 6.42 6.50 6.58 6.65 4.9 ?
† 12
235
Based upon C. ( ) indicates the mass number of the most stable isotope. For a description and the most accurate values and uncertainties, see J. Phys. Chem. Ref. Data, 26 (5), 1239 (1997). March 1999
236 CHAPTER 9. ATOMS WITH MANY ELECTRONS
Part III
Quantum Optics
237
Chapter 10
239
240 CHAPTER 10. THE BLOCH EQUATIONS
matrix elements of this density operator are the optical Bloch equations, and we must use them
instead of the Schrödinger equation. In order to appreciate the origin and the physical content
of the optical Bloch equations we begin by reviewing the rudiments of the density matrix theory.
where |ψk i is a complete set of orthonormal states of the system under study. We have a
statistical distribution
P of these states with pk being the probability of finding |ψk i in the set.
Obviously, k pk = 1. That is, the density operator acts on a member of the set |ψk i such as
to extract the probability of finding the system in |ψk i,
X
ρ̂|ψm i = pk |ψk ihψk |ψm i = pm |ψm i . (10.2)
k
If all members of the set are in the same state, for example |ψk i, the density operator reduces
to,
ρ̂ = |ψk ihψk | , (10.3)
and the system is in a pure state with P̂k = 1. Each time a quantum state can be expressed
by a single wave function, it is a pure state, but it does not have to be an eigenstate. Starting
from the equation (10.2) we find the diagonal elements of the density matrix, which are the
probabilities of finding the system in |ψm i,
Assuming that all |ψk i are orthonormal, the non-diagonal elements are necessarily zero. Besides
that, X
hψk |ρ̂|ψk i = 1 . (10.5)
k
So, ρ̂ contains all information about the system. When a state is unknown, ρ̂ describes the
probability of finding the system in each state. When the state is fully known, ρ̂ describes
a pure state, that is, a vector in the Hilbert space, which is unequivocally determined by a
complete set of observables with their respective quantum numbers.
1
In the presence of degeneracy or a continuous spectrum we can generalize the definition:
X Z X Z
ρ̂ ≡ pk P̂k + pλ P̂λ dλ where P̂k ≡ |kmihkm| and P̂λ ≡ |λµihλµ|dµ .
k m
Here, m and µ are degenerate quantum numbers, m, n are discrete, and λ, µ are continuous quantum numbers.
The set of quantum numbers is complete, when
X Z
|kmihkm| = 1̂ = |λµihλµ|dλdµ .
k,m
P
The degree of degeneracy of a state |ki is Tr P̂k = m 1. The probability of finding the system in the state |ki
P
is hP̂k i = pn m 1.
10.1. DENSITY MATRIX 241
ρ̂ = ρ̂†
hρ̂i ≥ 0
Tr ρ̂ = 1
. (10.6)
Tr ρ̂2 ≤ 1
det ρ̂ = 0
ρ̂ = ρ̂2 for a pure state
10.1.1.1 Entropy
In a very general sense, the entropy determines in what direction a reversible process will take
place. It is related to the size of the available phase space on both sides of the reaction. For
example, the coupling of discrete and continuous modes is governed by entropy considerations.
Entropy measures of the lack of information about a system from which we only know hĤi,
The information entropy (or von Neumann entropy) of statistically independent systems ρ̂ ≡
ρ̂1 ⊗ ρ̂2 is additive S = S1 + S2 . We can also define absolute temperatures by T −1 ≡ ∂S/∂hĤi.
The entropy of a pure state is 0. Hamiltonian processes conserve entropy, for they correspond to
non-dissipative unitary transformations. On the other side, relaxation increases the entropy and
the phase space volume. Another common definition is the so-called purity or Renyi entropy,
Quantum states can exhibit coherences. For example, if we express a state |ψi on a basis of
eigenstates |1i and |2i:
|hψ|1i|2 h1|ψihψ|2i
ρ̂ = |ψihψ| = . (10.9)
h2|ψihψ|1i |hψ|2i|2
The measurement process is not described by this equation. A pure state will always remain
pure. If the eigenstates do not interact, the density operator will remain diagonal. The von
Neumann equation conserves the properties of hermiticity, ρ̂ = ρ̂† , completeness, Tr ρ̂ = 1, and
purity det ρ̂ = 0.
The density operator for a statistical mixture in a canonical ensemble (where S is maximum,
U is variable, and N is fixed) follows from a variational problem with the Lagrange parameters
δ(S − kB αh1̂i − kB βhĤi) = 0, since Tr ρ̂ and hĤi are fixed by boundary conditions. We find,
1 −Ĥ/kB T
ρ̂ = Ze with Z ≡ Tr e−Ĥ/kB T . (10.11)
We also have the expectation values, hHi = −∂ ln Z/∂(1/kB T ) and (∆H)2 = −∂hĤi/∂(1/kB T ).
All quantities are fixed, except the kinetic energy, which balances the interaction with a heat
242 CHAPTER 10. THE BLOCH EQUATIONS
bath. T is the only equilibrium parameter. The density operator satisfies a Boltzmann distri-
bution 2 ,
Z
p2 ∂ 2 kB
U = hĤi = =− ln e−p /2mkB T dp = T . (10.12)
2m ∂(1/kB T ) 2
P
using the completeness relation (2.68), that is, n |nihn| = I, and defining,
as the projection of the state vector |ψk i on the basis vector |ni. Now, we can write the density
operator matrix representation within the basis {|ni} using the definition of ρ̂ in Eq. (10.1) and
replacing the expansions of |ψk i and hψk | of Eq. (10.13):
X X X X X
ρ̂ = pk |ψk ihψk | = pk |nihn|ψk ihψk |mihm| = pk cnk c∗mk |nihm| . (10.15)
k k m,n k m,n
P
with the diagonal elements hn|ρ̂|ni = k pk |cnk |2 e,
X X
ρ∗nm = hn|ρ̂|mi∗ = pk c∗nk cmk = pk hm|ψk ihψk |ni = hm|ρ|ni = ρmn , (10.17)
k k
Example 38 (Density operator for a single atom): For a very simple system such as
a single atom with several levels, that without spontaneous emission can be described by a
single wavefunction |ψ1 i, we can let pk = δ1k . That is, the equations (10.15) and (10.17)
reduce to,
X
ρ̂ = cn1 c∗m1 |nihm| and hn|ρ|mi = cn1 c∗m1 . (10.18)
m,n
2
The von Neumann entropy S of a mixture can be expressed in terms of the eigenvalues or in terms of the
trace and theP
logarithm of the density operator ρ̂. Since ρ̂ is a semi-definite positivePoperator, its spectrum λi ,
given by ρ = iP λi |ϕi ihϕi | where {|ϕi i} is an orthonormal basis, satisfies λi > 0 and λi = 1. Then the entropy
becomes S = − i λi ln λi = −Tr (ρ ln ρ).
10.1. DENSITY MATRIX 243
With the definition of the density operator (10.1) we can write the Eq. (10.19) as,
X
Tr ρ̂ ≡ pk hn|ψk ihψk |ni . (10.20)
n,k
which shows that the trace of the density operator representation is always 1 regardless of the
basis of the matrix representation.
Expectation values of observables are expressed by,
X
hÂi = pk hψk |Â|ψk i . (10.22)
k
The Eq. (10.26) says that the ensemble average of any dynamic observable  can be calculated
from the diagonal elements of the operator matrix ρ̂Â: Since the trace is independent of the basis
(this will be shown in Exc. 10.6.1.1), each unitary transformation taking the matrix representa-
tion from a basis {|ni} to another one {|ti} leaves the trace invariant. Using the definition of a
3
In the presence of degeneracy or a continuous part of the spectrum we can generalize the definition of the
expectation, X
hX̂i ≡ Tr ρ̂X̂ = hkm|ρ̂X̂|kmi .
k,m
244 CHAPTER 10. THE BLOCH EQUATIONS
unitary transformation we can easily show that the trace of a cyclic permutation of a product
is invariant. For example,
and in particular
Tr [ρ̂Â] = Tr [Âρ̂] = hÂi . (10.28)
In the Excs. 10.6.1.2 and 10.6.1.3 we apply the density operator to pure and mixed states of a
two-level system. In Excs. 10.6.1.4 and 10.6.1.5 we study thermal mixtures.
That is, after the measurement, the operator becomes diagonal on the basis of the eigenvalues
of  4 .
We note that the density operator (10.29) describes the whole ensemble after the measure-
ment. The sub-ensemble corresponding to a particular result an of the measurement is described
by a different density operator,
P̂n ρ̂P̂n
ρ̂0n = . (10.30)
Tr [ρ̂P̂n ]
This is true, when |an i is the only eigenvector with the eigenvalue an . If not, P̂n in the expression
(10.30) should be replaced by the projection operator onto the sub-space of an 5
|ψ1 i ∈ H1 to the state ψ ∈ H2 ⊗ H2 . If, for example, in the case of a two particle wavefunction
Ψ(x1 , x2 ) there is no way to construct a wavefunction (i.e. a pure state) ψ1 (x1 ) describing the
state of the first particle, then |Ψ(x1 , x2 )i =
6 ψ1 (x1 )ψ2 (x2 ).
The point of the discussion is that, even if the total system is in a pure state, the various
subsystems that compose it will normally be in mixed states. On the other hand, regardless of
whether the composite system is in a pure or mixed state, we can perfectly construct a density
matrix that describes the state. Therefore, the use of density matrices is inevitable. Let ρ̂ be
the density matrix of the system composed of two subsystems. Then the state in H2 is described
by a reduced density operator given by the partial trace of ρ̂ over H2 . In the particular case,
where the state the density matrix has the form ρ̂ = ρ̂1 ⊗ ρ̂2 , where ρ̂1 and ρ̂2 are the density
matrices in H1 and H2 , then the partial trace is simply, Tr H2 ρ̂ρ̂1 .
Writing, X
ρ̂(t0 ) = pk |ψk (t0 )ihψk (t0 )| , (10.32)
k
we see immediately,
ρ̂(t) = U (t, t0 )ρ̂(t0 )U † (t, t0 ) , (10.33)
where, for the common case of a time-independent Hamiltonian,
Now we find the time derivative of the density operator differentiating the two sides of (10.33)
and substituting the Eqs.
dU 1 dU † 1
= ĤU and = − U † Ĥ (10.35)
dt i~ dt i~
dρ̂(t) i
= [ρ̂(t), Ĥ] . (10.36)
dt ~
The commutator itself can be considered as a superoperator acting, not any more on states but
on operators, that is, we can write,
i
Lρ̂(t) ≡ [ρ̂(t), Ĥ] , (10.37)
~
246 CHAPTER 10. THE BLOCH EQUATIONS
where L is called Liouville operator. The equation (10.36) is called Liouville equation or von
Neumann equation. The Liouville equation describes the time evolution of the density operator
which, in turn, describes the distribution of an ensemble of quantum states. Even though the
form of the Liouville equation resembles the Heisenberg equation, Eq. (10.31) shows that ρ(t) is
in the Schrödinger picture.
For a two-level system perturbatively interacting with a light field, the Hamiltonian can be
decomposed as in (2.133) into a stationary part and a time-dependent part,
where Ĥa is the part of the Hamiltonian describing the atomic structure and V̂ (t) the interaction
of the dipole transition with the classical oscillating electric field. The interaction picture is the
natural choice for this type of problem. In this case, we can transform the density operator into
the interaction picture defined by (2.134),
where the ’tilde’ decoration (replacing the ’hat’) emphasizes, that we are now in the interaction
picture. We look for the time evolution rate of ρ̃(t) analogously to the Liouville equation.
Calculating the time derivatives on both sides of (10.39) and substituting Eq. (10.36) for dρ dt
results in,
dρ̃(t) i
= [ρ̃(t), V̂ (t)] . (10.40)
dt ~
This equation shows that the time evolution of the density operator in the interaction picture
depends only on the time-dependent part of the total Hamiltonian.
dρ̂(t)
hm| |ni = ~i hm|[ρ̂(t), Ĥ]|ni = ~i hm|[ρ̂(t), ĤA + V̂ (t)]|ni (10.41)
dt
= ~i (En − Em )hm|ρ̂(t)|ni + ~i hm|[ρ̂(t), V̂ (t)]|ni ,
10.2. BLOCH EQUATIONS FOR TWO-LEVEL ATOMS 247
where |mi and |ni are members of a complete set of vectors of a basis {|ki} which P are also
eigen-kets of ĤA and span the space of Ĥ. Now, we insert the closing expression k |kihk| in
the commutator on the right-hand side of Eq. (10.41):
X
hm|[ρ̂(t), V̂ (t)]|ni = [hm|ρ̂(t)|kihk|V̂ |ni − hm|V̂ |kihk|ρ̂(t)|ni] . (10.42)
k
For our two-level atom the complete set only includes two states: |1(t)i = |1i and |2(t)i =
e−iω0 t |2i.
In addition, the matrix elements of the dipole coupling operator V̂ are only non-
diagonal,
V ≡ h1|V̂ |2i = h2|V̂ |1i . (10.43)
Hence, Eq. (10.41) adopts the form,
dρ̂11 i
= ~ [ρ̂12 V − ρ̂21 V ]
dt
dρ̂22 i dρ̂11
= ~ [ρ̂21 V − ρ̂12 V ] = −
dt dt
, (10.44)
dρ̂12 i
= iω0 ρ̂12 + ~ [V (ρ̂11 − ρ̂22 )]
dt
dρ̂21 dρ̂∗12
= −iω0 ρ̂21 + ~i [V (ρ̂22 − ρ̂11 )] =
dt dt
remembering that the dash of the diagonal terms, called populations, must be unitary, and that
the non-diagonal terms, called coherences, must be complex,
The above set of equations constitutes the optical Bloch equations in the Schrödinger picture.
It does not include loss terms due to spontaneous emission. We transform the Bloch equations
to the interaction picture by replacing the Liouville equation (10.36) by (10.40), and calculating
the matrix elements. We obtain,
dρ̃22 i dρ̃12 i
= V (ρ̃21 − ρ̃12 ) and = V (ρ̃11 − ρ̃22 ) . (10.46)
dt ~ dt ~
We would also have obtained this expression via the substitution, ρ̂12 = ρ̃12 eiω0 t . The interaction
picture simplifies the expressions for the temporal dependence of the coherences by eliminating
the first term on the right-hand side. Transforming to the interaction picture removes the
temporal dependence of the basis vectors spanning the Hilbert space of the two-level atom.
We have derived the optical Bloch equations from the Liouville equation, which is the fun-
damental equation of motion of the density operator, and we have seen how a unitary transfor-
mation can be used to represent these equations in the Schrödinger, Heisenberg or interaction
picture. So far, the Bloch equations do not include the possibility of spontaneous emission. We
will learn later, how to include this phenomenon.
wave approximation (RWA) is good, when the Rabi frequency is sufficiently small, Ω ω.
Otherwise, we observe an energy correction of the levels called Bloch-Siegert shift. The RWA
can be implemented in the time dependence of the coupling operator,
which, applied to the Bloch equations in the Schrödinger picture Eq. (10.44), yields,
dρ22 iΩ dρ12 iΩ
= (ρ21 − ρ12 ) , = −i∆ρ12 + (ρ11 − ρ22 ) . (10.49)
dt 2 dt 2
In Exc. 10.6.2.1 we derive the Bloch equations from the equations of motion for the population
amplitudes a1 and a2 .
For arbitrary starting conditions, the solution of these equations is not simple. To solve the
problem we write the equations in a matrix form,
i
ρ11 0 0 2Ω − 2i Ω
ρ22 0 0 − 2i Ω 2i Ω
~≡
ρ
ρ12 , A≡
Ω − Ω −i∆
, ~˙ = A~
ρ ρ . (10.50)
i
2 2
i
0
ρ21 − 2i Ω 2i Ω 0 i∆
To solve this system of differential equations, we calculate the eigenvalues of the matrix,
The coefficients follow from the Bloch equations with particular starting conditions. With a
little algebra we get
(1) 1
2 ∗ ∗
ρ22 = ρ22 (0) + 2G 2 |Ω| (1 − 2ρ22 (0)) − ∆ (Ωρ12 (0) + Ω ρ12 (0)) (10.53)
(2) 1
2 ∗ ∗
ρ22 = 4G 2 −|Ω| (1 − 2ρ22 (0)) + (∆ + G)Ωρ12 (0) + (∆ − G)Ω ρ12 (0)
(3) 1
2 ∗ ∗
ρ22 = 4G 2 −|Ω| (1 − 2ρ22 (0)) + (∆ − G)Ωρ12 (0) + (∆ + G)Ω ρ12 (0)
(1)
ρ12 = 1
2G2
[∆Ω(1 − 2ρ22 (0)) + Ω (Ωρ∗12 (0) + Ω∗ ρ12 (0))]
(2)
ρ12 = ∆−G
4G2
−Ω(1 − 2ρ22 (0)) + (∆ + G) ΩΩ∗ ρ∗12 (0) + (∆ − G)ρ12 (0)
(3)
ρ12 = ∆+G
4G2
−Ω(1 − 2ρ22 (0)) + (∆ − G) ΩΩ∗ ρ∗12 (0) + (∆ + G)ρ12 (0) .
To begin the discussion of this solution, let us consider a sample of atoms initially in the
ground state when the light field is switched on at time t = 0,
such that,
The electronic states are orthonormal hi|ji = δij , and we define the transition operators by
Many times we will restrict ourselves to atoms of two or three levels. For a two-level system we
obtain the Pauli spin matrix defined in (2.45). Every 2 × 2 matrix can be expanded on a Pauli
matrix basis,
ρ11 ρ12
= |1iρ11 h1| + |1iρ12 h2| + |2iρ21 h1| + |2iρ22 h2| (10.61)
ρ21 ρ22
= ρ11 ( 12 + 12 σ̂z ) + ρ12 σ̂ − + ρ21 σ̂ + + ρ22 ( 12 − 21 σ̂z )
− +
+ − − + − + hσ̂ σ̂ i hσ̂ − i
= ρ11 σ̂ σ̂ + ρ12 σ̂ + ρ21 σ̂ + ρ22 σ̂ σ̂ = .
hσ̂ + i hσ̂ + σ̂ − i
This formalism can easily be extended to an atom with many levels. Solve the Exc. 10.6.2.3.
For the two-level case it is useful to introduce an alternative notation based on the Bloch
vector defined in (2.46),
2 Re ρ12 hσ − i + hσ + i hσx i
β~ ≡ 2 Im ρ12 = i(hσ − i − hσ + i) = hσy i . (10.62)
ρ22 − ρ11 + −
hσ σ i − hσ σ i− + hσz i
the length of which is simply the Rabi frequency. With this, we can write the Bloch equations,
dβ~
= G × β~ , (10.64)
dt
as will be shown in Exc. 10.6.2.4. ρ12 describes the polarization and ρ22 − ρ11 the population
inversion of the atom. The equation is analogous to the equation of motion for a rigid rotor
or spinning top (for example, a dipole in a homogeneous field). It displays phenomena such as
precession and nutation. The physical content and usefulness of the Bloch vector will become
clearer when we use the formalism to analyze electric and magnetic couplings. In Exc. 10.6.2.5
we verify that the Bloch vector is normalized (as long as spontaneous emission is not considered).
that is,
0
~ = sin Ωt .
β(t) (10.66)
− cos Ωt
That is, a resonant pulse rotates a Bloch vector initially pointing in the direction −z within the
π π
plane z-y, until it arrives, at time t = 2Ω , at the +y direction and at time t = Ω at the +z
10.2. BLOCH EQUATIONS FOR TWO-LEVEL ATOMS 251
direction. This means that the entire population has been transferred to the excited state. The
Bloch vector continues to rotate (the movement is called nutation) around the torque vector G
which, as can be seen from Eq. (10.64), points at the +x direction when ∆ = 0. The nutation
frequency is proportional to the force Ω of the atom-field interaction. With the Eq. (10.57) we
see that the population oscillates between the ground and excited state with the frequency Ω.
This means that the energy ~ω is periodically exchanged between the atom and the field. A
pulse of resonant light of duration such that τ = π/2Ω is called a π/2-pulse. The nutation is
illustrated in Fig. 10.1(a).
Once the coherence has been excited by a detuned radiation, ∆ 6= 0, the Bloch vector does
not stand still, even after the radiation has been switched off. To see this, we consider again the
general solution (10.53) now entering Ω = 0. If the Bloch vector is initially at a point in the
unitary circle of the plane z-y, it will rotate according to the formula,
that is,
ρ12 (0) sin ∆t
~ = ρ12 (0) cos ∆t .
β(t) (10.68)
2ρ22 (0) − 1
That is, the Bloch vector performs a motion of precession around the symmetry axis. The
precession is illustrated in Fig. 10.1(b).
1 1
2ρ22-1
2ρ22-1
0 0
-1 -1
1 1
0 0
1 1
Re ρ -1 -1 0 -1 -1 0
12 Re ρ
Im ρ12 12 Im ρ12
Figure 10.1: (Code: LM Bloch V ectorRabi.m) (a) Nutation of the Bloch vector. The red circles
show the evolution of the Bloch vector on the Bloch sphere for a resonant π-pulse. (b) Precession
of the Bloch vector.
The evolution of the Bloch vector on the surface of the Bloch sphere under the influence
of radiation fields can be considered a coherent trajectory of the wavefunction of the atomic
state, which is therefore subject to interference phenomena [132]. Interferometers can be real-
ized by sequences of consecutive pulses splitting populations, exciting coherences, and remixing
populations.
Sensors based on interferometry of atomic excitation are nowadays among the most accurate
and most sensitive. We will discuss the method of radiation pulse sequences in several exercises:
In Exc. 10.6.2.6 we present the Rabi method and in the Excs. 10.6.2.7, 10.6.2.8, and 10.6.2.9 the
Ramsey method.
252 CHAPTER 10. THE BLOCH EQUATIONS
Example 39 (Langevin equation): The Heisenberg equation for the evolution of the
internal degrees of freedom, including the phenomenologically introduced decay, is also called
Langevin equation. It can be written as,
dσ̂
i = ~1 [σ̂, Ĥ] − 2i Γσ̂ ,
dt
and analogously for σ̂z . With the Hamiltonian Ĥ = ~∆σ̂ † σ̂ + 12 ~Ω(eiωt σ̂ + h.c.) we obtain,
using the Pauli spin matrices, exactly the Bloch equations,
iσ̂˙ = ∆[σ̂, σ̂ † σ̂] + 12 Ωe−iωt [σ̂, σ̂ † ] − 2i Γσ̂ = −∆σ̂z − 12 Ωe−iωt σ̂z − 2i Γσ̂
iσ̂˙ z = ∆[σ̂z , σ̂ † σ̂] + 1 Ωe−iωt [σ̂z , σ̂ † ] + 1 Ωeiωt [σ̂z , σ̂] − i Γσ̂z = −Ω(σ̂ † − σ̂) − i Γσ̂z .
2 2 2 2
10.3. BLOCH EQUATIONS WITH SPONTANEOUS EMISSION 253
1 2 1 i
4 |Ω| 2 Ω(∆ − 2 Γ)
ρ22 (∞) = 1 1 2 and ρ12 (∞) = ei∆t . (10.73)
∆2 + 2
2 |Ω| + 4 Γ ∆2 + 21 |Ω|2 + 14 Γ2
This will be shown in Exc. 10.6.3.1. We see that the populations and coherences both have a
frequency dependence, which is similar but not identical to the one already found for the sus-
ceptibility χ, the absorption coefficient K, and the absorption cross section σ0a in the Eqs. (??),
(??), and (??). But here the denominators have an extra term 12 Ω2 contributing to an effective
widths of ρ22 and ρ12 , p
Γeff = 2|Ω|2 + Γ2 . (10.74)
This effect is called power broadening or saturation broadening. The phase factor ei∆t describes
the optical precession of the Bloch vector.
By introducing the saturation parameter,
2|Ω|2
s≡ , (10.75)
4∆2 + Γ2
we can rewrite the stationary dipole moment and the excited state population (10.73) as,
s/2 ∆ − iΓ/2 s
ρ22 (∞) = , ρ12 (∞) = ei∆t . (10.76)
1+s Ω 1+s
and
s/2
ρ12 (∞)2 = . (10.77)
(1 + s)2
Fig. 10.2(a) shows the Rabi oscillations damped by spontaneous emission. For long times the
population of the excited state ρ22 converges to the asymptote (10.76). Fig. 10.2(b) shows the
temporal evolution of the Bloch vector subject to spontaneous emission.
Example 40 (Resonant excitation and weak excitation): A case where the Bloch
equations can be analytically treated is under resonant excitation, ∆ = 0. In this case, for
the initial conditions, ρ12 (0) = ρ22 (0) = 0, the solution including decay is,
Ω2 −3Γt/4 3Γ
ρ22 (t) = 1 − e cos λt + sin λt and ρ12 (t) = 0 , (10.78)
2|Ω|2 + Γ2 4λ
√
where λ ≡ Ω2 − Γ2 . This solution (which will be derived in Exc. 10.6.3.2), describes the
optical nutation of the Bloch vector along the ρz axis. We note here that, due to spontaneous
emission, the norm of the Bloch vector is NOT conserved, i.e., the Bloch vector evolves to
the interior of the Bloch sphere.
Another case that can be solved analytically is the weakly excited atom, |Ω| γ,
Ω2 −iΩ −(i∆+Γ)t
ρ22 ' 2
1 + e−2Γt − 2 cos ∆t and ρ12 (t) ' e − 1 +ρ12 (0)e−(i∆+Γ)t .
4G 2i∆ + 2Γ
(10.79)
254 CHAPTER 10. THE BLOCH EQUATIONS
(a) (b)
1 1
2ρ -1
0.5 0
ρ22
22
0 -1
0 -1
5 0
Ω /Γ Im ρ
12 12 12
6 0.5 1
4 0
10 2 1 -0.5
0 -1
Γ t
12 Re ρ
12
Figure 10.2: (Code: LM Bloch T woLevelDecay.m) (a) Rabi oscillations damped by spontaneous
emission for Rabi frequencies between Ω/Γ = 0.2, .., 5. (b) Evolution of the Bloch vector subject
to spontaneous emission (Γ12 = 0.05Ω12 ) after of a resonant π-pulse (blue) and after a π-pulse
with detuning ∆12 = Ω12 /2.
~Ω
χ = −n ρ12 ei∆t + c.c. (10.81)
2|E|2
We can insert the new expression (10.73) for ρ12 into our previous expression for hd12 i (??)
and get new expressions for the polarization P(t), (??) and (??), and the susceptibility χ (??).
The modified expression for the susceptibility is,
N d212 −∆ + 12 Γ
χ= . (10.84)
30 ~V ∆2 + 12 |Ω|2 + 14 Γ2
10.3. BLOCH EQUATIONS WITH SPONTANEOUS EMISSION 255
The important new property is the effective width Γeff , which appears in χ, K, and σ0a . In
Exc. 10.6.3.3 and 10.6.3.4 we calculate the impact of the spontaneous emission on the determi-
nant of the density operator. Solve the Excs. 10.6.3.5, 10.6.3.6, 10.6.3.7, and10.6.3.8.
˙
ρ̂(t) = (L0 + Lsp )ρ̂(t) with
, (10.87)
i Γ +
L0 ρ̂(t) ≡ ~ [ρ̂(t), Ĥ] and Lsp = 2 (2σ̂ ρ̂σ̂ − σ̂ + σ̂ ρ̂ − ρ̂σ̂ + σ̂)
where σ̂ ± are the Pauli matrices. We show in Exc. 10.6.3.9, that the known Bloch equations
can be derived from the master equation.
q 2
1 i 1
2 ∆− 2Γ ± 2 ∆ − 2i Γ + Ω2 . (10.89)
The real parts of the eigenvalues Re E describe shifts and/or splittings of the transition line.
The imaginary parts Im E describe broadening effects of the lines.
In the simplest case, ∆ = 0 and Γ > 4Ω, we find the saturation broadening already discussed
in (10.74), and we will deepen it in Exc. 10.6.3.10. For the case Γ < 4Ω, we observe a splitting
of the line called Autler-Townes splitting, which will be studied in Exc. 10.6.3.11. If Ω 6= 0,
the spectrum becomes asymmetrical. In the case of weak excitation, Γ 4Ω, we observe a
shift of the transition line with dispersive dependence (near the resonance) on the frequency
of the incident radiation. This is the dynamic Stark shift (or light-shift). In the case of strong
256 CHAPTER 10. THE BLOCH EQUATIONS
excitation, Γ 4Ω, we observe again at the split spectrum, but now the two lines exhibit
an avoided crossing-type dependence on the radiation frequency. We study these effects in
Exc. 10.6.3.12.
Obviously, these effects can be studied by the Bloch equation formalism containing the terms
of spontaneous relaxation.
Ω= √1 Γ . (10.90)
2
10.4. LINE BROADENING MECHANISMS 257
We can use equation (10.90) to define the saturation intensity Isat for an atom with the transition
dipole d12 . From Eq. (1.112) we have,
s
2I¯
E0 = . (10.91)
0 c
Therefore, using the definition of the Rabi frequency, ~Ω = d12 E0 , and the relationship between
d12 and Γ given by Eq. (??), we have,6
g1 2π 2 c~
Isat = Γ , (10.92)
g2 3λ30
taking into account the degeneracies gj of the levels. In Exc. 10.6.4.4 we calculate the saturation
intensity of popular atomic transitions.
γ 0 = γ + γcol . (10.94)
The inverse of the collision rate is simply the time between phase interruptions or the time
between collisions. Now, for collisions between hard cores of atoms of mass m (with reduced
mass mred = m/2) and with radius ρ in a gas with density n consisting of a single species,
a standard analysis based on the kinetic theory of dilute gases shows that the time between
collisions is given by the collision rate,
−1
γcol = τcol = σnv̄ , (10.95)
q
8kB T
where v̄ = πm is the average collision velocity in a homogeneous gas at the temperature T
√ 2
red
8ρ2 n
γcol = p . (10.96)
mred /πkB T
6
Some authors define the saturation for s = 2, as happens when Ω = Γ.
258 CHAPTER 10. THE BLOCH EQUATIONS
We can now relate this simple result of gas kinetics to the phase interruption rate by reinter-
preting the meaning of the collision radius. When an excited atom propagating through space
suffers a collision, the long-range interaction will produce a time-dependent perturbation of the
energy levels of the radiating atom and a phase shift in the radiation,
Z ∞ Z ∞
η= [ω(t) − ω0 ]dt = ∆ω(t)dt . (10.97)
−∞ −∞
The integral is easily assessed for the two most frequent cases: non-resonant van der Waals
interactions n = 6 and resonant van der Waals interactions n = 3. The phase shifts are,
2π C6 4π C3
η6 (b) = and η3 (b) = √ . (10.100)
3~ b56 v 3 3~ b23 v
Now, if instead of using the hard core approximation, we define a collision as an encounter
causing a phase shift of at least 1 radians, we have a new condition for the collision radius,
2π C6 1/5 4π C3 1/2
b6 = and b3 = √ . (10.101)
3~ v 3 3~ v
Replacing these collision radiuses for the radius ρ in Eq. (10.96) and inserting the average
collision velocity, we find the collision rate,
√ 2 !2/5 3/2 2 3/10
2π C6 4πkB T 2 π C3
γc6 = 4n and γc3 = 4n . (10.102)
3~ µ 3 ~
Substituting the generalized γ 0 of (10.94) for γ in the Bloch equations (10.73), we find the
stationary solutions,
1 γ0 2 1 0
4 γ |Ω| 2 Ω(∆ − iγ )
ρ22 = and ρ12 = ei(ω−ω0 )t . (10.103)
1 γ0 0
∆2 + 2 γ |Ω|
2 + γ 02 ∆2 + 12 γγ |Ω|2 + γ 02
The effective linewidth (radiative and collisions) is,
q
0
Γ0eff = 2 γ 02 + 12 γγ |Ω|2 . (10.104)
When the excitation is sufficiently weak, so that power broadening can be neglected in compar-
ison to collision broadening, the second term can be discarded,
Γ0eff = 2(γ + γcol ) . (10.105)
The equations (10.74) and (10.105) express the linewidths in the limits of dominating power
and collision broadening, respectively. Note that the susceptibility, absorption coefficient, and
absorption cross-section retain their Lorentzian profile, but with a larger width due to collisions.
Since each atom is subject to the same broadening mechanism, the broadening is homogeneous.
Solve Excs. 10.6.4.5 and 10.6.4.6.
10.4. LINE BROADENING MECHANISMS 259
From Eq. (10.107) we can see that the line profile is,
1 m −(ω−ω0 )2 /2σ2
D(ω − ω0 ) ≡ √ e dω . (10.110)
2π kB T
The profile compares with the Lorentzian profile Eq. (??) associated with natural, power, or col-
lision broadening. Doppler broadening is a property of the atomic ensemble, each atom suffering
a unique but different displacement than the other atoms. Hence, it is called inhomogeneous
broadening.
The Heisenberg equation used to derive the Bloch equations assumes immobile atoms. How-
ever, we can easily apply the Galilei transformation to a system, where the atoms move with
the given velocity v,
(∂t + v · ∇)ρ(r, t) = − ~i [Ĥ, ρ(r, t)] . (10.111)
Since the light fields propagate as ei(ωt−k·r) , the solution of the above equation simply follows
from the immobile solution with the substitution ∆ → ∆ − k · v.
2
For a cloud obeying Maxwell’s velocity distribution, P (v) ∼ e−mv kB T ,
Z
1 2 2
ρ̄(∆) = √ e−(k·v) /2δ ρ(∆ − k · v)d(k · v) . (10.112)
2πδ R
The average of the density operator over all velocities, ρ̄, therefore follows as the convolution
of the density operator ρ (obtained as the solution of the Bloch equation) and the Gaussian
2 2
function G(∆) = (2πδ 2 )−1/2 e−∆ /2δ ,
where the matrix X contains the modulation index of the atomic motion kij v. The stationary
solution of the differential equation, averaged over the time of an oscillation period, can be
expressed as an infinite continuous fraction:
1
ρ(∞) = −(L + S+ + S− )−1 b where S± = −X 1 . (10.117)
L ± iΩa 1 − X L±iΩ 1 X
a 1−X ··· X
This solution replicates the correct excitation spectra even for a multilevel system.
Let us be more specific for a two-level system. Its Hamiltonian is, Ĥint = 12 ~Ωe−i[ωt−k·v/Ωa sin Ωa t] ,
such that the Bloch equation is,
ρ −Γ − 2i Ω 2i Ω 0 0 0 ρ22 0
d 22 i
ρ12 = − 2 Ω −Λ 0 + 2 cos Ωa t 0 − 2i kv 0 ρ12 + 2i Ω .
dt i
ρ21 i
2Ω 0 −Λ∗ 0 0 2 kv ρ21 − 2i Ω
P (10.118)
We look for the stationary solution by letting ρ̇n = 0 and ρ = ∞ n=−∞ n ρ e −inΩa t and projecting
on e−inΩa t by,
(L + inΩa 1)ρn + X(ρn+1 + ρn−1 ) + bδn0 = 0 . (10.119)
Now we define ρn±1 = S± ρn for n R 0. Then, equation (10.119) becomes,
− −1
ρ0 = −[L + X(S+
0 + S0 )] b para n = 0, (10.120)
S±
n∓1 = −[L + inΩa 1 + XS± −1
n] b for n≷0.
10.5. MULTI-LEVEL SYSTEMS 261
0.2
0.15
22
0.1
ρ
0.05
0
-10 -5 0 5 10
Δ
12
Figure 10.3: (Code: LM Bloch M icromotion.m) Spectral broadening due to the periodic move-
ment of the atom.
The levels have the energy ~ωi above the ground level.
Let us first have a look at the coherent part of the master equation. The Hamiltonian in the
semiclassical approximation (that is, the atom is quantized and consists of several levels |ii with
energies ~ωi , while the light fields are described by factors eiωij t , with frequencies ωij tuned near
the transitions |ii − |ji) includes the following contributions
X X
Ĥ = Ĥatom + Ĥatom−f ield = |ii~ωi hi| + |ii ~2 Ωij hj|eiωij t + c.c. . (10.123)
i
i<j com Ei <Ej
The Rabi frequency Ωij is a measure for the force at which the levels |ii and |ji are coupled
by the resonantly irradiated light field. The master equation can be simplified by applying the
262 CHAPTER 10. THE BLOCH EQUATIONS
rotating wave approximation and transforming to the coordinate system which rotates with the
light frequencies ωij :
ρij → ρ̂ij eiωij t , Ĥatom−f ield → e−iĤt/~ Ĥatom−f ield eiĤt/~ . (10.124)
Finally, the master equation can be reformulated by introducing a generalized Bloch vector, ρ,
and the matrix representation of the Liouville superoperator L as a linear system of n2 coupled
differential equations,
d
ρ = Lρ , ρ = (ρ11 .. ρnn ρ12 ρ21 .. ρn−1 n ρn n−1 ) . (10.125)
dt
Alternatively to the complex formulation, the differential equations can be written for the
real and imaginary part of the Bloch vector. The components ρii correspond to the population
probabilities of the levels |ii, the non-diagonal elements ρij describe the coherences between |ii
and |ji. Now, we must insert the Hamiltonian (10.123) and the density operator ρij into the
Liouville equation (10.36) in order to derive the generalized Bloch equations. In practice, these
calculations are simple but heavy. Therefore, we describe in Sec. 10.5.4 a simplified recipe for
compiling Bloch equations for arbitrary level systems for real atoms.
For example, for the two-level system with the definition of the external product (2.109):
L0 = −iĤ ⊗ I + iI ⊗ ρ̂ . (10.128)
Example 41 (Liouville equation for two levels): For example, for the two-level system,
2γ11 0 0 2γ21 2(γ11 + γ12 ) P 0 0 0
0 0 0 0 0 (kj) γ(kj) 0 0
L0 =
0
−
P .
0 0 0 0 0 (kj) γ(kj) 0
2γ12 0 0 2γ22 0 0 0 2(γ22 + γ21 )
Figure 10.4: Three level system (a) in Λ-configuration, (b) in V -configuration, and (c) in cascade
configuration.
Defining the Bloch vector by ρ ~, the Bloch equation matrix for three levels in Raman config-
uration (that is, in Λ-configuration) using the labeling of Fig. 10.4(a), is,
~˙ = L~
ρ ρ (10.130)
i
0 Γ12 Γ13 2 Ω12 − 2i Ω12 0 0 0 0 ρ11
0 −Γ12 − Γ23 0 − 2i Ω12 i
2 Ω12 0 0 i
Ω
2 23 − 2i Ω23 ρ22
0 Γ23 −Γ13 0 0 0 0 − 2i Ω23 i
2 Ω23 ρ33
i
− 2i Ω12 −Λ12 i
2 Ω12 0 0 Ω
2 23 0 0 0 ρ12
=
− 2i Ω12 i
2 Ω12 0 0 −Λ∗12 0 − 2i Ω23 0 0
ρ21
0 0 0 i
Ω
2 23 0 −Λ13 0 - 2i Ω12 0 ρ13
0 0 0 0 − 2i Ω23 0 −Λ∗13 0 i
Ω
2 12
ρ31
i
0 2 Ω23 − 2i Ω23 0 0 − 2i Ω12 0 −Λ23 0 ρ23
0 − 2i Ω23 i
2 Ω23 0 0 0 i
Ω
2 12 0 −Λ∗23 ρ32
7
For the same reason that the three-body problem has no general analytic solution.
264 CHAPTER 10. THE BLOCH EQUATIONS
These matrices serve to calculate, among others, the phenomena of Autler-Townes splitting
treated in Exc. 10.6.3.11, of the light-shift treated in Exc. 10.6.3.12, the dark resonances treated
in Exc. 10.6.5.2, the STIRAP method treated in Exc. 10.6.5.3, adiabatic sweeps treated in
Exc. 10.6.5.4, the dispersive interaction between atoms and light treated in Exc. 10.6.5.5, Fano
resonance-type line profiles of dark resonances treated in Exc. 10.6.5.6, and the quantum jumps,
which will be studied in later chapters. In Excs. 10.6.5.7 and 10.6.5.8 we will show, that an atomic
gas may have negative permittivity and negative permeability and, consequently, properties
usually only found in artificial metamaterials, as for example, a negative refractive index.
~red , has the length n2 −1, and from L we obtain the (trace-)reduced,
The resulting state vector, ρ
now invertible matrix Lred and the inhomogeneity vector b. The differential equation is now,
d
~red = Lred ρ
ρ ~red + b , (10.136)
dt
with the stationary and time-dependent solutions,
Once the matrix L or the matrix Lred and the inhomogeneity vector b are determined for
a system, the state of the atom can be calculated at any time, as well as the populations
and coherences. The system’s free parameters are the natural transition linewidths and the
detunings, as well as the intensities and emission bandwidths of the incident light fields.
time intervals ∆t sufficiently short, so that the Hamiltonian (or the Liouvillian) can be considered
constant during this interval, and we propagate the wavefunction (or the Bloch vector) to later
times via:
|ψ(t + ∆t)i = eiĤ(t)∆t |ψ(t)i or ~(t + ∆t) = eL(t)∆t ρ
ρ ~(t) , (10.138)
and insert the solution obtained again into equations (10.138) with the Hamiltonian Ĥ(t + ∆t)
(or the Liouvillian L(t + ∆t)) adjusted to the new time.
Example 42 (Electromagnetically induced transparency ): In some special cases, the
three-level Bloch equations can be solved analytically. The system in Λ-configuration schema-
tized in Fig. 10.4(a), where the two lasers satisfy the condition ∆12 = ∆23 can exhibit a dark
resonance leading to the phenomena of electromagnetically induced transparency (EIT) and
electromagnetically induced absorption. In these resonances a dramatic change of the refrac-
tive index is observed despite the fact that the atom becomes transparent, Re χ 0 and
|Im χ| Re χ: p
Re n = 1 + Re χ 0 ,
resulting in a high group velocity,
c
vg = dn
.
n + ω dω
EIT is usually studied in Λ-type systems, but similar phenomena can be found in cascade-
type systems [262, 259], which will be studied here. Disregarding the decay rate Γ13 , the
Bloch equations (10.130) and (10.133) give the coherences,
iΩ12 iΩ23
ρ̇12 = −Λ12 ρ12 + 2 (ρ11 − ρ22 ) − 2 ρ13
ρ̇13 = −Λ∗13 ρ13 − iΩ12 iΩ23
2 ρ23 − 2 ρ12
iΩ23 iΩ12
ρ̇23 = −Λ23 ρ23 + 2 (ρ22 − ρ33 ) − 2 ρ13 .
Assuming stationarity and negligible depletion of the ground state, ρ11 = 1,
iΩ12 iΩ23
0 = −Λ12 ρ12 + 2 − 2 ρ13
0 = −Λ∗13 ρ13 − iΩ12 iΩ23
2 ρ23 − 2 ρ12
iΩ12
0 = −Λ23 ρ23 − 2 ρ13 .
N d12 N |d12 |2
χ= ρ21 = ρ21 .
V ε0 E12 V ε0 ~Ω12
1
For a resonant probe laser, ∆23 = 0 and with Γ13 ' 0, we have Λ13 = 2 Γ23 + i∆12 and
1 Ω2
Λ23 = 2 (Γ23 + Γ12 ). The susceptibility in the probe transition is now, using Θ ≡ Γ23 + 2Λ1223 ,
2
N |d12 |2 Ω
Γ23 + 2Λ12 −2i∆12
χ= iΩ12 23
V ε0 ~Ω12
2
Ω
Γ23 + 2Λ12 −2i∆12 (Γ12 +2i∆12 )+Ω223
23
2
N |d12 |
= iΩ12 (Θ−2i∆12Θ−2i∆ 12
)(Γ12 +2i∆12 )+Ω223
= χ0 + iχ00 .
V ε0 ~Ω12
266 CHAPTER 10. THE BLOCH EQUATIONS
p 1
n= 1+χ'1+ χ .
2
Its imaginary part originates from the decay term of the atom: it is here responsible for the
absorbing nature of the cloud. EIT is characterized by a pronounced dispersion and a small
concomitant absorption.
−3
x 10
2 0.5
1.5
ρee
1 Re χ
0.5 0
0
−20 −10 0 10 20
0.2
−0.5
0.15 −20 −15 −10 −5 0 5 10 15 20
mm
0.1
ρ
0.05
1
0
−20 −10 0 10 20 0.8
1
0.6
Im χ
0.95
0.4
gg
0.9
ρ
0.2
0.85
0.8 0
−20 −10 0 10 20 −20 −15 −10 −5 0 5 10 15 20
Δge/Γge Δge/Γge
Figure 10.5: (Code: LM Bloch EitT hreeLevelCascadeStat.m) EIT signal for the cascade sys-
tem of strontium with the transitions at 689 nm and 497 nm with Ω12 = Γ12 , Ω23 = Γ23 and
∆23 = 0. The red lines are calculated by numerical integration of the Bloch equations. The
dotted lines are obtained from analytical formulas based on the assumptions of weak ground
state depletion (which is not really correct in the chosen parameter regime.
i
ρ̂˙ = − [H, ρ̂] + LΓ ρ̂ + Lβ ρ̂ . (10.139)
~
10.5. MULTI-LEVEL SYSTEMS 267
describe dissipation through spontaneous emission from the various excited states and decoher-
ence due to phase fluctuations of the laser fields, respectively. The operators σkl = |kihl| describe
the transition between two states. Γkl is the spontaneous decay of level |ki in to |li, and βkl
is the emission bandwidth of the laser coupling the two levels. Simple but tedious algebraic
transformations of the master equation lead, in the rotating wave approximation, to a set of
linear first-order differential equations in the populations of the atomic excitation levels and the
coherences between them. The equations are called the optical Bloch equations.
Alternatively, the Bloch equations may be found by breaking down the multi-level scheme
into a set of three-level systems. Respecting a few symmetry considerations, the multi-level
Bloch equations can then be reassembled from the three-level Bloch equations corresponding
to every possible combination of three levels. Based on such considerations, we provide in the
following a simple recipe for setting up Bloch matrices for arbitrary level schemes. Let us regard
a n-level atom. Its internal state is fully described by the populations ρkk and the (complex)
coherences ρkl , with k, l = 1, .., n. In this work we describe the coherences by their real and
imaginary parts. The labeling is such that the levels are sorted according to their excitation
energy, Ek < El for k < l. We define the Bloch vector,
~ ≡ (ρ11 ...ρnn Im ρ12 Re ρ12 Im ρ13 Re ρ13 ...
ρ (10.141)
...Im ρ1n Re ρ1n Im ρ23 Re ρ23 ...
...Im ρn−1,n Re ρn−1,n ) .
The Bloch equations then formally read,
~˙ = M ρ
ρ ~, (10.142)
where in the given Bloch vector basis the matrix M has the following structure,
(A) (B1 )
M = (C) (D) . (10.143)
(B2 )
(D) (C)
The different blocks of the matrix have the following significations. Block A handles the
transfer of populations by spontaneous decay. Its rank corresponds to the number of levels n.
The diagonal elements of this block are the decay rates Γ of the excited states. The off-diagonal
elements Γkl denote the gain of level k from a decaying level l. Conservation of energy P thus
requires that the sum of the transition rates cancels for every column of matrix A, Γ = k Γkl ,
as it is the case for the two-level Bloch matrix. If the levels are sublevels of a Zeeman and/or
hyperfine split multiplet, the rates have to be weighted with Wigner’s (3j) and {6j} symbols,
Γkl = ΓSkl . The relative oscillator strengths Skl are given in Sec. 10.5.4.2.
The blocks Bk treat the interdependence of the populations and the coherences. B1 describes
how the coherence between any pair of states driven by a light field generating a Rabi frequency
Ωkl influences the populations. The block consists of convoluted 2 × 2 matrices of the form,
ρ̇kk −Ωkl 0 Im ρkl
∼ . (10.144)
ρ̇ll Ωkl 0 Re ρkl
268 CHAPTER 10. THE BLOCH EQUATIONS
The Rabi frequencies have to be weighted not only with the relative oscillator strength Skl , but
also with the projection Hkl of the laser polarization onto the orientation of the magnetic field
and the laser polarization, Ωkl = Ωx Skl Hkl . Here Ωx is the Rabi frequency generated by a laser
on a transition, whose oscillator strength is 1. The projection is calculated in Sec. 10.5.4.3 for
the three possible laser polarizations, i.e. for σ ± and for π light.
The matrix C rules the influence of the decays of the coherences, of the detunings ∆kl =
ωx − ωatom , and the laser linewidths βkl . Note that the detuning of the laser frequency ωx is
negative for red-detuned light. In the chosen basis it breaks down into an array 2 × 2 matrices
aligned along the diagonal of M . Their shape is,
γkl
Im ρ̇kl − 2 −∆kl Im ρ̇kl
∼ . (10.146)
Re ρkl ∆kl − γ2kl Re ρkl
P
where γkl = m,Em <Ek ,El (Γkm + Γlm ) + 2βkl . Often the levels are sublevels of a Zeeman and/or
hyperfine split multiplets. In this case the frequency shift Zkl of the level is added to the detuning
∆kl . The shift is calculated in Sec. 10.5.4.4 for the example of the 6 Li D2 line.
The block D governs the interdependences of all laser-driven coherences of the atom. The
block contains 2 × 2 submatrices at any place of the matrix M , where the row index pair (mn)
and the column index pair (kl) have one index in common provided the two different indices
correspond to the Rabi frequency of an incident laser:,
Im ρ̇mn 0 ± 12 Ωkl Im ρpq
∼ . (10.147)
Re ρ̇mn ± 21 Ωkl 0 Re ρpq
The submatrix elements indexed by column (pq) and row (mn) are non-zero if one of the indices
p or q is equal to one of the indices m or n and the unequal indices correspond to a laser-driven
transition. In order to find the correct signs of the submatrix elements, we distinguish four
cases: 1. For m = p, n = k, and q = l the signs are: (− + ); 2. for n = q, m = k, and p = l the
signs are: (+ − ); 3. for m = q, n = k, and p = l the signs are: (+ + ); and 4. for n = p, m = k,
and q = l the signs are: (− − ). A proper parametrization is proposed in the next section.
µ(k, k) = k (10.148)
2
µ(k, l) = 2nk − n − k − k + 2l − 1 ,
so that,
Spontaneous transitions between hyperfine- and Zeeman split levels have to be weighted ac-
cording to the Wigner-Eckardt theorem using Clebsch-Gordan (3j) and Wigner {6j} symbols.
Consider the transition |(Jk , I)Fk , mk i ↔ |(Jl , I)Fl , ml i. The relative oscillator strength is,
2
Fk κ Fl
Skl = (10.158)
mk sign (ml − mk ) −ml
2
Jl Jk κ (2Fk + 1)(2Jl + 1)(2κ + 1)
.
Fk Fl I 2I + 1
The transition rates additionally depend on the relative orientation of the laser polarizations and
the magnetic field direction. This dependence is accounted for by decomposing the polarization
vector into the,
B ê3 × ĝ ê2 × ê3
ê3 = , ê2 = , ê1 = , (10.159)
B |ê3 × ĝ| |ê2 × ê3 |
where ĝ is an arbitrarily chosen direction, e.g. gravity. The relative amplitude of the transitions
∆mJ = 0 is proportional to the projection of the polarization vector on the magnetic field axis
ζ0 = (ε̂ · ê3 )2 for π-polarized light. To estimate the amplitude of the transitions ∆mJ = ±1, we
must project onto the coordinates,
With this generalization the Bloch equations can e.g. be employed to calculate Hanle res-
onances quantum mechanically. The Hanle effect occurs when a magnetic and an optical field
compete for the quantization axis.
The nuclear spin of the 6 Li atom is I = 1, its electron spin is S = 21 . The excitation states
are characterized by quantum numbers Jk , Fk , mk . The electron angular orbital momentum is
Lk = δk≥7 , and the electron angular orbital momentum is Jk = 21 δk≤6 + 32 δk≥7 . The hyperfine
structure of the excited state 2 P3/2 can be written as νhf 1 = −2.8 MHz, νhf 2 = 0 MHz, and
νhf 3 = 1.7 MHz. Hence, the hyperfine splitting is inferior to the natural decay rate Γ =
(2π) 6 MHz,
µB |B|
Zkl = 2π~ (gFk mk − gFl ml ) (10.162)
+ νhf 1 δ7≤l≤8 + νhf 2 δ8≤l≤13 + νhf 3 δ13≤l≤16 ,
10.6 Exercises
10.6.1 Density matrix
10.6.1.1 Ex: Trace of an operator
P
The trace of an operator  is defined by Tr  = n hn|Â|ni.
a. Show that the trace is independent of the chosen basis!
b. Show that Tr ÂB̂ = Tr B̂ Â!
e−nβ~ω 1
Pn = P −mβ~ω
with β≡ .
me kB T
Now imagine a device capable of measuring the speed of a single atom randomly chosen within
the cloud.
a. Express the probability of measuring a specific velocity v 0 for this atom using the density
operator.
b. Express the expectation value of the average velocity by the density operator.
Derive the Bloch equations explicitly based on the temporal evolutions of the coefficients a1,2
(5.63) knowing that ρij = a∗i aj .
Verify k~
ρk = 1.
Free atoms be illuminated by light pulses characterized by the Rabi frequency Ω, whose pulse
Rt
area is (i) Ω dt = π and (ii) = 2π. For which frequency tuning ∆ = ω − ω0 the excited state
0
population is maximum? Draw the spectral profile of the population in the range −5 < ∆/Ω < 5.
Many atomic clocks work according to the Ramsey spectroscopy method: The two-level atom
is resonantly excited by a microwave π/2-pulse. Then, the phase of atomic coherence precesses
freely over a period of time T accumulating an angle φ. Finally, a second π/2-pulse is applied
and the population of the upper-level is measured. Calculate this population as a function of
the angle φ. Neglect spontaneous emission.
Derive the analytic formula for the final population ρ22 for the Rabi and Ramsey experiments.
Derive and compare the line widths of the ’interference fringes’ in these two experiments.
10.6. EXERCISES 273
10.6.2.9 Ex: Atomic clocks by the Ramsey method with spontaneous emission
In this exercise we study the Ramsey method used in atomic clocks. For this, we will consider
a two-level system |1i and |2i excited by a microwave radiation field characterized by the Rabi
frequency Ω12 , and we will compare two cases: without and with spontaneous emission:
a. Write down the Hamiltonian of the system, propose a sequence of pulses allowing the observa-
tion of the Ramsey fringes, do a numerical simulation of the Schrödinger equation (based on the
prescription (10.138)), and prepare a graph of the type 10.1 illustrating the temporal evolution
of the Bloch vector during the sequence.
b. Calculate numerically from the Schrödinger equation the population ρ22 immediately after
the pulse sequence as a function of the detuning ∆12 of the radiation field, and prepare a graph
of the spectrum. Also, assuming a decay rate of Γ12 = 0.1Ω12 , calculate the population ρ22 as a
function of detuning ∆12 from the Bloch equations (making sequences of type (10.138)), prepare
a new graph, and compare it with the previous graph obtained by the Schrödinger equation.
c. What happens to the width of the fringes, when the free precession time τ between the Ram-
sey pulses is increased? Prepare a graph of the inversion 2ρ22 − 1 as a function of ∆12 and τ
and interprete the results.
’Photon echo’ is a powerful spectroscopic technique that allows circumvention of certain dephas-
ing processes, for example, the Doppler shift due to the atomic motion in a thermal sample of
atoms. The technique resembles the Ramsey method with the difference, that between the two
Ramsey π/2-pulses, that is, during the free precession time, we apply an additional π-pulse,
which inverts the imaginary part of the coherence. We will study this method by numerical
simulation of the Schrödinger equation and the Bloch equations for a two-level system with and
without spontaneous emission:
a. Write down the Hamiltonian of the system and do a numerical simulation of the Schrödinger
equation (concatenating the pulses as explained in Eq. (10.138)) for the following temporal pulse
sequence:
(i) resonant π/2-pulse (∆12 = 0) choosing Ω12 = 2,
(ii) evolution for a time T without radiation (Ω12 = 0),
(ii) resonant π/2-pulse using the same parameters as in (i),
(iv) evolution for a time T without radiation, and
(v) resonant π/2-pulse identical to the first pulse.
Prepare a graph of type 10.1 illustrating the temporal evolution of the Bloch vector during the
sequence. Now, repeat the sequence taking into account a possible Doppler shift leading to
∆12 6= 0.
b. Repeat the calculation of (a), now numerically solving the Bloch equations, which allow the
occurrence of spontaneous emission (Γ12 = 0.03Ω12 ). Interpret the results.
Derive the stationary solution of the Bloch equations including spontaneous emission.
274 CHAPTER 10. THE BLOCH EQUATIONS
Derive the solution (10.78) of the Bloch equations with spontaneous emission for resonant exci-
tation.
In Sec. 10.1.1 we already saw that det ρ̂ = 0 for conservative systems. Now, show explicitly for
the Bloch matrix of a two-level system, that det ρ = 0 only holds in the absence of spontaneous
emission.
A two-level atom with decay rate Γ = 2π × 6 MHz be excited by a light field detuned by ∆ = 2Γ
and whose intensity is a quarter of the saturation intensity. Write down the Bloch vector for
t → ∞.
~ ≡ (2Reσ+ , 2Imσ− , σz ),
Check the temporal evolution of the norm of the Bloch vector defined by ρ
where the σk are the Pauli matrices, for a resonantly excited two-level system with and without
spontaneous emission.
Show that the general form of the master equation: ρ̇ = − ~i [Ĥ, ρ] − Γ2 (2σρσ + − σ + σρ − ρσ + σ),
reproduces the Bloch equations including spontaneous emission.
10.6. EXERCISES 275
6π Γ2
σ(v) = .
k 2 4(ω − ω0 − kv)2 + Γ2
The normalized one-dimensional Maxwell distribution,
r
m 2
ρ(v)dv = e−mv /2kB T dv .
2πkB T
a. Calculate the absorption profile of the resonance line at 461 nm (Γ461 = (2π) 30.5 MHz) of a
strontium gas cooled to the Doppler limit (kB TD = ~Γ) of this transition.
b. Calculate the absorption profile of the resonance line at 689 nm (Γ689 = (2π) 7.6 kHz) of a
strontium gas cooled to the Doppler limit of the transition at 461 nm.
c. Compare the optical densities in case of resonance.
Help: To evaluate the convolution integral approximate the narrower distribution by a δ-
function maintaining the integral over the distribution normalized.
The gas is at T = 300 K, where the partial pressure of rubidium is around P = 10−1 mbar.
The length of the cell is L = 10 cm. The laser has an intensity below the saturation limit, such
that the cross section of an atom moving at velocity v is,
6π Γ2
σ(v) = .
k 2 4(ω − ω0 − kv)2 + Γ2
The saturation laser has high intensity. We suppose here, Ω ≡ 10Γ, where Ω is the frequency of
Rabi caused by the saturation beam. In this way it creates a population Na nd of atoms in the
10.6. EXERCISES 277
excited state. As this population lacks in the ground state, Ng = N − Ne , the absorption of the
proof beam is decreased by the factor,
Ne Ω2
= .
N 4(ω − ω0 + kv)2 + 2Ω2 + Γ2
R∞ Ng −Ne
Calculate for laser proof spectrum of optical density, OD(ω) = Ln −∞ N σ(v)ρ(v)dv, and
the intensity of light transmitted through the cell, II0 = e−OD .
XN 𝑁 Z N
𝑁
S= ∆s(vn ) ' ∆s(v )dn .
𝑆 = ∑ Δ𝑠(𝑣𝑛 ) ≈ 0∫ Δ𝑠(𝑣𝑛n) d𝑛
n=0
𝑛=0 0
Calcule a distância média necessária para que os átomos sejam freados até 𝑣 = 0 (ignore o limite
Doppler). Deixe em função de Γ, 𝑣0 , Δ𝑣, 𝑘 e 𝑊0. Dica: faça a mudança de variável 𝑛 → 𝑣 na integral,
isso pode economizar alguns cálculos. (1,5 pontos)
c) Tipicamente, a dessintonia da luz |𝛿| = 𝑘𝑣0 é bem maior do que a largura natural Γ da transição. O
que acontece com 𝑆 no limite em que 𝑘𝑣0 ≫ Γ? Interprete esse resultado, justificando a necessidade
278 CHAPTER 10. THE BLOCH EQUATIONS
Calculate the average distance required for the atoms to be slowed down to v = 0 (ignoring the
Doppler limit). Write the expression as a function of Γ, v0 , k, and W0 . Help: Do the following
change of variables to simplify the evaluation of the integral: n → v.
c. Typically, the detuning of the light, |δ| = kv0 , is much larger than the natural width Γ of the
transition. What happens to S in the limit when kv0 Γ? Interpret this result, justifying the
need for the Zeeman-slowing technique.
during a time interval of 2 ms. With this, solve the Schrödinger equation (10.130) iteratively
varying the detuning.
b. Write down the Liouville matrix of the system and do a numerical simulation of the Bloch
equations (10.130) using the same parameters as in (a). Interpret the results. What you observe
when you introduce a decay rate between adjacent levels of Γ/2π = 200 Hz?
Figure 10.8: Energy levels of an atom in the ground state with Zeeman structure (for example,
|J = 1, mJ = −1, 0, +1i) as a function of the applied magnetic field.
from the Bloch equations (10.130) with the Liouville matrix Lred reduced by the trace condi-
tion(10.134), the spectrum ρ22 (∆2 3) for the following set of parameters: Γ12 = 2, Γ23 = Γ12 /2,
Γ23 = 0.1Γ12 , Ω12 = 10Γ12 , Ω23 = 5Γ23 , ∆12 = −5Γ12 e ∆23 = [−1 : .01 : 1]Γ23 . Interpret the
spectrum in terms of a Fano resonance.
Figure 10.9
a. Consider a three-level system in Λ-configuration. The transitions |1i-|2i and |2i-|3i are as-
sumed to be electric dipoles and |1i-|3i a magnetic dipole, such that, Γ12 , Γ23 Γ13 . Extract
from the Bloch equation (10.130) the equations for the coherences ρ12 , ρ13 , and ρ23 .
b. Suppose, that the excitation on the probe transition be so weak, Ω12 Γ12 , that it does not
succeed to empty the ground state. In this approximation eliminate the dynamics of ρ23 and
deduce the stationary solution for ρ12 and ρ13 .
c. Calculate the magnetic susceptibility χm = [228] with the following parameters Γ12 =
7 · 107 s−1 , Γ23 = 3 · 107 s−1 , Γ13 = 2 · 107 s−1 , Ω12 = 0.1Γ12 , Ω23 = 2 · 108 s−1 , ∆23 = 0
in the regime ∆12 = [−15Γ23 , 15Γ23 ].
d. Simulate the Bloch equations (10.130) and compare with the numerical solution.
282 CHAPTER 10. THE BLOCH EQUATIONS
Chapter 11
283
284 CHAPTER 11. ATOMS IN QUANTIZED RADIATION FIELDS
teoria eletrodinâmica 1 . No espaço livre, onde não há cargas nem correntes, e dentro do calibre
de Coulomb temos a solução da equação de onda (??) generalizada para uma distribuição de
vetores de onda k 2 ,
X
A(r, t) = ~k [A+
0k e
i(k·r−ωk t)
+ A−
0k e
−i(k·r−ωk t)
], (11.1)
k
onde já isolamos o caráter vetorial devido à polarização ~k do modo k da luz. Obviamente,
A− + ∗
0k = (A0k ) . Como cada amplitude e a polarização da onda do potencial vetorial A0k e A0k
∗
Podemos calcular o médio sobre um perı́odo da energia do modo k-th na cavidade por uma
versão quantizada da Eq. (11.9),
Z
ε0
Ēk = 2 hnk |Êk · Êk |nk idV . (11.9)
O resultado (11.6) é exatamente o que Planck havia sugerido (embora estritamente falando,
sua sugestão foi a quantização dos osciladores nas paredes condutantes da cavidade, e não
do campo), para explicar a distribuição da intensidade espectral radiada por um corpo negro.
Vemos agora que segue naturalmente da quantização dos modos de campo na cavidade. Resolva
o Exc. 11.8.1.1.
1
Vide a apostila Eletrodinâmica do mesmo autor [55].
2
A interação átomo-luz pode depender da polarização da luz à respeito do eixo de quantização do átomo
definido, e.g. por um campo magnético. Nestes casos precisamos estender o ı́ndice k para incluir o estado de
polarização (k, λ).
11.1. QUANTIZATION OF THE ELECTROMAGNETIC FIELD 285
Ĥf ield é o hamiltoniano do campo quantizado, expresso pela Eq. (11.6), e Ĥatom−f ield o a in-
teração átomo-campo. Para o hamiltoniano sem interação, Ĥ = Ĥatom + Ĥf ield , os auto-estados
são simplesmente estados de produto dos estados atômicos e dos estados de número de fótons,
O lado esquerdo da Fig. 11.1 mostra como as auto-energias dos estados produtos consistem de
duas escadas deslocadas pela energia de dessintonização ~∆. Escrevemos o operador hamilto-
niano do átomo Eq. (11.11) como soma de projetores sobre os auto-estados não perturbados
utilizando a relação de completeza e a ortogonalidade dos auto-estados. Com a mesma ideia
podemos reescrever o operador dipolar definido na Eq. (??),
X X X
d̂ = |iihi|d| |jihj| = dij |iihj| . (11.13)
i j i,j
Figure 11.1: Esquerda: Estados de número de fótons e os dois estados estacionários do átomo de
dois nı́veis. Meio: Dobre escada mostrando a base de estados produtos de estados de número
e estados atômicos. Direita: Estados vestidos construı́dos por diagonalização do hamiltoniano
completo na base de estados produtos.
Agora vamos usar as Eqs. (11.8) juntos para descrever a interação átomo-campo através do
hamiltoniano Ĥatom−f ield = −d̂ · Ê,
r h i
XX ~ωk
Ĥatom−f ield = i dij · ~k âk ei(k·r−ωk t) − â†k e−i(k·r−ωk t) |iihj| . (11.14)
2ε0 V
k i,j
286 CHAPTER 11. ATOMS IN QUANTIZED RADIATION FIELDS
Para o nosso átomo de dois nı́veis interagindo com um único modo do campo apenas temos,
r h i
~ωk
Ĥatom−f ield = i dge · ~k âk ei(k·r−ωk t) − â†k e−i(k·r−ωk t) (|gihe| + |eihg|) . (11.15)
2ε0 V
Ĥatom−f ield = 2i ~Ω(r)σ̂ + âk e−iωk t − 2i ~Ω∗ (r)σ̂ − â†k eiωk t . (11.17)
|ψi = N1 |g, n − 1ie−iωg t + N2 |g, nie−iωg t + N3 |e, n − 1ie−iωe t + N4 |e, nie−iωe t , (11.18)
P √ P √
utilizando as notações ↠= n n + 1|n + 1ihn| e â = n n|n − 1ihn| e a ortogonalidade
hj, m|i, ni = δij δmn , vemos que (considerando por simplicidade um átomo localizado na origem),
onde ω0 ≡ ωe − ωg . Vemos, que negligenciar o primeiro e quarto processo (ou seja, os ter-
mos ∝ σ̂ ± â± do hamiltoniano) é equivalente a fazer a aproximação da onda rotatória (RWA),
onde desprezamos os termos girando com a frequência ωk + ω0 , e que realmente só precisamos
considerar o acoplamento entre os dois estados vestidos |g, ni e |e, n − 1i.
É importante ressaltar, que o primeiro e quarto termo podem ser usados para alguns estados
intermediários em processos de ordem superior, tais como absorção multifotônica ou processos de
espalhamento Raman. De fato, quando a frequência de Rabi é muito grande, Ω ' ω, os processos
de excitação e deexcitação se sigam tão rapidamente, que a conservação de energia pode ser
violada para tempos curtos. Deslocamentos energéticos devido às contribuições desprezadas na
RWA se chamam deslocamento de Bloch-Siegert 3 .
3
O deslocamento não é observado, quando os termos não-rotativos σ ± a± são proibidos por outras regras de
conservação ou de seleção. Por exemplo, quando uma ressonância é excitada por luz σ ± , a RWA fica exata.
11.2. QUANTUM CORRELATIONS IN THE JAYNES-CUMMINGS MODEL 287
Figure 11.2: Quatro termos na interação átomo-campo. Os termos (b) e (c) conservam a energia
em processos de primeira ordem, enquanto (a) e (d) não conservam.
(11.24)
P √ 1 0 P 0 1
â = n n|n − 1i 0 1 hn| e σ̂ − = n |ni 0 0 hn| .
À partir da transformação Ĥn Un = Un Ên , sob a condição que Un é unitário e hermitiano, Un† Un ,
√
e usando a abreviação tan 2φn ≡ nΩ/∆, obtemos:
cos φn sin φn
Un = . (11.29)
− sin φn cos φn
4
Valem as seguintes regras para determinantes
det(AB) = det A det B e (det A)−1 = det A−1 .
11.2. QUANTUM CORRELATIONS IN THE JAYNES-CUMMINGS MODEL 289
Os autovalores correspondentes aos autovalores Ên são obtidos por Ĥn |xi = enx |xi, ou seja,
Ên Un−1 |xi = enx Un−1 |xi. Conhecendo Ên |yi = eny |yi, obtemos |xi = Un |yi. A evolução temporal
do estado de Jaynes-Cummings, |ψ(t)i = e−iĤt |ψ(0)i, é descrito pela transformação,
Essa situação, como ilustrada na PFig.n 11.3, descreve bem os estado de um laser aproximado
2
por um estado coerente, |αi = n √αn! |nie−|α| /2 . Para n → ∞, a incerteza da distribuição
√
poissoniana é pequena, ∆n/n̄ = 1/ n → 0, tal que o modo luminoso é caracterizado pelo
número médio de fótons e flutuações são desprezı́veis. Isso nós permite substituir a distribuição
poissoniana, Pn = δnn̄ ,
ω0
(n̄ − 1)ω 0 2 0 0 $2n̄
Ĥsemi = Ĥlaser + Ĥatom + Ĥint = + + $n̄∗ . (11.34)
0 n̄ω 0 − ω20 2 0
Example √ 43 (Resonant π/2 pulse): Neste exemplo, consideramos pulsos π/2 ressonantes,
isto é, n̄Ωt = 12 π. A evolução de Jaynes-Cummings agora simplifica para,
1 i
e−iĤn̄ t = 21 e−i(n̄−1/2)ωt . (11.36)
i 1
290 CHAPTER 11. ATOMS IN QUANTIZED RADIATION FIELDS
Para grandes n̄, um pulso ressonante π/2 faz (suprimindo fases dinâmicas irrelevantes),
|2i|n̄ − 1i π/2 (|2i|n̄ − 1i + i|1i|n̄i)
y , (11.37)
|1i|n̄i (i|2i|n̄ − 1i + |1i|n̄i)
ou seja, para um campo coerente,
|2i|αi π/2 (|2i + i|1i)|αi
y . (11.38)
|1i|αi (i|2i + |1i)|αi
Obviamente, a estrutura do campo |αi não é afetada e recuperamos a dinâmica de um átomo
de dois nı́veis excitado por uma radiação clássica ressonante descrita pelas equações de Bloch
(10.50).
Figure 11.3: Esquema dos nı́veis atômicos para implementação de interações ressonantes com
campos clássicos (na transição inferior) e dispersivas com campos quânticos (na transição supe-
rior).
(1) (2)
0 X hn|Ĥn(1) |mihm|Ĥn(1) |ni nΩ2
hψn |Ĥn |ψn i = hn|Ĥn + Ĥn |ni + (0) (0)
=∓ , (11.40)
En − Em 4∆
n6=m
que é um resultado já obtido nos Excs. 5.5.1.4 e 10.6.3.12. Em notação matricial,
2
(1) nΩ /4∆ 0
Ĥn ' . (11.41)
0 −nΩ2 /4∆
O operador de propagação temporal (11.30) então simplifica para,
!
(1) e inΩ2 t/4∆ 0
e−iĤn t = 2 . (11.42)
0 e−inΩ t/4∆
O fato, que os estados atômicos fundamental e excitado evolvem com diferentes fatores de fase
é importante, como mostraremos no seguinte exemplo.
11.2. QUANTUM CORRELATIONS IN THE JAYNES-CUMMINGS MODEL 291
O fato, que o deslocamento de fase nϕ depende do número de fótons, e que ele vai em
sentidos opostos para os estados fundamental e excitado é interessante. Já estudamos no
Exc. 10.6.5.5, que a interação dispersiva do átomo com um campo de luz pode causar um
deslocamento de fase para o vetor de Bloch. Agora, percebemos que além disso, provoca um
deslocamento de fase da amplitude de probabilidade de ter n fótons no campo da luz por
um valor proporcional à n, ou seja (suprimindo fase dinâmicas irrelevantes),
inϕ
|2i|n − 1i nϕ e |2i|n − 1i
y . (11.44)
|1i|ni e−inϕ |1i|ni
Obviamente, a fase do campo luminoso é deslocada por um valor ±ϕ, que depende da
população do átomo.
Notamos aqui, que essa dinâmica estudada no último exemplo fornece um método de trans-
ferir coerência de uma superposição atômica para um correlação quântica de um campo de luz.
Basta colocar o átomo inicialmente numa superposição de estados |1i+|2i, e o campo vai evoluir
para um estado de gato de Schrödinger |αeiϕ i+|αe−iϕ i. A transferência de correlações quânticas
entre graus de liberdade acoplados é uma das caracterı́sticas do modelo de Jaynes-Cummings.
Como exemplos estudaremos o fenômeno do colapso e renascimento quântico no Exc. 11.8.2.2 e
o desdobramento de Rabi no vácuo no 11.8.2.3.
Para determinar o estado interno do átomo, devemos tracejar sobre o campo luminoso. Como
processos de dissipação são desprezados, temos um estado puro descrito por, ρ̂ = |ψihψ|. As
populações e coerências são, portanto,
X X
ρij = hi|Trf ield ρ̂|ji = hi| hn|ψihψ|ni|ji = ci,n c∗j,n . (11.51)
n n
Com isso, podemos calcular o vetor de Bloch atômico, cuja norma interessantemente NÃO é
preservada, pois,
√
2Re ρ 12
√
|~
ρ| =
2Im ρ12
2
= 2|ρ12 | − 2ρ11 ρ22 = −2 det ρ̂ (11.53)
ρ22 − ρ11
X X X X
=2 c1,n c∗2,n c∗1,n c2,n − 2 |c2,n |2 |c1,n |2 6= 1 .
n n n n
Para determinar o estado do campo luminoso, devemos tracejar sobre o estado atômico. Por
exemplo, a amplitude de probabilidade de encontrar o estado |ψi em |ni é,
tal que,
X
pn = hn|Tratom ρ̂|ni = hn| hi|ψihψ|ii|ni = |hn|ψi|2 = |c1,n |2 + |c2,n |2 . (11.55)
i=1,2
Para caracterizar o campo óptico separadamente do estado atômico podemos tentar, por
um cálculo similar à (11.49), projetar o estado de Jaynes-Cummings sobre uma base de estados
coerentes. Assim, a amplitude de probabilidade de encontrar o estado |ψi em |αi é,
X α∗n
2 /2
hα|ψi = e−|α| √ (c1,n |1i + c2,n |2i) (11.56)
n n!
2
X ∗n
α αm
|hα|ψi|2 = e−|α| √ √ (c∗1,m c1,n + c∗2,m c2,n ) ,
n n! m!
11.2. QUANTUM CORRELATIONS IN THE JAYNES-CUMMINGS MODEL 293
tal que,
2 2
X n
α X α
n
−|α|2
πQ(α) ≡ hα|Tratom ρ̂|αi = e c1,n √ + c2,n √ . (11.57)
n! n!
n n
Derivaremos este resultado no Exc. 11.8.2.4. Essa grandeza, que se chama função Q, permite a
ilustração do estado num plano Re α-Im α [31]. Ela geralmente é fácil de calcular, mas não exibe
muita informação, e.g. sobre fenômenos de interferências produzidos por correlações quânticas.
Na seção seguinte, vamos definir outras funções de distribuição, como a função de Wigner, que
também pode ser avaliada à partir dos coeficientes de Jaynes-Cummings [81].
0.4
0.1
1 1
12
0.2
pn
Reρ
11
ρ -ρ
0 0.5
22
0
1 0.05 10
-1 0
-0.2
-1 0 -10 0
0 Imρ -0.4 0 Imα
12 0
Reρ 1 -1 10 20 30 40 0 20 40 Reα 10 -10
12
t n
Assim, a função Q corresponde à uma função de Wigner suavizada que, por sua vez, corresponde
à uma função P suavizada. A transformação de Fourier complexa inversa dá imediatamente 5 ,
2 2 2 2 2 −2|λ|2 1 2
Q = W ? e−2|λ| = P ? e−2|λ| ? e = P ? e−|λ| . (11.59)
π π π π
Os estados são expressos por funções de peso bidimensionais P, Q, W . A seguir, damos alguns
exemplos para essas representações.
5 1
exp(−a|λ|2 + bλ + cλ∗ )d2 λ = 1
exp( bc
R
Usando a formula integral, π a a
).
11.2. QUANTUM CORRELATIONS IN THE JAYNES-CUMMINGS MODEL 295
ρ̂ = |nihn| . (11.62)
Pk = δnk
Este estado pode ser expandido em estado coerente |αi pelo seguinte procedimento. Para grandes
n calculamos primeiramente a função de distribuição coerente P ,
√
P|ni (α) = δ (1) (|α| − n) , (11.63)
pois ela nós permite derivar a matriz densidade pela formula (11.60),
Z
√
ρ̂|ni = δ (1) (|α| − n)|αihα|d2 α (11.64)
Z ∞ Z 2π Z 2π
√
= δ (1) (|α| − n)|αihα| |α|2 d|α|dϕα = n |αihα|dϕα .
0 0 0
X αn
|αi = e−|α|/2 √ |ni
n n!
X |α|n e−|α|
ρ̂ = |nihn| . (11.67)
n
n!
2 |α|2n
Pn = |hn|αi|2 = e−|α|
n!
a matriz densidade, Z
ρ̂|βi = δ (2) (α − β)|αihα|d2 α = |βihβ| . (11.69)
1 −|α−β|2
Q|βi (α) = e , (11.70)
π
e a função de Wigner,
2 −2|α−β|2
W|βi (α) = e . (11.71)
π
Fazendo a média, este estado pode ser escrito como uma superposição de estados de Fock,
X
ρ̂ = Pn |nihn| , (11.73)
n
X n̄n
ρ̂therm = |nihn| . (11.75)
n
(1 + n̄)1+n
1 −|α|2 /n̄
Ptherm (α) = e , (11.76)
πn̄
a matriz densidade, Z
1 2 /n̄
ρ̂ = e−|α| |αihα|d2 α , (11.77)
πn̄
a função de distribuição coerente Q,
1 2
Qtherm (α) = e−|α| /(n̄+1) . (11.78)
π(n̄ + 1)
e a função de Wigner,
1 2
Wtherm (α) = e−|α| /(n̄+1/2) . (11.79)
π(n̄ + 1/2)
11.2. QUANTUM CORRELATIONS IN THE JAYNES-CUMMINGS MODEL 297
a matriz densidade,
ρ̂ = , (11.82)
a função de distribuição coerente Q,
2 2
Q|β0 i|β1 i (α) = π1 e−|α−β0 | + π1 e−|α−β1 | . (11.83)
e a função de Wigner,
Z
∗ α−λa∗
W (α) = 1
π χS (λ)eλ d2 λ (11.85)
Z
2 /2 ∗ +(−α+β )λ ∗ −β ∗ )(α−β )
= 1
π e−|λ| e(α−β1 )λ 0
d2 λ = π2 e(α 1 0
.
e 2 2 ∗ ∗ −β ∗ )(α−β
e−2|α−β0 | + e−2|α−β1 | ± 2Re e2β1 β0 e−2(α 1 0
W (α) = ∗ . (11.87)
π 1 ± Re e2β1 β0
No nı́vel microscópico, temos o exemplo dos graus de liberdade internos de átomos, que
podem estar em estados de superposição |hi|ji| = δij . Por outro lado, os estados de gato
de Schrödinger são estados em campos contı́nuos de Schrödinger com superposições quânticas
mesoscópicas |hα|βi| = e−|α−β| . Também temos que enfatizar a diferença fundamental entre os
gatos de Schrödinger e os estados de superposição de modos,
|ψi = |αi + |βi =6 |αi|βi
ρ = |αihα| + |βihβ| + |αihβ| + |βihα| 6 |αβihαβ|
= . (11.88)
2 2 2
W (α) = π1 (e−2|α−α0 | + e−2|β−β0 | + termos de interferência 6= π1 e−2|αβ−α0 β0 |
Os gatos de Schrödinger exibem interferências no espaço de fase, enquanto que para super-
posições de modos, interferências aparecem apenas quando se varia um parâmetro (por exemplo,
o comprimento de um braço do interferômetro).
Interferências quânticas macroscópicas (isto é, interferências que são detectáveis com apar-
elhos macroscópicos, por exemplo, em esquemas heteródinos) são nomeadas gatos (fuzzy) de
Schrödinger, se os estados de interferentes estiverem visivelmente separados no espaço de fase.
298 CHAPTER 11. ATOMS IN QUANTIZED RADIATION FIELDS
Eles são muito sensı́veis à dissipação e são facilmente convertidos em misturas estatı́sticas.
Por exemplo, |αi ± | − αi contém apenas números de fótons ı́mpares (par) na função de dis-
tribuição Pn . Depois de algum tempo ∼ τcav /N , a distribuição é convertida em uma distribuição
poissoniana. Quanto maior N , mais rápida será a decoerência e, portanto, os estados do gato
de verdade nunca serão observados.
Seja,
|ψi = √1 (|1i|βi + i|2i|γi) . (11.89)
2
Comparando com os coeficientes de JC segue,
X√ X√
β = 21 nc∗1,n−1 c1,n e 1
2 nc∗2,n−1 c2,n . (11.90)
n n
As funções Q, P e W para gatos de Schrödinger usuais são conhecidas. Agora, só inserir β e γ.
Example 46 (Função de Wigner para dinâmica de Jaynes-Cummings): Outras
funções de distribuição costumam ser usadas para caracterizar o estado de um campo de
Heisenberg. A função de Wigner, por exemplo, é definida como [81],
Z
∗ ∗
W (α) = π12 χS (λ)eαλ −α λ d2 λ onde d2 λ = dRe λ dIm λ .
Função caracterı́stica,
X † ∗ † ∗
χn (λ) = c∗1,m c1n hm|eλâ e−λ â |ni + c∗2,m−1 c2,n−1 hm − 1|eλâ e−λ â |n − 1i
m,n
r
X n! m−n m−n
= (c∗1,m c1n + c∗2,m c2,n ) λ Ln (|λ|2 ) .
m,n
m!
Função caracterı́stica,
X
χS (λ) = (c∗1,m c1n + c∗2,m c2,n )unm (λ) .
m,n
Devemos distinguir os casos, porque unm (λ) = umn (−λ∗ ) = (−1)m−n umn (λ∗ ),
X X
χS (λ) = (c∗1,m c1n + c∗2,m c2,n )unm (λ) + (c∗1,m c1n + c∗2,m c2,n )unm (−λ∗ )
m≥n m<n
r
X n! −|λ|2 /2 m−n
2 −|λ|2 /2
= (|c10 | + |c20 |)e + e Ln (|λ|2 )
m>n
m!
∗
(c1,m c1n + c∗2,m c2,n )λm−n + (c1,m c∗1n + c2,m c∗2,n )(−λ∗ )m−n .
e
Z r Z
n! 2 x2λ ipα xλ
e +ipα xλ
unm ( x2λ )dxλ = e−xλ /2 xm−n
λ Lm−n
n ( 2 )e dxλ
m!
r r
n! π 1 m−n 2
= (−1)int 2 e−pα /2 Hen (pα )Hem (pα ) .
m! 2 n!
Função de Wigner,
r r
∗ π −1
1 m−n
W (α) = 1
π 2 (c1,m c1n
+ c∗2,m c2,n ) (−1)int 2
α 2 n!m!
h i
−p2α /2 −x2α /2
e Hen (pα )Hem (pα ) + e Hen (xα )Hem (xα ) .
Ĥef f = ω↠â + ω0 σ̂ + σ̂ − 1
2 + Ω
2 âσ̂ + + ↠σ̂ − + iΓ + −
2 σ̂ σ̂ . (11.91)
300 CHAPTER 11. ATOMS IN QUANTIZED RADIATION FIELDS
Ou em notação matricial,
1
− 2 ω0 0
0 1 i 1
2 0 − 2Γ
ω 2Ω
1
ω − 12 ω0
2Ω 0 √ 1
1 i
Hef f = 0 ω + 2 0 − 2Γ
√ ω 22Ω .
1
22Ω 2ω − 12 ω0 0
1 i
0 2ω + 2 0 − 2Γ
ω
..
.
(11.92)
O fluxograma da simulação é,
|1i(h1|+h2|)|ψi |ψi
|ψi → |ψi →
P |(h1|+h2|)|ψi| P ||ψi|
|1i n (c1n |ni+c2n−1 |n−1i) c1n |1i|ni+c2n−1 |2i|n−1i
= √P 2
= n√
P 2 2
n |c1n +c2n | | n (|c1n | +|c2n | )
projeção & evolução dinâmica . renormalização
|ψ(t + dt)i = e−iHef f dt |ψi
- ↓ %
?
sim ζ > hψ(t + dt). |ψ(t + dt)i no
variável aleatória
Absorção ou espalhamento de luz causa o decaimento do campo óptico como α(t) ∝ e−κt/2 .
A projeção (em notação por componentes) é implementada por,
1
c0jn ≡ pP cjn δj1 , (11.93)
2
n n(|c1n + c2n | )
a renormalização por,
1
c0jn ≡ pP cjn , (11.94)
2 2
n (|c1n | + |c2n | )
e a evolução dinâmica por c0jn ≡ e−κnt/2 cjn . Notamos, que a dissipação por perdas da cavidade
também pode ser tomada em conta pela equação mestre. Faz o Exc. 13.7.2.1.
Vamos ver, que com esse hamiltoniano podemos derivar, num cálculo é conhecido como teoria
de Weisskopf-Wigner, a equação de Schrödinger para as amplitudes dos nı́veis atômicos (10.69)
inclusive emissão espontânea.
Aqui, Ω0 é a frequência de Rabi da interação entre o átomo e o modo da bomba (que é tratado
como campo clássico), σ̂ − é o operador
p de abaixamento da excitação atômica, âk é o operador de
aniquilação de um fóton, e gk = d ω/(~ε0 V ) descreve o acoplamento entre o átomo e um modo
do vácuo com o volume V . O átomo tem dois estados, o estado fundamental |gi e estado excitado
|ei. Como estamos considerando apenas um átomo fixo no espaço 6 , podemos colocar-lhe na
posição r = 0. Além disso, considerando um laser incidente com alta potência,
√
âk0 |n0 ik0 = n|n0 − 1ik0 ' |n0 ik0 , (11.100)
âk0 é aproximadamente uma observável proporcional à raiz da intensidade. Como [âk0 , â†k0 ] ' 0,
√
podemos desconsiderar a natureza quântica e substituir, Ω0 ≡ 2 n0 gk0 . Com a aproximação da
onda rotatória (rotating wave approximation, RWA) o hamitoniano fica
h i Xh i
Ĥ = ~2 Ω0 σ̂ − â†k0 ei∆0 t + h.c. + ~ gk σ̂â†k ei∆k t + h.c. , (11.101)
k
introduzindo as abreviações
∆0 ≡ ω0 − ωa e ∆k ≡ ωk − ωa . (11.102)
d Ω0 X Z t 0
β(t) = −i α(t) − gk2 ei(ωk −ωa )(t −t) β(t0 )dt0 (11.109)
dt 2 0
k
Ω0 X Z t 00
= −i α(t) − gk2 e−i(ωk −ωa )t β(t − t00 )dt00 .
2 0
k
Rt 0
Usando a aproximação de Markov β(t − t00 ) ' β(t), com lim 0 e−i(ωk −ωa )t dt0 = πδ(ωk − ωa ), e
P R 3 t→∞
V
substituindo k −→ (2π) 3 d k, chegamos a,
d Ω0 X
β(t) ' −i α(t) − gk2 β(t)πδ(ωk − ωa ) (11.110)
dt 2
k
Z
Ω0 V
= −i α(t) − β(t) gk2 πδ(ωk − ωa )d3 k
2 (2π)3
Ω0 V 1 Ω0 Γ
= −i α(t) − 3
β(t)4πgk2 a πka2 = −i α(t) − β(t) .
2 (2π) c 2 2
No último passo introduzimos, como abreviação, a taxa de emissão espontânea,
V 2 2
Γ≡ k g , (11.111)
πc a ka
Finalmente,
d Ω0 d Ω0 Γ
α(t) = −i β(t) e β(t) = −i α(t) − β(t) . (11.112)
dt 2 dt 2 2
Esses são exatamente as equações das amplitudes de probabilidade (10.69) derivadas da
equação de Schrödinger, mas agora com o termo de emissão espontânea sendo explicitamente
derivado.
Em muitas situações na óptica quântica NÃO podemos desprezar o impacto do recuo fotônico
e omitir o movimento do centro de massa atômico [primeiro termo do hamiltoniano (11.113)].
Por isso, estenderemos no Cap. 12.3.2 a discussão para incluir o movimento atômico e as forças
da luz sobre átomos.
Radiação pode ser absorvida ou espalhada por um átomo em diferente maneiras, dependendo
se a interação for espalhamento elástico ou espalhamento inelástico, um processo coerente ou in-
coerente, espontâneo or (bosonicamente) estimulado. Estas propriedades são caracterı́sticas
para muitos processes, em particular, a fluorescência ressonante (isto é, absorção e emissão),
espalhamento de Rayleigh ou espalhamento de Raman. No seguinte, vamos esclarecer esta clas-
sificação.
Cada processo de espalhamento é espontâneo ou estimulado 9 . Processos espontâneos po-
dem ser entendidos como sendo estimulados flutuações do vácuo. Espalhamento de Rayleigh é
elástico, isto é, a energia cinética do átomo espalhador é a mesma antes e depois do processo
de espalhamento. Em contraste, o espalhamento de Raman é inelástico. Emissão espontânea
é devido ao decaimento da população de um estado excitado, o espalhamento espontâneo de
Rayleigh é devido ao decaimento de um momento dipolar induzido. Nos dois casos, o processo é
estimulado por flutuação do vácuo: Na emissão espontânea um autoestado de excitação interna
decai devido ao acoplamento à um reservatório eletromagnético. No espalhamento espontâneo
de Rayleigh a superposição coerente decai.
9
Teorias clássicas do espalhamento de luz através de excitação do movimento eletrônico dentro dos modelos
de Lorentz ou de Drude podem ser consultada nas Secs. apostila Eletrodinâmica do mesmo autor [55]. . Nestes
modelos entendemos muitos aspectos do espalhamento de Compton, de Thomson e de Rayleigh.
11.3. SPONTANEOUS EMISSION AND LIGHT SCATTERING 305
Por isso, o campo elétrico emitido por um átomo e a intensidade da luz espalhada são dados
por,
eω02 ˆ · r12
hÊ+
s (r, t)i = − ρ̃21 (t − rc )e−iω(t−r/c)
4πε0 c2 r
, (11.115)
4
α~ω0 |ˆ · r12 |2
I¯s = cε0 hÊ− +
s (r, t)Ês (r, t)i = ρ22 (t − rc )
4πc2 r2
com a definição da constante de Sommerfeld α = e2 /4πε0 ~c. Calculamos o fluxo total de fótons
emitidos,
Z ¯ 2 Z
(sp) Is r 1 α~ω04 |r12 |2 cos2 θ
Wf i = dΩ = ρ22 (t − rc ) sin θdθdφ (11.116)
~ω0 ~ω0 4πc2
8π α~ω04 |r12 |2 2α
= 2
ρ22 (t − rc ) = 2 ω03 |r12 |2 ρ22 (t − rc ) .
3~ω0 4πc 3c
O resultado coincide com taxa de emissão espontânea Γ calculada em (??).
A seção eficaz diferencial de espalhamento pode ser definida por,
dσ ω I¯s r2
≡ . (11.117)
dΩ ωs I¯0
I¯scoh hÊ− +
s (r, t)ihÊs (r, t)i
= . (11.118)
I¯s hÊ− +
s (r, t)Ês (r, t)i
s/2 s2 /2
Scoh = |ρ21 (∞)|2 = e Sincoh = ρ22 (∞) − |ρ21 (∞)|2 = . (11.120)
(1 + s)2 (1 + s)2
Figure 11.8: (Code: LM Quantumf ields InElastic.m) Espalhamento elástico versus inelástico.
hÊ − (r1 , t1 )..Ê − (rn , tn )Ê + (rn+1 , tn+1 )..Ê + (r2n , t2n )i
g (n) (r1 , t1 , .., r2n , t2n ) ≡ q . (11.122)
− + − +
hÊ (r1 , t1 )Ê (r1 , t1 )i..hÊ (r2n , t2n )Ê (r2n , t2n )i
onde, Z t
1
h· · ·it = lim · · · dt (11.124)
t→∞ t 0
denota a média temporal. Definindo a intensidade como Iˆ = 2ε0 cÊ + Ê − , as coerências ficam,
2
g (1) (τ ) ≡ 2εI0 c hT N Ê − (t)Ê + (t + τ )i e ˆ I(t
g (2) (τ ) ≡ 2εI0 c hT N I(t) ˆ + τ )i . (11.125)
As coerências devem ser calculadas à partir dos operadores de campo respeitando a ordem tem-
poral e a ordem normal. Elas são grandezas úteis para descrever fenômenos como o bunching
de fótons ou entender os espectros de fluorescência ou o espalhamento de luz por átomos cor-
relacionados. g (1) mede a coerência de um campo de luz (quanto ele se assemelha à uma onda
senoidal). g (2) mede, para um dado grau de coerência, o desvio do campo de luz do estado
quântico que mais se aproxima à luz clássica (quanto ele se assemelha à um laser).
As funções de correlação g (1) e g (2) são experimentalmente medidas em experimentos de
Young e Hanbury-Brown-Twiss. Eles são ilustrados nas Figs. 11.9.
Coerência e caos são propriedades contrárias da luz, que podem ser quantificadas pelo es-
pectro da luz ou a função de autocorrelação. A função de autocorrelação de primeira ordem
descreve flutuações de fase e amplitude, a função de autocorrelação de segunda ordem descreve
flutuações de intensidade. O perfil de emissão de uma fonte de luz emerge geralmente como
uma combinação de vários efeitos fı́sicos: O acoplamento a um banho térmico dá origem a
P −~ω(n+1/2)/k
uma distribuição térmica da energia de radiação, Pn = e −~ω(n+1/2)/k B T / ne BT .
Numa fonte de luz, o meio ativo exibe ressonâncias e alargamentos, e o ressonador imprime uma
estrutura modal.
11.3. SPONTANEOUS EMISSION AND LIGHT SCATTERING 307
Figure 11.9: (a) Esquema do experimento de Young. (b) Esquema do experimento de Hanbury,
Brown e Twiss. O experimento de Young revela a coerência de um campo, isto é, sua capacidade
de interferir. Em contraste, o experimento de Hanbury-Brown-Twist indica as correlações entre
as (quase-)partı́culas, que compõem o campo, isto é, efeitos devido às estatı́sticas quânticas ou
interações.
laser (11.126)
E = eRiωt
e−iωτ dt
=⇒ g (1) (τ ) = R
dt
= e−iωτ
(1) (11.127)
=⇒ F g (τ ) = δ(∆)
g (2) (τ ) = 1
Vemos, que os valores absolutos das coerências de primeira e segunda ordem são constantes, e
que o espectro é fino como uma função δ. Par um laser sujeito à ruı́do branco de fase (ζ seja
um is a número aleatório com distribuição normal) temos,
Agora, a coerência de primeira ordem decai exponencialmente, |g (1) (τ )| = e−γτ , tal que o espec-
tro tem um perfil lorentziano. Este resultado já foi derivado na Sec. ?? para a largura natural
de uma transição sujeito à emissão espontânea. Entendemos a conexão pela interpretação da
emissão espontânea como sendo induzida por flutuações do vácuo, que tem um espectro de ruı́do
branco.
Na Sec. 10.4.2 já vimos, que o alargamento por colisões pode ser tratado assumindo, que a
luz é emitida como uma superposição de ondas coerentes tendo a mesma frequência, En (t) =
308 CHAPTER 11. ATOMS IN QUANTIZED RADIATION FIELDS
Termos cruzados zeram. O alargamento por pressão é homogêneo, mas o fato que os pacotes
de onda são espalhados por átomos diferentes se traduz numa coerência de segunda ordem
modificada,
XZ Z ∞
∗ iωτ iφn (t+τ )−iφn (t) iωτ
hEn (t)En (t + τ )i = e e dt = e p(τ )dτ . (11.131)
n τ
2 2
1.5 1.5
g (1)
g (2)
1 1
0 0
-10 0 10 -10 0 10
τ τ
Luz mono-modo caótica pode ser vista como luz multi-modo incoerente, onde todos os modos
exceto um único são filtrados por um etalon. Essa luz é caracterizada por |g (1) (τ )| = 1 e
g (2) (τ ) = 2, apesar do comprimento de coerência sendo τ → ∞.
Dois modos sem flutuações interferindo satisfazem |g (1) (τ )| = 1, enquanto dois modos sem
flutuações satisfazem |g (1) (τ )| = cos 21 (ω1 − ω2 )τ .
é a função de autocorrelação e
Re (τ ) Se (ω)
g (1) (τ ) = e F (ω) = Fg (1) (ω) = . (11.141)
Re (0) Re (0)
A última grandeza é o perfil da linha. Podemos ver que, para calcular o espectro da fluorescência
ressonante, só precisamos calcular a função de correlação g (1) , isto é, as amplitudes do campo
310 CHAPTER 11. ATOMS IN QUANTIZED RADIATION FIELDS
Ê(t) que, por sua vez são relacionados aos operadores de campo (11.8). Os operadores de campo
seguem as soluções das equação de Bloch, que sendo linear, têm a seguinte forma geral,
X
ρij (t + τ ) = αijkl (τ )ρij (t) + βij (τ ) . (11.142)
k,l
Sabemos,
hAµ (t)Fν (t)i = hDµν i e hFµ (t)Aν (t)i = hDµν i (11.145)
e o teorema da regressão quântica dá,
d
hAµ (t)Aν (t0 )i = hDµ (t)Aν (t0 )i , (11.146)
dt
pois se t0 < t o termo hFµ (t)Aν (t0 )i zera para um processo Markoviano.
F (ν) = (Fg (1) )(ν) = F[e−iωτ ] ? F[G(τ )] = δ(ν − ω) ? F[G(τ )] = (FG)(ν − ω) . (11.147)
Portanto, podemos olhar para o espectro não deslocado, (FG)(ν). Como o espectro de flu-
orescência é determinado pela coerência de primeira ordem, que depende dos operadores de
campo que, em torno, dependem das populações e coerências atômicas, temos que resolver a
equação de Bloch.
Para um átomo de dois nı́veis as equações de Bloch reduzidas pela condição de normalização
(10.135) são,
−2γ − 2i Ω i
2Ω ρ22 0
~˙ = A~
ρ ρ + b = −iΩ −γ − i∆ 0 ρ12 + i Ω
2 (11.148)
i
iΩ 0 −γ + i∆ ρ21 −2Ω
com a solução (10.137), ou seja, ρ(t + τ ) = eLτ ρ(t) + (1 − eLt )L−1 b. Essa solução pode ser
colocada na forma seguinte,
X
ρkl (t + τ ) = α(kl)(mn) (τ )ρmn (t) + β(kl) (τ ) , (11.149)
(mn)
11.3. SPONTANEOUS EMISSION AND LIGHT SCATTERING 311
dando,
X
hσ̂kl (t + τ )i = e(l−k)iω0 (t+τ ) e(m−n)iω0 t α(kl)(mn) (τ )hσ̂mn (t)i + e(l−k)iω0 (t+τ ) β(mn) (τ )h1i .
(mn)
(11.152)
Ê − = γ σ̂21 , (11.153)
onde γ é simplesmente uma constante. Substituindo essa relação nas funções de correlação
(11.123) obtemos,
hσ̂21 (t)σ̂12 (t + τ )i hσ̂21 (t)σ̂12 (t + τ )i
g (1) (τ ) = = (11.154)
hσ̂21 (t)σ̂12 (t)i hσ̂22 (t)i
hσ̂21 (t)σ̂21 (t + τ )σ̂12 (t + τ )σ̂12 (t)i hσ̂22 (t)σ̂22 (t + τ )i
g (2) (τ ) = = .
hσ̂21 (t)σ̂12 (t)i2 hσ̂22 (t)i2
ou seja,
[L−1 b](21)
g (1) (τ ) = eiω0 τ eLτ (12)(12)
+ (1 − eLτ )L−1 b
(12) [L−1 b] (22)
. (11.158)
[(1−eLτ )L−1 b](22)
g (2) (τ ) = eiω0 τ [L−1 b] (22)
Essas funções de correlação podem facilmente ser calculadas pela resolução numérica das equações
de Bloch (11.148). A Fig. 11.11 mostra as funções de correlação e o espectro de fluorescência
derivado por transformação de Fourier (11.141).
(a) 1 (b) 0.05 (c) 2 (d) 100
0.8 80
1.5
0
0.6 60
g (τ)
g (τ)
kk
I
1
(1)
(2)
ρ
0.4 40
-0.05
0.5
0.2 20
0 -0.1 0 0
0 5 10 0 5 10 0 5 10 -50 0 50
Γt Γt Γt Δ
Figure 11.11: (Code: LM Quantumf ields ResF luM ollowSpectrum.m) (a) Evolução temporal
da população do estado excitação. (b) Função de correlação g (1) (τ ) e (c) g (1) (τ ). (d) Espectro
de Mollow.
O espectro exibe três linhas no espectro conhecidas como o tripleto de Mollow. Note, que
o tripleto não é observado em emissão espontânea (precisamos da excitação por um laser).
A posição das linhas é facilmente entendido na imagem dos átomos vestidos: Enquanto, na
aproximação semi-clássica os autovalores são calculados com o hamiltoniano Ĥatom + Ĥlaser ,
Ĥint sendo tratado com perturbação, na imagem do átomo vestido eles são calculados por
Ĥatom + Ĥlaser + Ĥint . O que, na imagem semi-clássica é interpretado com transição de dois
fótons aparece, na imagem do átomo vestido como uma transição simples entre estados vestidos.
de Bloch de dois nı́veis na presença de colisões são dadas por (10.73), onde γ 0 = γ + γcoll é a
largura da linha alargada por colisões. Dentro deste modelo e no limite Ω Γ, o espectro de
fluorescência ressonante, incluindo espalhamento elástico e inelástico, é dado por [164],
γ0 − Γ γ 0 /π Γ
F (ωs ) = + δ(ωs − ω) .
γ 0 (ω0 − ωs )2 + γ 02 γ 0
e as funções de correlações,
h ih i
hÊ − (t)Ê + (t + τ )i h Êσ−− (t) + Êπ− (t) + Êσ−+ (t) Êσ+− (t + τ ) + Êπ+ (t + τ ) + Êσ++ (t + τ ) i
g (1) (τ ) ≡ = h ih i .
hÊ − (t)Ê + (t)i h Êσ−− (t) + Êπ− (t) + Êσ−+ (t) Êσ+− (t) + Êπ+ (t) + Êσ++ (t) i
Rotulando os estados,
chegamos a,
h[γ σ̂21 (t) + γ σ̂31 (t) + γ σ̂41 (t)] [γ σ̂12 (t + τ ) + γ σ̂13 (t + τ ) + γ σ̂14 (t + τ )]i
g (1) (τ ) =
h[γ σ̂21 (t) + γ σ̂31 (t) + γ σ̂41 (t)] [γ σ̂12 (t) + γ σ̂13 (t) + γ σ̂14 (t)]i
P
α (τ )h[σ̂21 (t) + σ̂31 (t) + σ̂41 (t)] σ̂12 (t)i + β(12) (τ )hσ̂21 (t)i
P(mn) (12)(mn)
+ (mn) α(13)(mn) (τ )h[σ̂21 (t) + σ̂31 (t) + σ̂41 (t)] σ̂13 (t)i + β(13) (τ )hσ̂31 (t)i
P
+ (mn) α(14)(mn) (τ )h[σ̂21 (t) + σ̂31 (t) + σ̂41 (t)] σ̂14 (t)i + β(14) (τ )hσ̂41 (t)i
=
h[σ̂21 (t) + σ̂31 (t) + σ̂41 (t)] [σ̂12 (t) + σ̂13 (t) + σ̂14 (t)]i
α(12)(12) (τ )hσ̂21 (t)σ̂12 (t)i + β(12) (τ )hσ̂21 (t)i
+α(13)(13) (τ )hσ̂31 (t)σ̂13 (t)i + β(13) (τ )hσ̂31 (t)i
+α(14)(14) (τ )hσ̂41 (t)σ̂14 (t)i + β(14) (τ )hσ̂41 (t)i
=
h[σ̂21 (t) + σ̂31 (t) + σ̂41 (t)] [σ̂12 (t) + σ̂13 (t) + σ̂14 (t)]i
α(12)(12) (τ )ρ22 (∞)i + β(12) (τ )ρ21 (∞) + α(13)(13) (τ )ρ33 (∞)
+β(13) (τ )ρ31 (∞) + α(14)(14) (τ )ρ44 (∞) + β(14) (τ )ρ41 (∞)
= .
ρ22 (∞) + ρ33 (∞) + ρ44 (∞) + ρ23 (∞) + ρ32 (∞) + ρ24 (∞) + ρ42 (∞) + ρ34 (∞) + ρ43 (∞)
† +↠b̂)
|ψ(t)i = eiΩt/2(âb̂ |ψ0 i . (11.161)
e as equações de Heisenberg,
com a solução,
â(t) cos 12 Ωt −i sin 21 Ωt â0
= . (11.165)
b̂(t) −i sin 21 Ωt cos 12 Ωt b̂0
p
Introduzindo a abreviação η ≡ cos(Ωt/2), podemos descrever a evolução como,
√ √
â(t) η −i 1 − η â0
= √ √ . (11.166)
b̂(t) −i 1 − η η b̂0
Percebemos 11 , que o feixe refletido na superfı́cie de um meio opticamente mais denso sofre
um deslocamento de fase de π/2.
11
Este fato é uma consequência da time-reversal invariance no divisor de feixe.
11.4. BEAM SPLITTING AND QUANTUM AMPLIFICATION 315
onde, √
0 1nb p
√
1nb
0 2(nb − 1)
p ..
Ĥa+b = ~2 Ω
2(nb − 1) . .
(11.169)
.. √
. na 1
√
na 1 0
Os sub-espaços com na +nb +1 fótons são completamente degenerados, pois det(λIa+b −Ha+b ) =
λna +nb +1 = 0. A degenerescência é removida, quando introduzimos mecanismos de perda para
um dos modos. Assim, o hamiltoniano pode ser entendido como um sistema de Dicke com
multiplicidade 1 (n\+ n ) = n + n + 1.
2 a b a b
satisfazem U −1 HU = E. Portanto,
XN
1 N 1
|ψi = √ + (â†1= â†2 )N |0, 0i (â†1 )n (â†2 )N −n |0, 0i
√ (11.170)
2 N/2 N! 2 N/2 N ! n=0 n
s s
N N
1 X N X N
= N/2 |n, N − ni = 0.5n 0.5N −n |n, N − ni
2 n n
n=0 n=0
N
r n
X n
(N/2) −N/2 X αn
' e |n, N − ni = e−|α|/2 √ |n, N − ni ,
n! n!
n=0 n=0
Além disso,
n N −1
1 X N 2 N X N N (N + 1)
hn̂21 i = n n = n (n + 1) = (11.173)
2 n 2 n 4
n=0 n=0
s
n
1 X N N (N − 1)
hn̂1 n̂2 i = n n(N − n) = N hn̂1 i − hn̂21 i =
2 n 4
n=0
Example 50 (Matriz densidade para o divisor de feixe com 0 ou 1 fóton): E.g. para
estados de superposição, |ψi = √12 (|0, 1i ± |1, 0i),
0 0 0
0 1
± 12
2
ρ̂ = 0 ± 12 1 .
2
..
.
11.4.3 Shot-noise
O limite de Heisenberg nas fases de quadratura de um campo de luz determina o ruı́do de shot-
noise na intensidade do feixe de luz. Para medir este ruı́do dividimos o feixe por uma separatriz
em dois feixes â e b̂ e recombinamos estes feixes numa segunda separatriz,
ĉ = √1 (â
2
+ b̂) , dˆ = √1 (â
2
− b̂) . (11.181)
Detectados por fotodetectores com o coeficiente de ganho g estes dois feixes produzem correntes,
Iˆ+ ≡ Iˆc + Iˆd = g(↠â + b̂† b̂) , Iˆ− ≡ Iˆc − Iˆd = g(↠b̂ + b̂† â) . (11.183)
Iˆ+
2
= g 2 [n̂2a + n̂2b + 2n̂a n̂b ] , Iˆ−
2
= g 2 [(↠b̂)2 + (b̂† â)2 + ↠â + b̂b̂† + â↠+ b̂† b̂] . (11.184)
ˆ 2 i ≡ hIˆ2 i − hIi
Agora, com a definição, h(∆I) ˆ 2 , obtemos o ruı́do de intensidade do campo,
h∆Iˆ+
2
i = h(∆Iˆ+ )2 i = g 2 h(∆n̂a )2 i , (11.187)
hâiout
G= . (11.190)
hâiin
The quantum amplifier can also be described in the Heisenberg picture; the changes are
attributed to the amplification of the field operator. Thus, the evolution of the operator âout is
given by
âout = Û † âÛ , (11.191)
while the state vector remains unchanged. The gain is then given by,
hâout iin
G= . (11.192)
hâiin
In general, the gain G may be complex, and it may depend on the initial state. For laser
applications, the amplification of coherent states is important. Therefore, it is usually assumed
that the initial state is a coherent state characterized by a complex-valued initial parameter α,
such that |ini = |αi. Even with such a restriction, the gain may depend on the amplitude or
phase of the initial field.
In the following, the Heisenberg representation is used; all brackets are assumed to be eval-
uated with respect to the initial coherent state,
noise = hâ†out âout i − hâ†out ihâout i − (h↠âi − h↠ihâi) . (11.193)
The expectation values are assumed to be evaluated with respect to the initial coherent state.
This quantity characterizes the increase of the uncertainty of the field due to amplification. As
the uncertainty of the field operator does not depend on its parameter, the quantity above shows
how much output field differs from a coherent state.
320 CHAPTER 11. ATOMS IN QUANTIZED RADIATION FIELDS
From the unitarity of U , it follows that b̂ satisfies the same commutation relations. The c-
numbers are then c2 − s2 = 1. Hence, the phase-invariant amplifier acts by introducing an
additional mode to the field, with a large amount of stored energy, behaving as a boson. Calcu-
lating the gain and the noise of this amplifier, one finds G = c, and
noise = c2 − 1 . (11.196)
The coefficient g = |G|2 is sometimes called the intensity amplification coefficient. The noise
of the linear phase-invariant amplifier is given by g − 1. The gain can be dropped by splitting
the beam; the estimate above gives the minimal possible noise of the linear phase-invariant
amplifier. The linear amplifier has an advantage over the multi-mode amplifier: if several modes
of a linear amplifier are amplified by the same factor, the noise in each mode is determined
independently;that is, modes in a linear quantum amplifier are independent.
To obtain a large amplification coefficient with minimal noise, one may use homodyne de-
tection, constructing a field state with known amplitude and phase, corresponding to the linear
phase-invariant amplifier. The uncertainty principle sets the lower bound of quantum noise in
an amplifier. In particular, the output of a laser system and the output of an optical generator
are not coherent states.
The multiplicative amplifier D also adds additive noise F . We have DD† = 1,
† †
aout a F1
= D in + . (11.197)
aout ain F2
where the field mode is expressed by the Hermitian quadrature components â0 = 2−1/2 ·(x̂θ +iŷθ ).
The expectation value of ∆Jˆ is afflicted with the Heisenberg uncertainty and can be expressed
as the first moment of the Wigner function W (α):
Z
ˆ
hψ|∆J|ψi = W|ψi (α)∆Jd ˆ 2α (11.202)
√ Z
= 2|αLO | W|ψi (xθ , pθ )xθ dxθ dpθ (11.203)
√ Z ∞
= 2|αLO | wθ (xθ )xθ dxθ .
−∞
Here, the distribution function integrated over a rotated quadrature component pθ is given by,
Z ∞
wθ (xθ ) ≡ Wθ (xθ , pθ )dpθ . (11.204)
−∞
This is called the radon transform. The distribution function wθ (pθ ) as well as the Wigner
function are normalized to 1. Multiple measurements of the expectation value R xθ = hψ|x̂θ |ψi now
yields a histogram H|ψi (xθ ) reflecting, if normalized, w|ψi (xθ ) = H|ψi (xθ )/ H|ψi (xθ ) exactly the
distribution function.
Considering the finite detector efficiency [160] wθ (xθ ) must be generalized to a convolution
with an apparatus function ζ(x):
√ 1 2
wθreal ( ηxθ ) = (wθideal ? ζ) onde ζ(x) = p e−ηα /(1−η) . (11.205)
π(1 − η)
A finite detector efficiency degrades the contrast of the quantum interference structures.
With the procedure of optical homodyne tomography or quantum state endoscopy the Wigner
function for
R e.g. a Schrödinger cat state can be reconstructed from a set of distribution functions
wθ (xθ ) = W (αeiθ )dpθ measured for various phases θ [160]. To do this the data set is exposed
to an inverse radon transform:
Z ∞ Z πZ ∞
1
wθ (xθ )|ζ|eiζ [Re (2 θ]
−1/2 αe−iθ )−x
W (α) = 2 dxdθdζ . (11.206)
4π −∞ 0 −∞
In contrast to the conventional homodyne detection, where the phase dependency of amplitude
noise is recorded, the homodyne tomography allows the complete reconstruction of a quantum
state through measurement of the distribution of the amplitude noise power,
Z
ωs ωs /2π
Ps = dt|hI(t)i|2 (11.207)
2π 0
for various phases.
Alternatively, to the homodyne method, one may reconstruct the photon distribution in field
modes from their temporal evolution [250]. Another method could be to use atoms as sensors
for the quantum state of a light field in a Jaynes-Cummings type dynamics.
emitir patentes de Gould para o bombeamento óptico e o laser baseado no princı́pio de descarga
elétrica num gás. A questão de como atribuir crédito para inventar o laser continua por resolver
pelos historiadores.
Em 16 de maio de 1960, Theodore Maiman operou o primeiro laser funcionando, no Hughes
Research Laboratories, Malibu, Califórnia, à frente de várias equipas de pesquisa, incluindo as
do Townes na Universidade de Columbia, do Schawlow na Bell Labs e do Gould na empresa
TRG (Grupo Técnico de Investigação). O laser operacional do Maiman utilizou um cristal de
estado sólido de rubi sintético bombeado por uma lâmpada de relâmpago para produzir luz laser
vermelho a 694 nm de comprimento de onda; no entanto, o dispositivo só era capaz de forma
pulsada, devido ao seu sistema de criação de bombeamento de três nı́veis. Mais tarde em 1960,
foi construı́do o primeiro laser de gás, usando hélio e neon que foi capaz de operação contı́nua no
infravermelho. Basov e Javan propuseram o conceito de diodo laser semicondutor. Em 1962, foi
realizado o primeiro dispositivo de dı́odo laser, feito de arseneto de gálio emitindo luz perto do
infravermelho. Hoje em dia existem diodos laser em vários regimes até o ultravioleta ’blue-ray’.
Interessantemente, apesar de muitas tentativas, ainda não foi possı́vel fabricar diodos laser
amarelos ou verdes.
Figure 11.14
Como o estado excitado tem mais energia, ele pode obviamente decair sozinho (isto é, espon-
taneamente) para um estados de energia inferior. A ideia brilhante do Einstein agora foi de
postular um terceiro processo, que ele chamou de emissão estimulada,
Neste processo, um fóton incidente estimula um átomo excitado de transferir o elétron para
uma órbita inferior. A energia liberada é então usada para formar um segundo fóton em todos
os respeitos idêntico ao primeiro. Este processo é necessário para garantir que em equilı́brio
térmico a população dos estados segue a lei de Boltzmann.
Figure 11.15
I(ν)B21 N
N2 = [1 − e−(A21 +2B21 I(ν))t ] < N1 . (11.212)
A + 2B21 I(ν)
1000
600
N1 , N2
400
200
aumentando I(ν)
0
0 1 2 3 4 5
tempo
Figure 11.16
Figure 11.17
muito claro como os átomos, que são os constituintes elementares, de qualquer objeto de matéria
devem cooperar para produzir os fenômenos bem conhecidos de espalhamento de luz, como a
refração, a difração ou o espalhamento de Bragg ou de Mie.
A nossa ideia é, que o melhor sistema para estudar essas questões são nuvens de átomos
resfriados para temperaturas extremamente baixas. Uma temperatura baixa significa, que o
movimento dos átomos fica muito lento, tão lento que o movimento não perturba a interação e
atrapalha as medidas. Assim, podemos medir a interação coletiva entre a luz e os átomos sob
condições puras, autênticas.
Nós jogamos agora um pulso curto de um feixe laser e observamos a reação dos átomos que
nos estudamos em circunstâncias variadas.
Q̂† (ξ)âQ̂(ξ) = â cosh r + e−iϕ ↠sinh rQ̂† (ξ)↠Q̂(ξ) = ↠cosh r + eiϕ â sinh r .
e
(11.215)
As formulas (11.215) descrevem uma transformação de Bogolubov. Resolve o Exc. 11.8.6.3.
Com hâi = 0,
∆x̂2b = e2r 1
2 Re hâ2 i + 1
4 + 12 h↠âi (11.222)
∆ŷb2 = e−2r − 12 Re hâ2 i + 1
4 + 12 h↠âi .
and,
sech r −(|α|2 +|β|2 )+(α∗ β+αβ ∗ ) sech r− 21 [eiθ (α∗2 −β ∗2 )+e−iθ (α2 −β 2 )] tanh r
Q|βi (α) = e (11.226)
π
and,
∗ 2 (α−α∗ )2
W|βi (α) = 1
2π exp(− (α+α
2e−2r
)
+ 2e2r
) . (11.227)
if the local oscillator can be considered classical, α = |α|eiθ . I.e. the phase of the local oscillator
permits us to select either one of the quadrature components.
Cavities are good for producing squeezing. However, the unused ports of the cavity let
vacuum fluctuations come in, which partially overrule squeezing.
The numbers of photons in the squeezed state,
†2 /2−ξ ∗ â2 /2 ∗2 /2−ξ ∗ α2 /2 |α|n e−|α| ξα∗2 /2−ξ∗ α2 /2
hn|α, ξi = hn|eξâ |αi = eξα hn|αi = e (11.232)
n! !
(eiθ tanh r)n/2 − 1 (|β|2 −e−iθ β 2 tanh r) βe−iθ/2
= e 2 Hn √ .
2n/2 (n! cosh r)1/2 2 cosh r sinh r
In the photon representation we can easily see that the squeezed vacuum is (unlike the coherent
and the Fock vacuum) not empty,
α→0
hα, ξ|n̂|α, ξi = |α|2 + sinh2 |ξ| −→ sinh2 |ξ| (11.233)
α→0
∆α,ξ n̂ = |α| + 2 cosh2 |ξ| sinh2 |ξ| −→ 2 cosh2 |ξ| sinh2 |ξ| .
Remember that the single-mode squeezing is obtained if â = b̂. In a number state base
1 X
|r, φi = (tanh r)n einφ |nia |nib . (11.239)
cosh r n
RT
where B̂(T ) ≡ ξ 0 dτ b̂† (τ )b̂(τ ).
Carrying out the ordering we get now,
*∞ Z Z Z +
X (−ξ)k+l T T T
† † †
Pk (T ) = TN dτ1 dτ2 .. dτk+l b̂ (τ1 )b̂(τ1 )b̂ (τ2 )b̂(τ2 )..b̂ (τk+l )b̂(τk+l )
k!l! 0 0 0
l=0
(11.254)
*∞ +
X (−ξ)k+l Z T Z τ1 Z τk+l−1
= dτ1 dτ2 .. dτk+l b̂† (τ1 )b̂† (τ2 )..b̂† (τk+l )b̂(τk+l )..b̂(τ2 )b̂(τ1 ) ,
k!l! 0 0 0
l=0
We have,
X ?
hn(n − 1)i(T ) = k(k − 1)Pk (T ) = hT N Â2 (T )i (11.257)
k
Z T Z T D E
?
=η 2
dτ dτ 0 T N b̂† (τ )b̂† (τ 0 )b̂(τ 0 )b̂(τ ) . (11.258)
0 0
hence,
R t+T R t+T
t dτ t − dτ 0 S 2 (τ τ 0 ) hT N E − (τ )E − (τ 0 )E + (τ 0 )E + (τ )i
Q(T ) = hni(T ) R 2 − 1 (11.260)
t+T
S t dτ hE − (τ )E + (τ )i
Z T h i
hni(T ) (2)
=2 dτ (T − τ ) g (τ ) − 1 .
T2 0
I.e. we can follow the evolution of the photon number variance as time goes on [243, 233, 238].
11.8. EXERCISES 333
g (2) (τ ) ∝ P̃ (τ ) . (11.262)
11.8 Exercises
11.8.1 Quantization of the electromagnetic field
11.8.1.1 Ex: Estatı́stica fotônica
Um ressonador óptico contem 10 fótons no modo T EM00q . Qual é a probabilidade de encontrar,
num qualquer tempo, 1 fóton resp. 10 fótons, quando a luz é (a) térmica, (b) coerente? Qual é
no caso (a) a temperatura da luz?
Mostraremos neste exercı́cio como, por operações coerentes num sistema de três nı́veis, podemos
criar correlações quânticas tipo gato de Schrödinger num modo óptico. No sistema ilustrado
na Fig. 11.18 imaginamos a transição inferior excitada por pulsos π/2 de radiação clássica
ressonante microonda (como descrito pela operação (11.35)). A transição superior é excitada
por pulsos π de uma luz laser quântica e sintonizada muito fora de ressonância, assim criando
uma dinâmica dispersiva (como descrito pela operação (11.35)). No tempo t = 0 o átomo seja
localizado
√ no estado |1i. Agora, aplica a seguinte sequência de pulsos: (i) pulso microonda
com n̄Ω12 t = π/2, (ii) pulso óptico com Ω223 t/4∆23 = π, (iii) outro pulso microonda π/2 e
finalmente (iv) um pulso óptico ressonante com a transição |2i-|3i projetando a população do
átomo em um dos estados da transição microonda. Descreve a evolução do estado do sistema
ao longo da sequência. Determine o estado final do modo óptico?
337
338 CHAPTER 12. ATOMIC MOTION IN ELECTROMAGNETIC FIELDS
caso de uma partı́cula livre, |ψiext = |pi. Para partı́culas num confinadas num potencial os
estados externos são os autoestados vibracionais, |ψiext = |ni. As evoluções temporais dos
graus de liberdade in- e externos são governadas por equações de Schrödinger independentes.
Para nuvens atômicas frias a energia cinética é bem menor do que a energia de excitação, o que
permite a separação das escalas de energia. Isto é, os graus de liberdade internos são gelados
no estado fundamental. Muitos fenômenos, por exemplo, a condensação de Bose-Einstein e a
dinâmica dos condensados são descritos neste regime.
Figure 12.1: Os graus de liberdade internos de átomos frios são termicamente gelados.
Não obstante, o fato de ser termicamente gelado não impede a excitação intencional do
grau de liberdade interno por irradiação de campos eletromagnéticos sintonizados perto de res-
sonâncias acoplando nı́veis eletrônicos de energia. Em caso de acoplamento, os graus de liberdade
externos e internos devem ambos ser considerados.
1 −~2 2 i~q
Ĥint = (−i~∇ − qA)2 + qΦ ' ∇ + A · ∇ + qΦ . (12.5)
2m 2m m
caracterı́stica quântica não levamos muito à sério até agora. Ele aparece na frequência de Rabi
e também no termo eik·r̂ . Agora, devemos nós lembrar, que
Umomentum = e−ik·r̂ = |p + ~kihp| (12.7)
e o operador unitário do recuo fotônico na absorção. Mostraremos em breve, que é justamente
este termo no hamiltoniano que dá origem à todos fenômenos de força da luz sobre átomos.
A presença do operador de posição no hamiltoniano de Jaynes-Cummings introduz um novo
grau de liberdade. Sem potencial externo (o sistema é invariante à translações espaciais), este
grau de liberdade é simplesmente o momento do centro de massa atômica, tal que o novo conjunto
de números quânticos é |j, n, pi. Estritamente falando deverı́amos abrir o espaço Hilbert total
pelo produto externo, Ĥele ⊗ Ĥrad ⊗ Ĥext .
Átomos naturalmente não exibem momento elétrico dipolar permanente, quando não são su-
jeitos à campos elétricos externos. Em contraste, moléculas polares, como por exemplo, dı́meres
heteronucleares, têm. Isso permite, influenciar seu movimento por campos elétricos inomogêneos
(vide Sec. 14.5.3).
Para simplificar, assumimos uma geometria unidimensional, B = B(z), mas podemos facilmente
generalizar para três dimensões. Os autovalores de H são
q
1
E± (z) = ± ~2 Ω2 + [µB gF B(z) − ~ω]2 . (12.13)
2
Suficientemente longe da ressonância, ~Ω |µB gF B(z) − ~ω|, obtemos
1 ~2 Ω2
E± (z) ≈ ± [µB gF B(z) − ~ω] ± , (12.14)
2 4[µB gF B(z) − ~ω]
onde o segundo termo pode ser interpretado como o deslocamento Stark dinâmico dos nı́veis de
energia.
Para ilustrar a ação da radiofrequência, calculamos a energia potencial e os estados vestidos
para átomos 6 Li. Por simplicidade, assumimos um gradiente de campo magnético 1D linear
B(z) ≡ zb. Fig. 12.2(a) mostra o acoplamento de radiofrequência e Fig. 12.2(b) os estados
vestidos para dois subestados magnéticos acoplados por uma radiofrequência. A mı́nimo apare-
cendo na curva superior da Fig. 12.2(a) pode servir como potenciais de confinamento. Usando
uma radiação rf com várias frequências, mı́nimos de potenciais podem ser gerados em várias
distâncias z. No Exc. 12.3.1.3 calculamos um exemplo.
A força na base dos estados vestidos é calculada a partir de,
d
F= p = i~−1 [H, p] , (12.15)
dt
com o hamiltoniano (12.12),
X
F(r) = hF̂(r)i = −Tr atom,laser ρ̂Or Ĥ = − hn, j|ρ̂∇r Ĥ|j, ni .
n,j
1
2
0.5
Δ/2π (kHz)
Δ/2π (kHz)
0 ω 0 Ω Δ
-0.5
-2
-1
-4 -1.5
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
z (μm) z (μm)
Figure 12.2: (Code: LM F orces P otencialAdiabatico.m) (a) Energias potenciais para um es-
quema de nı́veis F = 12 com um fator g de g = − 23 (como no caso do estado fundamental 2 S1/2
do 6 Li). Uma radiofrequência (seta) acopla os subestados mF = ± 12 . Aqui b = 200 G/cm e
ω = 2π × 5 kHz. (b) Estados vestidos não-acoplados (linha pontuada), estados vestidos acopla-
dos (linha sólida) e deslocamentos dinâmicos de Stark (dash-dotted) calculados na aproximação
longe de ressonância. A frequência de Rabi é Ω = 2π × 700 Hz.
A força da luz exercida sobre um átomo pode ser de dois tipos: uma força dissipativa
espontânea e uma força conservativa dipolar. A força espontânea surge do impulso experimen-
tado por um átomo quando absorve ou emite uma quanta de impulso fotônico. Como vimos na
342 CHAPTER 12. ATOMIC MOTION IN ELECTROMAGNETIC FIELDS
Seç. ??, quando um átomo espalha luz, a seção transversal ressonante de espalhamento pode
πλ2
ser escrita pela Eq. (??), σ0a = gg12 2 0 , onde λ0 é o comprimento de onda ressonante. Na região
óptica do espectro eletromagnético os comprimentos de onda da luz são da ordem de várias cen-
tenas de nanômetros, e as seções transversais ressonantes para espalhamento se tornam muito
grande, (∼ 10−9 cm2 ). Cada fóton absorvido transfere uma quanta de impulso ~k ao átomo na
direção de propagação. A emissão espontânea após a absorção ocorre em direções aleatórias, e ao
longo de muitos ciclos de absorção-emissão, ela cancela para zero em médio. Por consequência,
a força espontânea total age sobre o átomo na direção da propagação da luz, como mostrado
esquematicamente no diagrama da Fig. 12.3. A taxa saturada de espalhamento de fótons por
emissão espontânea (o valor recı́proco do tempo de vida do estado excitado) fixa o limite superior
para a magnitude da força. Esta força é às vezes chamada de força de pressão radiativa.
A força de gradiente dipolar pode ser facilmente entendida considerando a luz como uma
onda clássica. É simplesmente a força no médio temporal decorrente da interação do dipolo
da transição, induzida pela oscilação do campo elétrico da luz, com o gradiente da amplitude
do campo elétrico. Focalização do feixe luminoso controla a magnitude deste gradiente, e a
sintonização da frequência óptica abaixo ou acima da transição atômica controla o signo da
força que age sobre o átomo. Ajustando a luz abaixo da ressonância atrai o átomo para o centro
do feixe de luz, enquanto ajustando a luz acima da ressonância repele-lo. A força dipolar é
um processo estimulado em que não há troca de energia entre o campo e o átomo. Fótons
são absorvidos de um modo e reaparecem por emissão estimulada em outro. Conservação de
momento exige que a mudança de direção de propagação de fótons desde um modo inicial
para um modo final deixe o átomo com um recuo. Contrário a força espontânea, não há, em
princı́pio, nenhum limite superior para a magnitude da força dipolar, pois é uma função apenas
do gradiente de campo e dessintonização.
Dentro da teoria do electromagnetismo calculamos forças radiativas sobre cargas pelo tensor
de estresse de Maxwell 2 . A interação da radiação com átomos dotados de graus de liberdade
internas exibindo ressonâncias pode ser tratado qualitativamente pelo modelo de Lorentz 3 .
No seguinte, mostraremos cálculos quantitativos semiclássicos e quânticos: A força de um
feixe luminoso sobre um átomo pode ser calculada de muitas maneiras diferentes, cada uma
enfatizando um aspecto ligeiramente diferente: À partir da força de Lorentz clássica exercida
sobre um átomo por campos eletromagnéticos pode ser derivada uma equação semiclássica de
Fokker-Planck [236]. Na Sec. 12.2.1 derivaremos as duas contribuições (força dipolar e pressão
radiativa) em uma teoria semiclássica [97]. Wineland et al. [256] escolha como ponto de partida
a seção cruzada do processo elementar de espalhamento (Sec. 12.2.3). Dalibard et al. [60]
desenvolve uma teoria quântica usando a representação dos estados vestidos (Sec. 12.2.2). Cirac
et al. [46] mostram uma abordagem baseada na equação mestre (Sec. 12.2.3).
12.2.1 The dipolar gradient force and the radiation pressure force
Para calcular as forças da luz sobre um átomo, descrevemos o átomo como um sistema de dois
nı́veis: Um nı́vel fundamental |1i e um nı́vel excitado |2i decaindo para o nı́vel fundamental com
a taxa Γ. A diferencia de energia dos nı́veis é ω0 ≡ E2 − E1 . A luz seja um feixe laser com a
frequência ω, que pode ser dessintonizada da transição atômica, ∆ ≡ ω − ω0 . Para descrever
a interação, consideramos a parte (12.6) do hamiltoniano total descrevendo a interação [60].
2
Vide a apostila Eletrodinâmica do mesmo autor [55], Sec. 6.2.3.
3
Vide a apostila Eletrodinâmica do mesmo autor [55], Sec. 7.2.4.
12.2. OPTICAL FORCES ON FREE AND CONFINED ATOMS 343
Utilizando o operador de densidade ρ̂ 4 , podemos calcular a força que o campo da luz exerce
sobre o átomo,
F(0) = − 12 ~∇r Ω(0)(ρ12 e−i∆t + ρ21 ei∆t ) − 2i ~kΩ(0)(ρ12 e−i∆t − ρ21 ei∆t ) . (12.19)
As grandezas ρ12 ≡ h1|ρ̂|2i = ρ∗21 são as coerências, que se desenvolvem num sistema de dois
nı́veis excitado por um feixe laser. Inserindo as soluções estacionárias das equações de Bloch,
Ω2 (2∆ − iΓ)Ω
ρ22 = e ρ12 = e−i∆t . (12.20)
4∆2 + 2Ω2 + Γ2 4∆2 + 2Ω2 + Γ2
obtemos
4∆Ω ΓΩ2
F(0) = − 12 ~ ∇ r Ω + ~k . (12.21)
4∆2 + 2Ω2 + Γ2 4∆2 + 2Ω2 + Γ2
Γ 2
Com a definição da seção cruzada dessintonizada, σa (∆) = σa (0) 4∆2 +2Ω2 +Γ2 ,
2Ω2 Ω2 σa (∆)
F(0) = − 12 ~∆∇r ln 1+ + ~k . (12.22)
4∆2 + Γ2 Γ σa (0)
A seção cruzada ressonante para uma transição ’clássica’ é σa (0) = 3λ2 /2π.
Aparentemente, a força contem duas contribuições. A força de gradiente dipolar pode ser
derivada de um potencial. Ela é proporcional ao gradiente de intensidade e pode ser interpre-
tada como resultando de processos de absorção seguido por emissão auto-estimulada. Perto da
ressonância ela é dispersiva. Longe da ressonância pode ser aproximada por
A força de pressão radiativa é dissipativa. Perto da ressonância ela é absortiva. Ela é propor-
cional ao gradiente da fase e a única força ativa em ondas planas. Ela pode ser interpretada como
resultando de processos de absorção seguido por emissão espontânea. Com Ω2 = σa (0)ΓI/~ω
obtemos uma formula,
I
Frp = ~k σa (∆) = ~kγsct , (12.24)
~ω
que descreve a força como produto do número de fótons no feixe incidente, I/~ω, a seção
transversal para absorção, σa (∆) e o recuo por fóton, ~k. γsct é a taxa de espalhamento. A força
de gradiente dipolar (e o potencial associado) é frequentemente usado para aprisionar átomos, e a
força de pressão radiativa é frequentemente usada para resfriar-lhes. Note que ainda precisamos
4
O operador denside consiste de uma parte interna (excitação do átomo) e uma parte luminosa ρ̂ = ρ̂atom ⊗
ρ̂laser , onde a parte do movimento ρ̂motion foi desprezada, apesar de ter um papel importante em óptica atômica.
344 CHAPTER 12. ATOMIC MOTION IN ELECTROMAGNETIC FIELDS
Figure 12.3: Esquerdo: Um átomo com a massa m e a velocidade vA se move para o direito e
absorve um fóton propagante para esquerdo com o momento ~kL . Centro: Um átomo excitado
sofre uma mudança de momento pA = mvA − kL . Direito: A reemissão isotrópica de um fóton
resulta, no médio de muitos ciclos absorção-emissão, em uma mudança de momento para o
átomo de hpA i = mvA − kL .
corrigir as Eqs. (12.23) e (12.24) para tomar em conta o quadrado do médio sobre as orientações
do elemento da matriz do momento de transição, d12 /3.
O parâmetro de saturação,
1 2
Ω
s = 22 1 2 , (12.25)
∆ + 4Γ
permite escrever a força de gradiente dipolar e a força de pressão radiativa como,
~∆ 1 ~∆ ~kΓ 1
Fdp = − ∇s = ln[1 + s] e Frp = . (12.26)
6 1+s 6 6 1+s
Eq. (12.26) mostra que ’satura’ a força de pressão de radiação quanto s aumenta, e é, portanto,
limitada pela taxa de emissão espontânea. O parâmetro de saturação descreve essencialmente a
importância relativa dos termos que aparecem no denominador da função de perfil de linha para
as forças da luz sobre o átomo. A taxa de emissão espontânea é uma propriedade intrı́nseca do
átomo, proporcional ao quadrado do momento da transição atômica, enquanto que o quadrado
da frequência de Rabi é uma função da intensidade do laser incidente. Se s 1, a emissão
espontânea é rápida comparado a qualquer processo estimulado, e o campo de luz é dito ser
fraco. Se s 1, a oscilação de Rabi é rápida comparado a emissão espontânea e o campo é
considerado forte.√ O fator de perfil da linha indica um ’alargamento de potência’ por saturação
de um fator de 2. Note, que a força e o potencial de gradiente dipolar, Eqs. (12.26), não
saturam quando a intensidade do campo da luz é aumentado. Geralmente Fdp e Udp são usados
para manipular e aprisionar átomos com um laser sintonizado longe de ressonância para evitar
absorção.
Muitas vezes, o momento de transição pode ser orientado usando luz circularmente polar-
izada. Nesse caso, todas as expressões anteriores para Fdp , Frp e Udp devem ser multiplicados
por 3. A partir de agora vamos abandonar a mediação sobre orientações e usar só d212 para o
quadrado do momento de transição. Resolva os Excs. 12.3.2.1 e 12.3.2.2.
12.2. OPTICAL FORCES ON FREE AND CONFINED ATOMS 345
descreve, sob a influencia da interação dispersiva uma precessão em torno do eixo polar. Isso
foi discutido no Exc. 10.6.5.5. O deslocamento de Stark provoca uma rotação de ~−1 U (r)t.
Simultaneamente, o átomo é sujeito à uma força, que corresponde ao gradiente do potencial
−∇U (r), como ilustrado na Fig. 12.4(a). Vemos, que os deslocamentos de fase da onda de
Broglie e do vetor de Bloch são iguais. Finalmente, a fase do modo luminoso é deslocada pelo
mesmo montante num efeito chamado de refração. Ou seja, os graus de liberdade internos,
externos e luminosos são emaranhados.
Figure 12.4: O diagrama esquerda mostre estados produtos e estados vestidos para sintonização
azul. Note que a população está no nı́vel superior e o átomo está sujeito à força (repulsiva)
procurando campos fracas quando entra no feixe laser. O diagrama direito é similar, más para
sintonização vermelha. A população está no nı́vel inferior e o átomo está sujeito à força (atrativa)
procurando campos fortes.
Este fato tem, tem um uso prático em interferômetros atômicos, porque muitas vezes é mais
fácil detectar uma interferência dos estados de excitação interna, do que das ondas de Broglie.
Com o emaranhamento, basta medir uma interferência para conhecer a outra.
Por meio de variações locais do potencial U (r), e.g. por um laser focalizado, é possı́vel ma-
nipular uma frente de onda de Broglie do mesmo jeito que, em óptica clássica, manipulamos um
frente de onda de luz por lentes ou outros componentes, e.g. pelo ı́ndice de refração representado
por uma nuvem atômica perto de ressonância, como ilustrado na Fig. 12.4(b).
intensidade do laser, é obvio que um aumento dessa intensidade (potência óptica por unidade
de área) amplifica a força sobre o átomo (F ' ∇R Ω).
No seguinte vamos supor potenciais independentes do estado atômico interno. Como as-
sumimos o nosso potencial harmônico, podemos tratar as dimensões espaciais
p separadamente e
quantizar canonicamente como no caso de modos ópticos, definindo aho ≡ ~/mζ:
1 ẑ iaho † 1 ẑ iaho
b̂ ≡ √ + p̂ e b̂ ≡ √ − p̂ . (12.43)
2 aho ~ 2 aho ~
Obtemos:
m 2 2 1
V (z) = ζ ẑ = ~ζ(b̂ + b̂† )2 . (12.44)
2 4
onde ~Ω ≡ he|y|giE0 é a frequência de Rabi da transição interna. O último termo vem da trans-
formação na imagem de interação. O operador e−ikẑ corresponde à uma transformação do tipo
(2.184), ou seja, uma transição para um sistema se movendo com uma velocidade correspondente
à um recuo fotônico ~k. Obviamente, à partir do estado fundamental, o hamiltoniano somente
acopla estados vibracionais excitados,
hn, g|Ĥint |0, gi = 0 e hn, e|Ĥint |0, gi = 12 ~Ωhn, e|eikẑ |0, gi . (12.46)
O fato que, obviamente, somente termos proporcionais à hn|eikẑ |mi contribuem, é devido ao fator
de overlap deFranck-Condon e permite de analisar transição diretas entre nı́veis vibracionais do
mesmo oscilador.
Os elementos da matriz de transição são, com ẑ ∼ b̂ + b̂† :
X X
|hm|eikẑ |ni|2 = |hm|1 + ikẑ|ni|2 = 1 − ik 2 z 2 n2 + n+1
2 , (12.51)
m m
e,
1 −m~k2 /mζ
hm|eikẑ |0i = e . (12.52)
m!
Isso vale no regime de Lamb-Dicke.
Isto é, átomos frios em estados vibracionais n baixos não podem ajustar o deslocamento do recuo
fotônico ~2 k/m dentro do espectro vibracional da armadilha. Por consequência, o recuo deve
ser absorvido pela armadilha inteira. Este é o regime de aprisionamento forte (strong binding
regime) em armadilhas de ı́ons, que é análogo ao efeito de Mößbauer discutido embaixo.
fóton γ deve carregar sua própria energia de recuo, enquanto o momento do recuo é recebido pela
rede inteira. Isto é, a frequência não fica alterada, e a largura da transição é a largura natural.
Por isso, no regime de Lamb-Dicke, o efeito Doppler de primeira ordem cancela, e o átomo fica
no mesmo estado vibracional do seu potencial de localização. Isto é o efeito de Mößbauer. Além
do mais, ξ Γ. Um efeito similar pode ser observado em armadilha magneto-ópticas operadas
em transições muito finas, como discutido embaixo.
12.3 Exercises
12.3.1 Electromagnetic forces
12.3.1.1 Ex: O efeito Stern-Gerlach
Considere um condensado de Bose-Einstein de 87 Rb em uma superposição de dois estados Zee-
man. De repente um gradiente magnético de ∂z B = 100 G/cm é aplicado por 2 ms. Calcule a
separação dos centros de massa das duas partes do condensado depois de 10 ms de tempo de
voo.
Quantum measurement
Desde a fundação da mecânica quântica, a questão da relação entre o mundo e o que pode-
mos aprender sobre ele, isto é, entre a realidade e o observador, erá em primeiro plano das
discussões. Cientistas como Bohr, Heisenberg, Schrödinger e Einstein ocuparam posições con-
troversas e brigaram sobre a interpretação correta da mecânica quântica. O processo da medição
é suposto fornecer informação sobre o mundo lá fora, mas não erá claro se esta informação pode
ser completa e exata ou se existem limitações ou variáveis escondidas. Também, não erá claro se
a medida pode ser não-invasiva ou se o medidor sempre deve atrapalhar o fenômeno sob inves-
tigação. A etapa mais importante nesta questão foi a interpretação de Copenhague formulada
por Bohr, Heisenberg e Born em 1927 e elaborada por von Neumann e Dirac depois. Contestada
muitas vezes no passado, ela ainda permanece válida hoje em dia.
Neste capı́tulo vamos estudar o processo da medição visto pela mecânica quântica e discutir
alguns efeitos aparentemente paradoxos, que permitem aprofundar nosso entendimento. Entre
eles são o salto quântico, o gato de Schrödinger, o efeito Zeno quântico e paradoxo de Einstein-
Podolski-Rosen.
353
354 CHAPTER 13. QUANTUM MEASUREMENT
nós acostumamos com este fato simplesmente aceitando de considerar partı́culas como ondas.
Mas ainda não entendemos muito bem ”Porque o mundo clássico é tão diferente do mundo
quântico?”, ”Porque a mecânica quântica permite superposições quânticas de estados classica-
mente proibidas?”, ”Porque as leis fundamentais da mecânica quântica são invariáveis à respeito
da flecha do tempo, enquanto o mundo macroscópico sempre vai do passado para a futuro?”,
”Como pode ser que na mecânica quântica permite efeitos sem causa como a emissão espontânea,
enquanto o mundo cotidiano parece ser determinado?”
Claramente, em algum limite, a mecânica quântica deve abranger a fı́sica clássica. Más
apesar do teorema de correspondência de Ehrenfest, esse fato é longe de ser trivial. Algumas
previsões da fı́sica clássica e da fı́sica quântica são fundamentalmente diferentes e, em alguns
casos, contraditórias. Os estados do gato de Schrödinger são o epı́tome desse fato: Em uma
versão do famoso paradoxo, uma partı́cula atravessa uma fenda dupla. Por trás de uma das
fendas é um detector. Se ele registra uma partı́cula, um aparelho matando um gato está acionado.
Sabemos que, na realidade quântica a partı́cula atravessa as duas fendas em um estado de
superposição, tal que o gato deveria ser num estado de superposição também. Na mecânica
quântica os gatos podem estar em uma superposição de ”morto” e ”vivo”.
Figure 13.2: A medida de um sistema quântica pressupõe a interação do sistema com um reser-
vatório perturbando a sua dinâmica.
Acreditamos hoje que as respostas das perguntas acima são, de alguma maneira escondidas
nos processos que destroem as superposições dos gatos de Schrödinger na transição do mundo
13.1. THE OBSERVER AND THE REALITY 355
microscópico até o mundo macroscópico. No entanto, os detalhes deste processo de destruição das
coerências quânticas, chamado de decoerência, são muito complicadas e representam um assunto
de pesquisa contemporânea. Na pesquisa moderna, isso é uma das motivações para tentar criar
em laboratórios os maiores (quase macroscópicos) sistemas quânticos possı́veis, colocar-lhes em
estado de gato de Schrödinger e estudar a sua decoerência 2 .
Este processo é irreversı́vel, isto é, separa o passado do futuro. Esta evolução não é descrita pela
equação de Schrödinger. Em vez disso, a redução súbita do estado deve ser postulada, como
feito no famoso axioma de von Neumann.
Num segundo passo, o observador olha para o dispositivo de medição confirmando um dos
resultados possı́veis. Assim, ele transforma o estado em um auto-estado do dispositivo:
ρproj y ρmedidor = |kihk| . (13.3)
A partir deste momento, podemos de novo deixar o sistema quântico sozinho até a próxima
medida.
Visto dentro do sistema da amostra, a evolução do processo de medida é descontinua, pois
destrói todas coerências possı́veis entre seus estados. De fato, o problema vem do comportamento
não-ideal do dispositivo de medida (simbolizado por | ↑i antes da medida). Uma medida ideal
não-invasiva 4 deixaria o estado quântico |ψi inalterado:
H
|ψi| ↑i −→ |ψi| %i , (13.4)
enquanto o dispositivo de medida muda para um estado (| %i depois da medida) indicando o
estado atual da amostra. No entanto, isto é normalmente, impossı́vel sem correlação previamente
estabelecida entre |ψi e | ↑i. Num dispositivo de medida real, o acoplamento entre |ψi e | ↑i
requer que estes sistemas sejam não ortogonais.
2
Existem tentativas de introduzir o conceito da seta temporal também no mundo microscópico: ”In an isolated
system, spontaneous processes occur in the direction of increasing entropy.” [180].
3
Notamos, que somente se todas as observáveis comutandas do sistema são medidas e registradas, torna-se
ρmedidor um estado puro. Senão ρmedidor continua sendo uma mistura parcial.
4
Vide a discussão da medida de não-demolição quântica.
356 CHAPTER 13. QUANTUM MEASUREMENT
Figure 13.3: A emissão espontânea pode ser visto como acoplamento do sistema sob investigação
a um reservatório externo.
Example 51 (Medida quântica num sistema de dois qubits): Para discutir isso num
exemplo, consideramos o sistema mais simples imaginável: Dois átomos de dois nı́veis, onde o
primeiro representa o sistema que nós queremos medir e o segundo o medidor. Introduzimos
a seguinte base:
1 0 0 0
0 1 0 0
|1i ≡ | ↓i| ↓i =
0 , |2i ≡ | ↑i| ↓i = 0 , |3i ≡ | ↓i| ↑i = 1 , |4i ≡ | ↑i| ↑i = 0 .
0 0 0 1
13.1. THE OBSERVER AND THE REALITY 357
H = | ↓ih↓ | ⊗ | ↑ih↑ | .
no âmbito de incluir processos de dissipação de energia. O problema deste hamiltoniano é, ele
†
† −iĤef f t 6= eiĤef f t .
também não é comutável, [Ĥef f , Ĥef f ] 6= 0, e a sua dinâmica não é unitária, e
Isso significa, que já a possibilidade de emissão espontânea impede a reversibilidade da dinâmica.
Observamos um decrescimento temporal da norma hψ(t)|ψ(t)i indicando uma perda de energia.
que a norma está diminuindo
A perda da normalização durante a evolução até a ocorrência do próximo salto quântico é devido
a dissipação de energia dentro do reservatório,
ρamostra 0
Tr ρamostra → 0 enquanto Tr =1, (13.11)
0 ρreserv
e representa uma medida da probabilidade que um processo irreversı́vel tem sido ocorrido.
358 CHAPTER 13. QUANTUM MEASUREMENT
Processos dissipativos podem ser simulados jogando dados com números aleatórios ζ. Dividi-
mos o tempo em pequenos intervalos dt e propagamos a função de onda de ψ(t) para ψ(t + dt).
Depois de cada intervalo perguntamos, qual fui a probabilidade acumulada no tempo do inter-
valo para ocorrência de uma transição, 1 − hψ(t)|ψ(t)i. A transição é abrupta e se chama de
salto quântico. Agora, geramos um número aleatório ζ, uniformemente distribuı́do entre 0 e 1,
que nós comparamos á probabilidade. No caso, ζ > 1 − hψ(t)|ψ(t)i, concluı́mos que não houve
salto quântico, e deixamos o sistema prosseguir em paz, só renormalizando a função de onda
para compensar as perdas [178, 59]. No caso contrário, ζ < 1 − hψ(t)|ψ(t)i, concluı́mos que
houve salto quântico, e o sistema está projetado no auto-estado ψ0 . Agora, a evolução começa
de zero. A simulação pode ser feita assim,
!
(1−iĤdt)|ψ(t+dt)i
se ζ > 1 − hψ(t)|ψ(t)i
|ψ(t)i y |ψ(t + dt)i ≡ hψ(t)|ψ(t)i . (13.12)
|ψ0 i se ζ < 1 − hψ(t)|ψ(t)i
11
0.5
ρ
0
0 50 100 150 200
1
11
0.5
ρ
0
0 50 100 150 200
t (ns)
1. se existem processos de redução súbita de estados na realidade e se eles podem ser obser-
vados em saltos quânticos entre estados 5 ;
Tornamos nossa atenção para o átomo de três nı́veis: Obviamente, o átomo vai preferencial-
mente espalhar fótons na transição forte. No entanto, quando o elétron de valência é ’arquivado’
ou ’engavetado’ no estado excitado, nada de fluorescência pode ser observada na transição forte
dipolar. Entendemos a transição fraca como o objeto sendo medido pelo medidor, que neste
caso, é a transição forte.
Figure 13.5: (a) Medida quântica no exemplo do átomo de três nı́veis incorporando uma transição
fraca observada e uma transição forte observando. (b) Signal Random Telegraph na fluorescência
ressonante devido a saltos quânticos.
5
’If we have to go on with these damned quantum jumps, then I’m sorry that I ever got involved.’
(E. Schrödinger, 1952).
360 CHAPTER 13. QUANTUM MEASUREMENT
D5/2
0.5
0
1
S1/2 0.5
0
1
P1/2
0.5
0
0 200 400 600 800 1000
t (ns)
reservatório dos modos do vácuo) para o subespaço de estados tendo exatamente n de fótons no
reservatório,
ρ(n) = TraceF (P (n) ρAF ) , (13.14)
e derivamos, à partir da equação de von Neumann, a equação mestre para o estado atômico
ρ(n) sob a restrição de um número fixo de fótons n no campo. A partir da equação mestre,
derivamos agora as equações de movimento para o vetor de Bloch %(n) . Eles diferem apenas em
(n)
um termo das equações de Bloch: A expressão Γ12 ρ22 , que descreve o decaimento espontâneo
da população do nı́vel excitado da ressonância de monitoramento, é substituı́da pela expressão
(n−1)
Γ12 ρ22 :
d (n)
% = (L − |1iΓ12 h2|) %(n) + |1iΓ12 h2|%(n−1) , (13.15)
dt
P (n)
A condição do traço para o subestado com o número de fótons n é violada: j ρjj 6= 1. A
explicação fı́sica para isso é a seguinte: Enquanto a emissão induzida e a absorção mantêm o
número de fótons no sistema combinado luz-átomo, a emissão espontânea diminuı́ o número de
fótons. À redução do estado de superposição atômica nas equações de Bloch corresponde, nas
equações modificadas (13.14), o colapso do subespaço descrito por %(n) com a constante de tempo
Γ12 e a transição do sistema para o outro subespaço %(n−1) , cuja evolução é guiada por outra
equação diferencial correspondente, agora, para n − 1 fótons. Cada detecção de fluorescência
no momento t = 0 determina a condição inicial para o desenvolvimento futuro do sistema para
ρ(n) (0) = 0 e ρ(n−1) (0) = |1ih1|. A densidade de probabilidade c(t) para uma nova observação
de emissão espontânea no tempo t com a eficiência de detecção η, ou em outras palavras, o
histograma da durações dos perı́odos de escuridão no sinal de fluorescência relacionada com a
362 CHAPTER 13. QUANTUM MEASUREMENT
é um operador sobre o espaço Hilbert total e chamado de operador de Kraus. A respeito dos
operadores de Kraus o estado do sistema combinado depois da medida é,
Mq |ψi
|Ψq i = q ⊗ |qi . (13.21)
hψ|Mq† Mq |ψi
que ainda etiquetamos com o resultado da medida q. De fato, estas consideração permitem
derivara uma trajetória quântica.
Como exemplo canônico de um operador de Kraus usamos [18, 42] pegamos H = x ⊗ p, onde a
posição e o momento satisfazem a relação de comutação, [x, p] = i. A função inicial da ancilla
seja uma distribuição gaussiana,
Z
1 02 /4σ 2
|φi = dq 0 e−q |q 0 i (13.23)
(2πσ 2 )1/4
1 2 2
φ(q) = hq|φi = 2 1/4
e−q /4σ . (13.24)
(2πσ )
1 2 2
M (q) = hq|e−ix⊗p |φi = 2 1/4
e−(q−x) /4σ , (13.25)
(2πσ )
pois o operador e−ix⊗p faz uma translação espacial quando aplicado no grau de liberdade da
posição. Os elementos POVM correspondentes são,
1 2 /2σ 2
E(q) = Mq† Mq = √ e−(q−x) , (13.26)
2πσ 2
R
que obedecem dqE(q) = 1.
Calcular hψq |ψq i = hψ 0 |M (q)† M (q)|ψ 0 i.
Note que limσ→0 E(q) = |x = qihx = q|. Isto é, num limite particular, estes operadores
convergem para uma medida forte da posição. Para σ → ∞, falamos de medida fraca.
Outro exemplo seria o átomo de três nı́veis do exemplo do amplificador quântico de Dehmelt.
364 CHAPTER 13. QUANTUM MEASUREMENT
dρ̃tot
i~ = [Ṽ (t), ρ̃tot ] . (13.27)
dt
onde ρ̃tot (t) ≡ U0† ρtot U0 , e Ṽ (t) ≡ U0† V U0 e U0 = e−i(Ĥsys +Ĥres )t/~ . Essa evolução é, por
enquanto, muito geral, e a solução pode ser escrita formalmente
Z t
1
ρ̃tot (t) = ρ̃tot (0) + dt1 [Ṽ (t1 ), ρ̃tot (t1 )] . (13.28)
i~ 0
Esta maneira de terminar uma equação iterativa por ρtot (0) é geralmente conhecida como a
aproximação de Born. Aqui, vamos apenas até a segunda ordem em Ṽ . Tomando o traço sobre
o banho,
ρ̃sys (t) = Trres ρ̃tot (t) , (13.31)
extraı́mos a matriz densidade só do sistema, dando,
Z t Z t Z t1 h i
1 1
ρ̃sys (t) = ρ̃sys (0)+ dt1 Trres [Ṽ (t1 ), ρ̃tot (0)]+ dt1 dt2 Trres Ṽ (t1 ), [Ṽ (t2 ), ρ̃tot (0)] .
i~ 0 (i~)2 0 0
(13.32)
13.2. OPEN SYSTEMS AND THE MASTER EQUATION 365
Um outro pressuposto não é essencial, mas muitas vezes válido é, que Trres [Ṽ (t1 ), ρ̃tot (0)] = 0.
Neste caso, o termo de primeira ordem zera. Em segunda ordem, podemos escrever,
Z t Z t1 h i
M(t) 1
ρ̃sys (t) = e ρ̃sys (0) onde M(t)χ ≡ dt1 dt 2 Trres Ṽ (t1 ), [Ṽ (t2 ), χ ⊗ ρ res ] ,
(i~)2 0 0
(13.34)
é um superoperator agindo sobre o operador densidade do sistema. Pegando a derivada temporal,
temos a equação mestre explicita,
Z t h i
dρ̃sys (t) d 1
= Lρsys (t) = (M(t)ρ̃sys (t)) = dτ Trres Ṽ (t), [Ṽ (τ ), ρ̃sys (t) ⊗ ρres ] .
dt dt (i~)2 0
(13.35)
O superoperador L se chama operador de Lindbladt.
Em outras palavras, a função de correlação de dois pontos é significativo somente quando t ' τ ,
e é válido estender o limite superior para o infinito. Esta é a aproximação de Markov.
Trres (V (t)V (s)ρres ) = ~2 a† ahΓ(t)Γ† (t)ires eiΩ(t−s) + ~2 aa† hΓ† (t)Γ(t)ires e−iΩ(t−s) , (13.43)
onde hΓ(t)Γ† (t)ires ≡ Trres Γ(t)Γ† (t)ρres , e para o banho termico, hb†j bk i = δjk nk and hbj b†k i =
−1
δjk (1 + nk ) and nk = eβ~ωk − 1 . Portanto,
X D E Z t X Z ∞
†
T = −a aρ̃(t) gj gk bj b†k dτ e i(ωj t−ωk τ ) iΩ(t−τ )
e †
' −a aρ̃(t) gk2 nk dτ ei(ωk −Ω)(t−τ ) .
res 0 0
j,k k
(13.44)
Em seguida, teremos de usar a relação
Z ∞
dτ e±iετ = πδ(ε) ± iP V , (13.45)
0
−1 P 2
onde N ≡ eβ~Ω − 1 . Define γ2 ≡ k gk δ(ωk − Ω). O procedimento pode ser repetido
por todos os termos da equação mestre. Temos, então, a equação mestra para um oscilador
harmônico amortecido,
dρ̃ γ γ
= (N + 1) 2aρ̃a† − a† aρ̃ − ρ̃a† a − N 2a† ρ̃a − aa† ρ̃ − ρ̃aa† . (13.48)
dt 2 2
13.2.6 Thermalization
Para completar a discussão, vamos considerar o tempo de desenvolvimento do número de fótons
significa ha† ai. Note que Trres aa† ρ̃ = Trres aa† ρ, e pode ser útil para usar (com n̂ = a† a)
n̂a = an̂ − a e n̂a = an̂ − a para simplificar o lado direito da equação mestre. temos
dha† ai
= −γha† ai + γN , (13.49)
dt
ea solução para esta equação é
i 1
hψ0 |ψ(δt)i = |ψ0 i − Ĥδt/~|ψ0 i − 2 Ĥδt2 |ψ0 i + ... = |ψ0 i + |δψi , (13.52)
~ 2~
tal que
i 1 1
hψ0 |ψ(t)i ' 1 − Ĥδt − 2 Ĥδt2 e P (t) ' 1 − (hĤ 2 i0 − hĤi20 )δt2 . (13.53)
~ 2~ ~2
q
Dessa forma podemos explicitar o tempo de Zeno a partir das equações acima, τZ = ~2 / hĤ 2 i0 − hĤi20 .
Realizamos agora medidas de von Neumann sucessivas. Realizando ao total N medidas ao longo
de um tempo t, isso nos leva a uma frequência de medida τ −1 . Iremos fazer essas medidas a fim
de verificar se o sistema esta ainda no seu estado inicial, porém a cada medida nosso sistema é
projetado de volta ao estado inicial seguindo novamente seu processo de transição. A população
do estado inicial então será dada por:
P (N ) (τ ) = P (N ) (t/N )N . (13.54)
A Fig. 13.8 mostra a trajetória para cinco medidas com intervalos de tempo τ . A linha tracejada
mostra a população caso não fosse feita medida alguma, notamos uma diferença muito grande,
pois a cada medida o sistema retorna ao estado de evolução que está em regime de tempos
curtos. Há ainda de se notar que caso extrapolássemos o numero de medidas para infinito nossa
probabilidade iria cada vez mais se aproximar da unidade,
" 2 #N
t N grande 2 /N τ 2 N →∞
1− −→ e−t Z −→ 1 . (13.55)
N τZ
clássico? O efeito de Zeno quântico supõe a redução completa do estado para um autoestado.
No entanto, podemos imaginar medições clássicas que também reduzem o estado (como p.ex. a
medida da polarização de um feixe de luz).
pi numa única medida. No entanto, medindo as populações com uma amostra de n átomos ou
repetindo a medida n vezes com um único átomo em condições idênticas, podemos reduzir a
incerteza. A probabilidade de encontrar um átomo r vezes no estado |2i é,
n r
Pn,r,2 = p (1 − p2 )n−r . (13.56)
r 2
Assumimos que em qualquer instante de tempo o sistema fica num autoestado |n(r)i,
que a fase γn : C → γn (C) não é integrável. Berry mostra que o sistema acumulou, depois de
um ciclo completo no espaço dos parâmetros r(T ) = r(0) e |n(r(0))i = |ψ(0)i a fase,
ZZ
γn (C) = Vn (r)dS , (13.63)
C
onde
X hn(r)|∇r H(r)|m(r)i × hm(r)|∇r H(r)|n(r)i
Vn (r) = Im . (13.64)
(Em (r) − En (r))2
m6=n
Consideramos o exemplo de um sistema de dois nı́veis sem decaimento descrito pelo vetor
de Bloch,
Re Ω σx
1
H = Rσ ≡ Im Ω σy . (13.65)
2
∆/2 σz
e
i sin θ
ihn± (r)|∇r |n± (r)i = iêφ .
r
Finalmente,
I I I
− sin θ − sin θ
γ+ = êφ dR = r sin θdφ = sin2 θφ̇dt .
C r C r C
Definindo o espaço dos raios por |ψ̃(t)i = e−if (t) |ψ(t)i, obtemos,
A fase dinâmica δ pode ser zerada por uma escolha apropriada de Ĥ(t), mas não a fase topológica
β. β não depende de Ĥ(t), mas é uma propriedade geométrica da curva projetando H sobre P.
Em contraste com eiβ , a fase β é somente determinada modulo 2πn.
Figure 13.9: (a) Esquema para medir o efeito de Aharonov-Bohm. (b) Efeito de Aharonov-
Casher: os eletrodos φ não produzem campos elétricos dentro dos condutores; mesmo assim,
observa-se na saı́da do interferômetro interferência construtiva ou destrutiva dependendo do
potencial aplicado.
De (13.83) temos:
q
∇g(r) =A(r) . (13.84)
~
Agora, já mostramos na Sec. 2.5.3, que a função de onda
onde ΦB = πa2 B(0, t) é o fluxo do vetor magnético B pela seção transversal do solenoide.
Num ponto rsim do anteparo, localizado no plano de simetria do sistema, calculemos g(rsim )
por dois caminhos diferentes: ambos começam na fonte e terminam no anteparo final, mas
um atravessa pela fenda da esquerda e outro pela fenda da direita:
Z Z
q rsim qΦB 1 qΦB
g(rsim ) = A(x) · dx = φ̂ · (ρφ̂dφ) = ± .
~ 0 2π~ ρ 2~
13.7 Exercises
13.7.1 The observer and the reality
13.7.2 Open systems and the master equation
13.7.2.1 Ex: Equação mestre para cavidades
Seja κ a taxa de perda de fótons por transmissão através dos espelhos da cavidade. Mostre que
a equação mestre dá
dρ̂ κ κ
= (n̄ + 1) 2âρ̂↠− ↠âρ̂ − ρ̂↠â − n̄ 2↠ρ̂â − â↠ρ̂ − ρ̂â↠,
dt 2 2
P
onde o operador de densidade é definido por ρ ≡ m,n Pm,n |mihn|. e n̄ é o número de fótons
térmicos.
378 CHAPTER 13. QUANTUM MEASUREMENT
381
Part V
Atom Optics
383
Chapter 14
385
386 CHAPTER 14. MANIPULATION OF ATOMIC GASES
transition, soon after the dynamic response of condensates to the perturbations was the subject
of in-depth investigations, followed by the study of superfluid characteristics, quantum transport
phenomena, the interaction of condensed with light, of condensed gas mixtures [190, 235] and
the behavior of condensates in periodic potentials. To name only a few landmarks, we mention
the creation of vortices [171, 167] and quantum turbulence [122], the realization of various
types of atom lasers [174, 5, 30, 110]and atom interferometers with condensates [112, 152],
the coherent amplification of waves of matter [134, 153, 135, 72], the creation of the Mott
state in optical lattices [99], the study of condensates in reduced dimensions [186], Anderson’s
location of waves of atomic matter [43, 215], the observation of resonances of Feshbach type
collisions [53, 133, 248] and Efimov states of [154, 20], the creation of homonuclear molecular
condensates [100, 143, 56, 264, 260, 123] and heteronuclear [198] and degenerate Fermi gases
[69], to the observation of the BCS pairing [128, 101], the matter wave superradiance [134] and
the interaction of condensates with optical cavities [231, 40] and with surfaces [26].
It is clearly unthinkable to discuss all matters in this course. Let us, however, give a basic
and practical introduction to atomic optics with condensates. The course begins in this chap-
ter with a presentation of the most important experimental techniques for cooling, confining,
manipulating and detecting atomic gases. Knowledge of these techniques will allow a better
understanding of how it is possible to generate and analyze all the effects mentioned above. The
chapter ?? introduces the Bose-Einstein condensation phenomenon and the following chapters
focus on the thermodynamic, superfluid, coherent and dielectric properties of condensates.
The comparable success of atomic optics can be documented by Nobel laureates who have
been assigned to this area of physics:
• de Gennes (1991)
• Leggett (2003)
• Ashkin (2018)
For review articles on BEC see [200], [58], [50], [93], [217], [234], [146] and check in internet
sites http://amo.phy.gasou.edu/bec.html and http://jila.edu/bec.html.
In this chapter we review the basic techniques of Atomic Optics, emphasizing the cooling,
entrapment and measurement of cold atomic gases.
of a massive particle, we write the Hamiltonian of a free particle for a one-dimensional motion,
~2 d2
Ĥ = − . (14.1)
2m dx2
Therefore, the general solution of the Schrödinger stationary equation,
is r
ikx −ikx 2mE
ψ(x) = Ae + Be com k= . (14.3)
~2
Note well, that the functions eikx are not quadratically integrable. On the other side, they do
not represent actual physical systems. In practice, we need to consider wave packets or specify
a finite volume for the particle.
Note also that the spectrum of eigenvalues is continuous. To R guarantee the interpretation
2 3
as density of probability we will require quadratic integrability, |ψ| d r = 1. This means that
the wave function can not be infinite in a finite volume. But it can be infinite in an infinitely
small volume.
The description of the atomic motion by a wave equation emphasizes the fact that microscopic
particles have wave properties. Each atom corresponds to an de Broglie wave
h
λdB = , (14.4)
p
the width of the Γ transition. The Doppler limit is given by (see Exc. 14.7.1.1),
~
kB TD = Γ. (14.5)
2
We can also compare the kinetic energy with the energy transferred to an atom by the absorption
of a single photon. The photonic recoil is given by,
~2 k 2
kB Trec = . (14.6)
2m
Atomic clouds with temperatures around TD = 100..1000 µK they call it cold. Clouds with
temperatures around Trec = 100..1000 nK they are called ultracold.
In most experiments of atomic optics we do not work with individual atoms (or ions), but
with a set of atoms, called clouds. In general, sets can not be described by a single wave function.
Or we describe each atom by a separate wave function (which only works when atoms do not
interact), or we describe the cloud by probability distributions (such as a density matrix). We
now consider a thermal cloud. The Maxwell-Boltzmann distribution of velocities is,
r 3
m 2
g(v) = e−mv /2kB T . (14.7)
2πkB T
R R∞
This distribution is normalized, g(v)d3 v = 0 4πv 2 g(v)dv = 1. Average velocity is now
Z r
kB T
v̄ = vg(v)dv = . (14.8)
m
0.5
N
0
-5 0 5
v
This represents the average Broglie wave of the atoms in the set. When in a dense gas this
quantity exceeds the average distance between atoms,
where n is the density of atoms, we enter into a new scheme, where Maxwell-Boltzmann law is to
be valid. As λtherm ∝ T −1/2, this regime is at low temperatures. The amount ρ is called phase
14.1. THE ATOMIC MOTION 389
space density. When the density in the phase space approaches 1, it means that it increases the
probability of finding more than one atom per elementary cell of the phase space. Here we enter
the regime of quantum degeneration, where the Boltzmann statistic must be replaced by the
Bose-Einstein statistic in the case of bosons or Fermi-Dirac in the case of fermions. This will be
dealt with in the next chapter. Of condition nλ3therm ' 1 we obtain
2
1 2π~ (2π~)2 n2/3
kB Tc = = . (14.11)
m λtherm m
~2 d2
Ĥ = − + U (x) . (14.12)
2m dx2
As the wavefunction is now localized, the spectrum of possible energies organizes itself into
discrete levels. The atoms are allocated in populations of these levels.
Often, a 3D potential can be written in the form,
This is the case, for example, for a rectangular well characterized by Ux (x) = Uy (y) = Uz (z) =
U0 /3 inside the well. The relationship (14.13) also holds for a harmonic potential,
m 2 2
U (r) = ωx x + ωy2 y 2 + ωz2 z 2 . (14.14)
2
In these cases, it is generally useful to do the following ansatz for the wavefunction,
~2 ψx00 (x)
− + Ux (x) = const. ≡ Ex , (14.17)
2m ψx (x)
and similarly for y and z. Since E = Ex + Ey + Ez , we can have the same energy for different
combinations of Ex , Ey and Ez . That is, multidimensional systems are often degenerate.
Figure 14.3: (Code: AO T echniques Iof f eP otential.m) (a) The figure shows two dimensions of
a Ioffe-Pritchard type magnetic trapping potential (characterized by being approximately linear
at large distances from the center and harmonic near the center). (b) Harmonic approximation
(most experimentally feasible potentials are approximately harmonic near the center). (c) One-
dimensional cut through the potential of (a,b). (d) Density of states for a harmonic (dotted
line) and a Ioffe-Pritchard type potential (solid line).
or U (x) = m 2 2 2 2 2 2 2
2 ωr ρ , where ρ = x +y +λ z with λ = ωz /ωr . We define ω̄ = (ωr ωz )
2 1/3 = λ1/3 ω .
r
The individual particle levels of this Hamiltonian are, nx ny nz = ~ωx nx + ~ωy ny + ~ωz nz , where
nj are integer numbers.
We now introduce the density of states η() for nan arbitrary potential via,
Z Z Z Z p
1 3 3 (2m)3/2 3
η()d ≡ d rd k = d r d − U (r) . (14.19)
(2π) 3 (2π) ~
2 3
For the cylindrical harmonic trap defined in (14.18), we find with a little help from Dr. Bronstein
[38],
Z r
(2m)3/2 3 m
η() = d r − ωr2 ρ2 (14.20)
(2π) ~
2 3 2
Z 1 Z √1−x̃2 Z √1−x̃2 −ỹ2 p
1 82
= dx̃ √ dỹ √ dz̃ 1 − x̃2 − ỹ 2 − z̃ 2 .
(2π)2 (~ω̄)3 −1 − 1−x̃2 − 1−x̃2 −ỹ 2
I Ω2 Ω2
s= = 2 = 2 , (14.23)
Isat Ωsat Γ /2
e
~kΓ s
Frp = . (14.24)
2 (2∆/Γ)2 + 1 + s
Agora, se consideramos um átomo propagando na direção +z com a velocidade vz contrapropa-
gante á onda de luz dessintonizada da ressonância por ∆, a dessintonia total será
∆ −→ ∆ + kvz . (14.25)
392 CHAPTER 14. MANIPULATION OF ATOMIC GASES
onde o termo kvz é o deslocamento de Doppler. A força F− agindo sobre o átomo será na direção
contrária ao movimento. Em geral,
~kΓ s
F± = ± . (14.26)
2 (2(∆ ∓ kvz )/Γ)2 + 1 + s
Supõe que temos dois campos propagando nas direções ±z. Pegamos a força total F = F+ + F− .
Se kvz é pequeno em comparação com Γ e ∆, achamos através de uma expansão de Taylor
kvz (2∆/Γ)
Fz ' 4~ks . (14.27)
1 + s + (2∆/Γ)2
Esta expressão mostra, que se a dessintonia ∆ é negativa (isto é, para o lado vermelho da
ressonância), então a força de resfriamento vai se opor ao movimento e ser proporcional à
velocidade atômica. A Fig. 14.4 mostra essa força dissipativa restauradora em função de vz
numa dessintonia ∆ = −Γ e I = Isat /2. O movimento unidimensional do átomo, sujeito a
uma força de oposição proporcional à sua velocidade, é descrito por um oscilador harmônico
amortecido. O ’amortecimento Doppler’ ou ’coeficiente de atrito’ é o fator de proporcionalidade,
−4k 2 (2∆/Γ)
αd = s . (14.28)
1 + s + (2∆/Γ)2
No entanto, o átomo não vai resfriar indefinitivamente. Num certo ponto a taxa de resfria-
mento Doppler será equilibrada pela taxa de aquecimento vindo das flutuações de momento do
átomo absorvendo e reemitindo fótons. O limite de resfriamento Doppler é dado por,
Γ
kB T = ~ . (14.29)
2
This limit is generally, by alkaline atoms, in the order of hundreds of microkelvin. In the early
years of cooling and imprisonment, before 1988, the Doppler threshold was thought to be a
true physical barrier. But this year, several groups have shown that, in fact, atoms could be
cooled well below
114 the Doppler
CHAPTERthreshold. The FROM
6. FORCES effect is caused by the
ATOM-LIGHT hyperfine structure of the
INTERACTION
ground state. Let us describe simplified one-dimensional models in the next section. Resolve
û4ÿ û
!"#%$&
û4ÿ ý '" #(
!)$+*(," -. /#0
Γ
û4ÿ ûû
∆ω
4 132 Γ
ü4
û ÿý
4 5687
139
ü û4ÿ û
üªý û ü|þ û û þ û ý û
Γ
kB T = ~ (6.18)
2
which is called the Doppler-cooling limit. This limit is typically, for alkali atoms,
on the order of a few hundred microkelvin. For example the Doppler cooling
limit for Na is T = 240 microkelvin. In the early years of cooling and trapping,
prior to 1988, the Doppler limit was thought to be a real physical barrier, but
in that year several groups showed that in fact Na atoms could be cooled well
14.2. OPTICAL COOLING 393
Figure 14.5: The upper line shows the change in polarization as a function of distance (in units
of the wavelength) for the ’lin-perp-lin’ stationary waveform configuration. The figure below
shows a simplified image of the Sisyphus cooling mechanism for a two-level atom Jg ↔ Je .
desdobramento do light shift entre os dois estados fundamentais mostrado na Fig. 14.5. Como
o átomo continua a mover-se para +z, as forças relativas do acoplamento são revertidos perto
de 3λ/8, onde a polarização é essencialmente σ+ . Assim, os nı́veis relativas de energia dos dois
estados fundamentais hiperfinos oscilam ’fora de fase’, quando o átomo se move através da onda
estacionária. A idéia fundamental é, que a taxa de bombeamento óptico, sempre redistribuindo
população ao nı́vel hiperfino inferior, retarda o light shift do átomo se movendo através do
campo. O resultado é um ’efeito de Sı́sifo’, onde o átomo fica a maioria do tempo em subnı́veis
escalando uma colina de potencial, assim convertendo energia cinética em energia potencial. Essa
energia potencial acumulada é posteriormente dissipada por emissão espontânea para os modos
eletromagnéticos do vácuo. Simultaneamente, a emissão espontânea transfere a população de
volta para a os dois nı́veis mais baixos do estado fundamental. O diagrama inferior da Fig. 14.5
ilustra o atraso de fase do bombeamento óptico. Para que esse mecanismo de resfriamento
funciona, o tempo de bombeamento óptico, que é controlado pela intensidade da luz, deve ser
menor do que o tempo do light shift, essencialmente dado pela velocidade do átomo. Como o
átomo está se movendo lentamente, tendo sido previamente resfriado pelo mecanismo Doppler,
o campo de luz deve ser fraco, a fim de diminuir a taxa de bombeamento óptico e de atrasar
o bombeamento com respeito á modulação do light-shift. Essa imagem fı́sica combina a força
óptica dipolar conservadora, cujo integral espacial dá origem aos montes e vales do potencial,
sobre os quais o átomo se move, e a dissipação de energia irreversı́vel por emissão espontânea,
que é necessária para obter um resfriamento. Podemos tornar a discussão mais precisa e obter
expressões simples para o coeficiente de atrito e a velocidade de captura através de algumas
definições. Como no modelo de resfriamento Doppler nós definimos o coeficiente de atrito αlpl
como constante de proporcionalidade entre a força F e a velocidade atômica v.
F = −αlpl v . (14.30)
∆ = ω − ω0 , (14.31)
λ/4
τp = (14.32)
v
ou
Γ0 ' kv , (14.33)
onde Γ0 = 1/τp e λ/4 ' 1/k. Agora, a energia W dissipada durante um ciclo de escalada e
emissão espontânea é, essencialmente, a diferencia de energia média dos dois estados fundamen-
tais deslocados pelo light-shift, ∆ls ≡ ∆+ + ∆− , por exemplo, entre os pontos z = λ/8 e 3λ/8
ou W ' −~∆ls . Portanto, a taxa de dissipação de energia é
dW
= Γ0 ~∆ls . (14.34)
dt
No mesmo tempo, cada mudança de energia temporal de um sistema sempre pode ser exprimida
como dW
dt = F · v. Portanto, nesse modelo unidimensional, considerando a Eq. (14.31), podemos
14.2. OPTICAL COOLING 395
escrever,
dW
= −αlpl v 2 = −Γ0 ~∆ls , (14.35)
dt
tal que com (14.33),
Γ0 ~∆ls ~∆ls ~k 2 ∆ls
αlpl = − ' −kv ' − . (14.36)
v2 v2 Γ0
Note que desde que ∆ < 0, αlpl é uma grandeza positiva. Também note, que em grandes
dessintonias, (∆ Γ) a Eq. (12.23) dá,
U ∆ls Ω2
= = . (14.37)
~ 4 4∆
Também é verdade que para grandes light shifts em comparação com a largura natural
do estado fundamental (∆ls Γ0 ), e para grandes dessintonias no vermelho da ressonância
(∆ & 4Γ),
Γ ∆2
' (14.38)
Γ0 4Ω2 .
Portanto, o coeficiente de atrito sub-Doppler também pode ser escrito
~k 2 ∆
αlpl = − (14.39)
4Γ
Eq. (14.39) faz duas predições remarcáveis: Primeiro, o coeficiente de fricção sob-Doppler ’lin-
perp-lin’ pode ser um grande número em comparação com αd . Note que da Eq. (14.28), com
I . Isat e ∆ Γ,
3
2 Γ
αd ' ~k , (14.40)
∆
e
4
αlpl ∆
' . (14.41)
αd Γ
Segundo, αlpl é independente da intensidade do campo aplicado. Esse último resultado é difer-
ente do coeficiente de fricção, que é proporcional a intensidade do campo até saturação (vide
Eq. (14.28)). No entanto, mesmo quando αlpl parece impressionante, a gama de velocidades
atômicas onde pode operar é restrita pela condição,
Γ0 ' kv . (14.42)
A razão das velocidades de captura para resfriamento Doppler versus sob-Doppler é, portanto,
somente
vlpl 4∆ls
' . (14.43)
vd ∆
Fig. 14.6 ilustra graficamente a comparação entre os mecanismos de resfriamento Doppler e sub-
Doppler ’lin-perp-lin’. A diferencia dramática na gama de captura é evidente da figure. Note
também que as pentes nas curvas dão os coeficientes de fricção para os dois regimes e que, dentro
da gama de captura, a pente fica bem mais rı́gida para o mecanismo sub-Doppler.
396 CHAPTER 14. MANIPULATION OF ATOMIC GASES
118 CHAPTER 6. FORCES FROM ATOM-LIGHT INTERACTION
NPO Q
l
k NPOR
Γ
ij
h_
g bc NPO N
d ef
a bc S NPOR
^ _`
S NPO Q
S RUT S R#N ST N T RUN R#T
VW
XY[Z\]
14.2.2.2 σ + -σ − molasses
and second, that α is independent of the applied field intensity. This last
lpl
result differs from the Doppler friction coefficient which is proportional to field
intensity up to saturation (cf. Eq. 6.15). However, even though αlpl looks
O segundo mecanismo opera comthedois
impressive, range offeixes
by the condition that,
contrapropagantes
atomic velocities circularmente polarizados em
over which it can operate is restricted
sentidos opostos. Quando os dois feixes contrapropagantes Γ ' kv têm a mesma amplitude, a polar-
0
The ratio of the capture velocities for Doppler vs. sub-Doppler cooling it there-
ização resultante é sempre fore
linear
only.
e ortogonal ao eixo de propagação, mas a ponta do vetor de
4∆
polarização traça uma hélice com um passo de v /vλ' (vide Fig. 14.7). A fı́sica do mecanismo sob-
∆ω
lpl d
L
40 ~kv
hJz i = = ~(Π+ − Π− ) . (14.45)
17 ∆0
Note que a medida da orientação somente é diferente de zero quando o átomo se move. Na
Eq. (14.45) denotamos as populações dos subnı́veis |± > como Π± , e interpretamos os elementos
da matriz que não são zero como uma medida direta da diferencia de população entre os nı́veis
14.2. OPTICAL6.3.
COOLING
SUB-DOPPLER COOLING 119 397
σ+ σ−
o
n λprq λps
Figure 14.7: Polarização como função da distancia (em unidades do comprimento de onda) para
Figure 6.5: Polarisation as a function of distance (in wavelength units) for the
a configuração deσonda
+ estacionária σ + -σ − .
, σ − standing wave configuration.
it is sufficient to irradiate a single red-detuned running-wave light field: In a real ion trap the
cylindric symmetry cannot be realized with absolute precision so that we get different secular
frequencies ζx 6= ζy 6= ζz and a coupling of the degrees of freedom for all directions of space.
The cooling of the ionic motion in a single direction results in a cooling of the motion in the
other directions.
In the rest system of an ion oscillating in a harmonic trap the Doppler-shift of the laser
frequency changes periodically: v(t) = v0 cos ζr,z t. The ion absorbs therefore in its rest system
the light on a comb of sidebands whose distance and strength depend on the oscillation frequency
and amplitude. The absorption profile of a transition in such a harmonically vibrating ion follows
as a convolution of the Lorentz profile LΓ , describing the naturally broadened resonance, with
a function S, describing the splitting of the absorption profile into sidebands [138]:
X
A(∆) = (LΓ ? S)(∆) , S(∆) = Jn (k · v0 /ζr,z )2 δ(∆ − nζr,z ) . (14.49)
n
Jn denotes the Bessel function of nth order. In essence, the system is governed by three time-
constants: The natural decay width of the cooling transition Γ is a measure for the inneratomic
time scale, since it determines the average duration of absorption-emissions cycles. The secular
frequencies ζr,z determine the time scales for changements in the external degrees of freedom,
i.e. for changements of the ion’s location and velocity. The Doppler-shift kv0 of the resonance
frequency in the return point of the ion motion finally, is a measure for the kinetic energy of the
ion.
The relative importance of these three characteristic frequencies reveal the state of the ion
in the trap. The modulation index kv0 /ζr,z decides on the hight and the number of sidebands in
the excitation spectrum. The better the ion has been cooled, the smaller the modulation index
and the smaller the hight and number of sidebands. The kinetic energy of the ion is,
1 1 2 2
Ekin = mv02 = mζr,z x0 . (14.50)
2 2
The modulation index kv0 /ζr,z = kx0 = 2πx0 /λ is also called Lamb-Dicke parameter. By
cooling the der Lamb-Dicke parameter is so much reduced and the ion is so well localized that
its motional sidebands are smaller than the wavelength of the exciting light. It then is in the
so-called Lamb-Dicke regime x0 λ [?] and has so small motional sidebands that they do not
contribute to the line shape and do not influence the line width. Therefore the linear Doppler
effect vanishes.
The quantity ζr,z /Γ defines the resolution of the sidebands. If the resolution is poor, we talk
about weak confinement, else about strong confinement. Therefore the same ion can be weakly
confined with respect to an allowed transition and strongly confined with respect to a forbidden
transition. The cooling processes in the two cases of strong and weak trapping must be described
differently. At weak confinement the oscillation frequency ζr,z is so slow that many absorption-
emission cycles with the time constant Γ−1 can occur during one oscillation period. Cooling
process and cooling limit are approximately the same as for free particles and are described by
Doppler cooling.
by the narrow transition split into vibrational sublevels |nr,z i, which are populated in thermal
equilibrium according to the Bose-Einstein distribution and have the kinetic energies Ekin ,
1
nr,z = and Ekin = ~ζr,z (nr,z + 21 ) . (14.51)
e~ζr,z /kB T −1
To perform the so-called optical cooling sideband cooling [255] the laser is tuned to the first
lower sideband. The laser light is then scattered in a Raman-Anti-Stokes process at the excited
electronic state with a vibrational quantumnumber lower by 1 |e, nr,z − 1i. The subsequent
spontaneous decay occurs most probably to the same vibrational substate of the ground state
|g, nr,z − 1i. The net effect of such a scattering process therefore is a transition to the next lower
vibrational quantum number. The zero point energy of the ion in the trapping potential cannot
be underscored by cooling, Ekin > 12 ~ζr,z (for the Yb+ ion it is Ekin > 2 neV). However, the
uncertainty of the kinetic energy, and the temperature T given by (14.51) have no lower limit
[75].
At every absorption process, free particles carry away the momentum of the photons ~k.
The recoil of a free Yb+ ion corresponds to the frequency shift ε/2π = 5.3 kHz. On a narrow
transition, it yields a resonance at the frequency ε. For trapped ions, this is not the case, because
the momentum is absorbed by the whole trap (s. analogy to the efeito Mößbauer).
mωtrap
hni = . (14.52)
2~k 2
This means that higher trap frequencies ease the required temperature at which sideband cooling
can start to work.
|ω̇trap |
ωtrap . (14.53)
ωtrap
A população dos nı́veis quantizados não deve mudar sob descompressão adiabática, e~ωi /kB Ti =
e~ωf /kB Tf , e a densidade do espaço de fase fica inalterada, ni λdB,i = nf λdB,f . Se isto é verdade,
então a temperatura e a densidade mudam como,
3/2
ωf Tf nf
= = . (14.54)
ωi Ti ni
400 CHAPTER 14. MANIPULATION OF ATOMIC GASES
|G}~>
| }~ {
y ∆
|G}~#
vwx
tu
σ+ σ−
| }~ {
z
{
The atom will therefore experience a net restoring force pushing it back to
como es esquemathe de origin.
aprisionamento funciona,
If the light beams consideramos
are red-detuned ∼ Γ, um
then átomo de dois
the Doppler shiftnı́veis
of com uma
transição J = 0 →
theJatomic
= 1 movendo-se ao longo
motion will introduce da direção z. Aplicamos
a velocity-dependent term to theum campo magnético
restoring
force such that,
B(z) crescendo linearmente for small
com displacements
a distância and velocities,
da origem. the total restoringZeeman
Os deslocamentos force dos nı́veis
can be expressed as the sum of a term linear in velocity and a term linear in
eletrônicos dependem da posição,
displacement1 ,
.
FM OT = F1z + F2z = −αz − Kz. (6.33)
µB gF mF dB
∆B = z ≡ ∂ωzeem z , (14.55)
~ the equation
From Eq. (6.33) we can derive dz of motion of a damped harmonic
oscillator with mass m,
.. 2α . K
z+ z + z = 0. (6.34)
m m
The damping constant α and the spring constant K can be written compactly
in terms of the atomic and field parameters as
2
16 |∆0 | (Ω0 ) (k/Γ)
α = ~kΓ h i2 h i2 (6.35)
2 4(∆0 )2
1 + 2 (Ω0 ) 1 + 1+2(Ω 0 )2
14.3. OPTICAL AND MAGNETO-OPTICAL TRAPS 401
nas direções x, y, tal que κz = 2κx = 2κy . O termo de amortecimento proporcional à velocidade
implica que a energia cinética E se dissipa do átomo (ou de um conjunto de átomos) como
é tipicamente dezenas de microssegundos. É importante ter em mente que uma MOT é anisotrópica,
desde a força de restauração ao longo do eixo z é o dobro da força de restauração na plano x-y.
Além disso, uma MOT fornece uma armadilha dissipativa, não conservadora. Por isso é mais
correto caracterizar-a pela velocidade de captura máxima, em vez de dar a ’profundidade’ da
armadilha.
Os primeiros experimentos com átomos presos em MOTs foram realizadas desacelerando
um feixe atômico para carregar a armadilha. Mais tarde, uma fonte contı́nua não resfriada
foi usada para isso, mostrando que a armadilha pode ser carregada com átomos lentos de um
vapor em temperatura ambiente. Assim fui possı́vel utilizar uma célula de vapor para capturar
átomos frios da extremidade de baixa velocidade da distribuição de Maxwell-Boltzmann, e as
montagens experimentais tornaram-se mais simples. Agora muitos grupos no mundo utilizam
esses montagens para aplicações que vão desde a espectroscopia de precisão para o controle
óptico de colisões reativas. Resolva os Excs. 14.7.3.1 e 14.7.3.2.
%)P
+ ¢¡ £¤ ¥¦
§©¨«ª¬®ª %
+ ¢¡ £¤ ¥& 3
Figure 6.7: Hyperfine structure in sodium atom showing the usual cooling,
Figure 14.9: Estrutura hiperfina em átomos de sódio mostrando as transições usuais de resfria-
pumping, and repumping transitions.
mento, de bombeamento e rebombeamento.
red of resonance. The obvious advantage of large detunings is the suppression
of photon absorption. Note from Eq. 6.2 that the spontaneous force, involving
absorption and reemission, falls off as the square of the detuning while Eq. 6.8
shows that the potential derived from dipole force falls off only as the detuning
itself. At large detunings and high field gradients (tight focus) Eq. 6.8 becomes
14.3.2 Optical dipole traps ~ |Ω0 |
2
U' , (6.39)
4∆ω
Para aplicar variações temporais
which shows thatnothepotencial
potential becomesdedirectly
confinamento,
proportional to light o uso de campos magnéticos é
intensity
and inversely proportional to detuning. Therefore at far detuning but high
um procedimento lento e desajeitado. Do outro lado, temos feixes
intensity the depth of the FORT can be maintained but most of the atoms lasers que podem ser variados
rapidamente e de maneira MOTs bemare:localizada. A força dipolar exercida
will not absorb photons. The important advantages of FORTs compared to
12 −3
(1) high density (∼ 10 cm ) and (2) a well-defined polarization
por um feixe laser é igual ao
gradiente da frequência Rabi axis F = −∇(d · E), isto é, ela dá origem a um potencial óptico além do
along which atoms can be aligned or oriented (spin polarized).
disadvantage is the small number of trapped atoms due to small FORT volume.
The main
onde Pcav é a potência total do feixe e zR ≡ πw02 /λdip o comprimento de Rayleigh num dado
comprimento de onda λdip . O potencial dipolar é dado por (14.66). Usando a profundidade o
potencial,
3πc2 Γ 2P
U0 ≡ <0 , (14.70)
2ω03 ∆ πw02
podemos aproximar, perto do centro do potencial, isto é, próximo do eixo óptico, r 12 w0 , e
dentro do comprimento de Rayleigh, z πw02 /λ, o potencial por potencial harmônico 4 ,
(−2x2 −2y 2 )/w02 −z 2 /zR
2 2x2 + 2y 2 z2
U (r) ' U0 e e ' U0 1 − − 2 (14.71)
w02 zR
m 2 2 m 2 2 U0 r2 z2
≡ U0 + ωr r + ωz z ≡ kB T + 2+ 2 .
2 2 kB T 2r̄ 2z̄
Isso leva às equivalências,
q √ q
2 U0 2U0
ωr = w0 q m e ωz = zR q m
. (14.72)
w0 kB T zR kB T
r̄ = 2 U0 e z̄ = √
2 U0
Potenciais fracos podem ser deslocados ou distorcidos por gravidade. Por exemplo, numa
armadilha harmônica U = m 2 2 m 2 2 m 2 2 m 2 2 2 m 2 2
2 ωr r + 2 ωz z − mgz = 2 ωr r + 2 ωz (z − g/ωz ) − 2 g /ωz , os
átomos cedem uma altura g/ωz2 . Em armadilhas dependendo do tempo a gravidade causa um
comportamento mais complexo [111]. Trabalhos importantes têm sido feitos por [45, 207, 84,
106, 1, 62, 158, 159, 73].
4
O diâmetro de um feixe gaussiano pose ser caracterizado de várias maneiras,
r̄1/e2 -radius √ r̄1/2-radius
r̄1/√e-radius = √ = 2 r̄1/e2 -radius = ,
2 2 ln 2
e r̄-rms ≡ r̄1/√e-diam e r̄-hwhm ≡ r̄1/2-diam e r̄-diam = 2r̄-radius .
406 CHAPTER 14. MANIPULATION OF ATOMIC GASES
Figure 14.10: (Code: AO T echniques Dipoletraps.m) (esquerda) Potencial dipolar criado por
um feixe gaussiano. (direita) Potencial dipolar criado por uma onda de luz estacionária.
onde para a armadilha quadripolar, 2∂r B = ∂z B. Para calcular o raio rms de uma nuvem de
temperatura T aprisionada neste potencial colocamos U (r̄) ≡ kB T e obtemos a distribuição de
densidade,
√
r2 +4z 2 /r̄
n(r) = n0 e−U (r)/kB T = n0 e− . (14.81)
408 CHAPTER 14. MANIPULATION OF ATOMIC GASES
A normalização requer,
Z Z ∞Z ∞ √
2 2
N= n(r) d3 r = n0 e− r +4z /r̄ 2πr drdz (14.82)
R 3 −∞ 0
Z ∞Z ∞ Z ∞ 2|z|
2|z|
= n0 2πr̄ 2 −ξ
ξe dξdz = n0 2πr̄ 2
e− r̄ 1 + r̄ dz
−∞ 2|z|/r̄ −∞
Z ∞
= n0 2πr̄3 e−ζ (1 + ζ) dζ = n0 4πr̄3 .
0
Através da sua velocidade v o átomo veja uma mudança na orientação do campo magnético,
v · ∇|B| v
' . (14.84)
|B| r
Portanto, as transições de Majorana, que acontecem quando essa mudança está mais rápida do
que a frequência Larmor [205], ocorrem dentro de um raio rsf definido por,
v
ωLarmor (rsf ) < . (14.85)
rsf
Eles ocorrem com uma taxa correspondendo ao fluxo de átomos através do elipsoide,
1 N 3 v
' 3 rsf = ... . (14.86)
τsf Rcloud rsf
Em 1995 foram propostas várias técnicas para superar este problema. Foram propostas
geometrias de armadilhas magnéticas, como a armadilha cloverleaf ou a armadilha TOP (time-
orbiting potential), tendo no centro um campo magnético finito, ou armadilhas hı́bridas uti-
lizando um plugue óptico para tapar o vazamento de átomos. Estes avanços técnicos foram os
últimos necessários para atingir a condensação de Bose-Einstein. Resolva o Exc. 14.7.4.6.
a nuvem esfria, uma vez que os átomos se acumulam no fundo aproximadamente harmônico (e
portanto separável) do potencial. Este fato tem inibida uma evaporação eficiente em H↑ abaixo
de 120 µK [89].
Uma segunda técnica de evaporação foi demonstrada em armadilhas chamadas time-orbiting
potential (TOP) [205]. É uma caracterı́stica das armadilhas TOP de exibir uma região espacial
chamada ”cı́rculo fatal” (death-circle) onde átomos de passagem são ejetados. Este cı́rculo fatal
pode agir como uma superfı́cie de evaporação 2-dimensional se o raio do cı́rculo é suficientemente
grande [113]. Sob a influência da gravidade a dimensionalidade é ainda reduzida a 1D [144].
A técnica de evaporação de maior sucesso implementado até agora é baseada num acopla-
mento radiativo de estados confinados e livres. Discutimos essa técnica nas seções seguintes.
Publicações sobre resfriamento evaporativo são [166], [177], [205], [211], [?], [175], [185], [83],
[85], [118], [193], [125], [170], [64], [145], [?], [114], [168], [78], [126]. See ([50], Sec. 3.1.4) para
uma visão geral.
P = 1 − e−ξ , (14.106)
com ξ = 2π|V12 |2 /~gµB ∂z Bv. O teoria de Landau-Zener é estritamente válida apenas para um
sistema de dois nı́veis, que nós usamos aqui só para uma discussão qualitativa dos dois casos
limitantes.
Para um campo de rf fraco, ξ 1, P é muito menor do que 1, isto é, os átomos permanecem
predominantemente na superfı́cie diabática mostrado na Fig. 14.11(a). A probabilidade para
uma transição de spin (spin flip) é, P ≈ t, o que descreve o limite diabático do resfriamento
evaporativo induzido por rf: Os nı́veis atômicos de energia são quase não-perturbados, os átomos
de esparrinham muitas vezes através da superfı́cie de ressonância, e só depois de 1/P oscilações,
eles fazem um spin-flip do estado hiperfino |F, F i para o |F, F − 1i.
O limite adiabático é claramente a situação ideal para o resfriamento evaporativo. No en-
tanto, o processo de evaporação numa armadilha (com o tempo de oscilação Tosc ) satura em
uma potência rf menor. A condição para saturação é P ≈ Tosc /τel , onde τel é o tempo médio
entre duas colisões. Isso significa, que um átomo energético está evaporado antes que colidir
novamente.
Apenas a componente do campo magnético da radiação rf perpendicular ao campo magnético
de aprisionamento induz spin-flips. Em certas geometrias do potencial de confinamento, por
exemplo a armadilha quadrupolar, o campo magnético cobre o ângulo sólido inteiro. Por con-
sequência, há dois pontos onde o campo de aprisionamento e do campo rf são paralelos e os
elementos da matriz de transição são zero. Numa área em torno destes pontos, o caso de acopla-
mento diabático é realizado, mas na prática, a transição rf pode ser suficientemente saturada de
modo a que esta área é pequeno e não afeta fortemente a eficiência da evaporação.
Note também, que gravitação deforma a superfı́cie do potencial de confinamento e pode
reduzir a eficiência da evaporação [144]. Resolva o Exc. 14.7.4.8.
414 CHAPTER 14. MANIPULATION OF ATOMIC GASES
Figure 14.13: Criação de um buraco repulsivo por luz sintonizada pelo azul de uma transição
atômica.
Ωa /2π denota a frequência do campo aplicado, Vdc a amplitude da parte dc da voltagem, Vac
No último passo, assumimos que o potencial tem forma quase cilı́ndrica. Para uma dada geome-
tria As curvaturas bz,r podem ser extraı́das de simulações numéricas. A partir da equação de
continuidade, encontramos bz = −2br . A polaridade dos electrodos é modulada com a frequência
Ω,
φ(r, t) = φ(r)(ζ − cos Ωt) . (14.119)
As equações de movimento são derivadas de mr̈ = −e∇φ(r, t),
Introduzindo os parâmetros a e q,
8eφa bz ζ 4eφa bz
az = = −2ar e qz = = −2qr , (14.121)
mΩ2 mΩ2
14.5. OTHER TRAPS 417
Essas equações predizem órbitas estáveis, desde que os parâmetros a e q estejam dentro do
chamado diagrama de estabilidade mostrado em Fig. 14.16.
φ(r,z) (u.a.)
0
−2
10
10
0 0
z (mm) −10 −10 r (mm)
De acordo com essas equações, o ı́on passa por movimentos oscilatórios que são definidos
pelos parâmetros da armadilha ai e qi . Para que o movimento do ı́on seja finito, a sua amplitude
de oscilação pode não ultrapassar o limite definido pelos elétrodos. Esta condição impõe um
regimes possı́vel para os parâmetros da armadilha chamado diagrama de estabilidade [173].
0.2
-0.2
ar
-0.4
-0.6
0 0.5 1
az
No limite |ai |, qi 1 o ı́on se desloca só de uma pequena distância s r0 durante um perı́odo
de modulação Ωa . Então o ı́on sofre um movimento lento periódico chamado de macromovimento
dentro do potencial de aprisionamento com frequência secular ζi . Este movimento é modulado
por uma oscilação rápida chamada de micromovimento excitada pelo campo de modulação Ωa .
Sem voltagem dc aplicada entre o anel é as tampas, ai = 0, o movimento do ı́on é descrito pela
seguinte equação simples:
ri (t) = ri0 1 − 21 qi cos Ωa t cos ζi t , ζi = √1 qi Ωa
8
, i = r, z . (14.123)
A órbita do ı́on está confinada à região interna da armadilha, se sua energia cinética é inferior a
mζr2 r02 + M ζz2 z02 . Uma vez que a armadilha é, a qualquer momento, focando em algumas direções
e desfocando em outra, não se trata de um potencial conservador. O movimento oscilatório
418 CHAPTER 14. MANIPULATION OF ATOMIC GASES
y (µm)
r (µm)
0 0
−5
−2
−4 −2 0 2 0.2 0.4 0.6
x (µm)
1
Figure 14.17: (Code: AO T echniques IonsP aulT rapSimul.m) Movimento micro e macro do
v (cm/s)
ı́on. 0
−1
Outras geometrias para os eléctrodos, desviando do quadrupolo perfeito, também0.2 são
0.4 possı́veis.
0.6
Estas armadilhas também são bem descritas pela equação (14.116), quando o ı́on fica perto do
40
centro. Para nuvens de ı́ons, por exemplo, armadilhas multipolares de ordem superior [252],
eF (eV/cm)
20
armadilhas Paul-Straubl [225] e anéis de armazenamento [249] têm sido utilizados. Particu-
0
larmente importante para o armazenamento de muitos ı́ons resfriados −20
com aplicações na com-
putação quântica é a chamada armadilha de Paul linear [209, 88,−40 208]. Aqui os ı́ons imóveis
estão alinhadas numa cadeia linear. A vantagem da armadilha linear, em comparação0.2 0.4com out-0.6
t + 0 µs
ras armadilhas de muitas partı́culas, é o acesso óptico mais fácil de ı́ons individuais por lasers e
a possibilidade de zerar o micromovimento.
14.5.2 Micromotion
O movimento do ı́on na armadilha de Paul é uma superposição de duas vibrações com as
frequências de oscilação respectivas Ωa (frequência da modulação) e ζr,z (frequências seculares
para as vibrações em direção radial e axial). Para o ı́on em equilı́brio térmico, isto é não resfriado,
as energias cinéticas médias do micro- e do macromovimento são iguais [29].
O macromovimento pode ser reduzido por resfriamento ao contrário do micromovimento é
que constantemente excitado pelo modulação do campos elétrico aplicado [47]. Do outro lado,
a amplitude do micromovimento diminui com a distância do ı́on do centro da armadilha e, no
mı́nimo do pseudo-potencial desaparece totalmente. Portanto, para suprimir o micromovimento,
é impreterı́vel resfriar o macromovimento e pular o átomo para o centro, se necessário, usando
campos elétricos adicionais. Como a frequência do micromovimento é muito mais elevada do
que aquela do macromovimento, as bandas laterais dinâmicas podem ser facilmente resolvidas.
Quando a frequência de modulação Ωa é muito alta, as frequências seculares do macromovimento
também são altas. Neste caso, até as bandas laterais do macromovimento podem ser resolvidas
por transições atômicas dipolares largas. Este é chamado regime de acoplamento forte.
Devido a repulsão coulombiana só um único átomo pode ficar no centro da armadilha de Paul,
tal que é difı́cil zerar o micromovimento. Uma solução consiste em usar armadilhas lineares,
onde o centro, onde o micromovimento zera é estirado numa linha.
Outras geometrias para os eletrodos também são possı́veis. Para nuvens iônicas e.g. potenci-
ais multipolar de ordem superior [252], armadilhas de Paul-Straubl [225], e potenciais toroidais
[249] tem sido usados. Particularmente importante para o aprisionamento de vários ı́ons resfri-
ados é a chamada armadilha de Paul linear [209, 88, 208]. Aqui os ı́ons imóveis estão alinhados
14.5. OTHER TRAPS 419
Example 55 (Cálculo numérico do campo elétrico criado por uma superfı́cie car-
regada): Para calcular o potencial de aprisionamento de uma partı́cula carregada acima
de uma estrutura de microtrap planar, procedemos da seguinte maneira. A energia de uma
carga num campo elétrico é H = −eφ. O potencial eletrostático é dado pela lei de Coulomb,
Z Z Z
1 X ρ(r0 ) 0 1 X r − r0 0 1 X E(r0 ) 0
φ(r) = dV − φ n df + df ,
4π0 n Vn |r − r0 | 4π n Sn |r − r0 |3 4π n Sn |r − r0 |
where φn is the voltage applied to the n-th boundary. In practice, electric fields are generated
by electrodes set to specific voltages. Using the Dirichlet boundary conditions, we only retain
the second term. Furthermore, to account for the planar geometry of the chip electrodes, we
only consider surface boundaries in the y 0 = 0 plane,
Z
1 X ydx0 dz 0
φ(r) = − φn p 3 .
4π n Sn (x − x0 )2 + y 2 + (z − z 0 )2
This implies that the field lines cross the chip surface orthogonally, which in reality is only
true if the chip electrodes cover the whole area. Therefore, we only consider small gaps
f1
10 mm
f2
f3
f4
frf
Figure 14.18: Possible design for a microchip ion trap. φn are static potentials except for φ0 ,
which is alternates sign with radio frequency.
between the electrodes. We digitize the integral by dividing every electrode φn into a number
of identical surface elements ∆fm ,
1 X y∆fm
φ(r) = − φn p 3 .
4π n,m (x − xm )2 + y 2 + (z − zm )2
This formula can easily be evaluated numerically. A concrete example for a microchip ion
trap is shown in Fig. 14.18.
In practice magnetic fields are created by current-carrying wires. Those can be parametrized
420 CHAPTER 14. MANIPULATION OF ATOMIC GASES
Figure 14.19: (Code: AO T echniques IonsBecCalcs.m) Two-dimensional cuts through the elec-
tric potential generated by the microchip shown in Fig. 14.18 for φrf = 100 V and φj = 0.
14.5.3 QUEST
Átomos e dı́meros homonucleares não possuem momento dipolar elétrico permanente, mas po-
dem ter momento dipolar magnético permanente. Por isso, dı́meros homonucleares devem ser
confinados por gradientes de campo magnético ou um momento de dipolo elétrico deve ser in-
duzido por um campo eletromagnético oscilante. No regime óptico, isso foi demonstrado com a
armadilha quase eletrostática (QUEST).
Em contraste, dı́meros heteronucleares são moléculas polares com um momento dipolar
elétrico permanente, que pode ser bastante grande se eles estiverem profundamente ligados.
De acordo com o teorema de Earnshaw, não tem máxima de campo magnético ou de campo
magnético estático em espaço livre. Assim, nenhum estado ’high-field seeking’ pode ficar preso.
Em princı́pio, armadilhas ópticas dipolares tipo QUEST também podem ser usadas para dı́meros
heteronucleares. Mas o problema é que, em contraste com as moléculas homonucleares, são
possı́veis transições entre os estados fundamentais da vibração. Assim, a luz criando a QUEST
induz transições redistribuindo os ı́ons sobre todos os estados vibracionais.
14.6. ANALYSING TECHNIQUES 421
rf
Figure 14.20: Circuito de ressonância para detecção eletrônica do ı́ons. A armadilha de Paul
é operada por uma radiofrequência enquanto uma voltagem contı́nua é escaneada através do
diagrama de estabilidade. Simultaneamente, um oscilador é sintonizado perto de uma das
frequências seculares da armadilha. Quando o ponto de estabilidade é tal que a frequência sec-
ular coincide com a frequência do oscilador, o movimento dos ı́ons é parametricamente excitado
e o circuito de ressonância é amortecido. Este amortecimento é detectado por um amplificador
de banda estreita.
Assim, dois dipolos idênticos com 1 Debye = 10−27 /2.998 Cm= 10−19 /c Cm2 /s= 39.36 eaB
paralelamente orientados numa distancia r = 1 µm tem a energia,
1 p2
Ĥint = ≈ h × 1.5 MHz ≈ kB × 73 µK .
4πε0 r3
α 2δ
Im χ = e Re χ = . (14.128)
ω/c ω/c
Agora, os coeficientes de absorção e de dispersão podem ser relacionados com a secção transversal
óptica σ(∆) [164], onde ∆ é a dessintonia entre a frequência da luz e uma ressonância atômica,
cuja largura de linha é Γ. Esse resultado se chama teorema óptico,
∆
α = nσ(∆) e δ = nσ(∆) , (14.129)
Γ
onde n e a densidade da nuvem. Finalmente, obtém-se a lei de Lambert-Beer,
Z ∞
i ∆
E = E0 e iωL/c
exp iσ(∆) − n(r)dz ≡ E0 eiωL/c e−α/2 eiδ , (14.130)
2 Γ −∞
A absorção α descreve as perdas de intensidade para o feixe laser por espalhamento por
átomos desordenados. Ela é forte perto de ressonância, mas diminuı́ quadraticamente com a
dessintonia ∆. Ela dá origem à força de pressão radiativa. A dispersão δ descreve a refração
do feixe por uma distribuição de densidade continua [51, 12]. Ela desaparece em ressonância é
diminuı́ lentamente (∝ ∆) com a dessintonia. Ela está ligada à força dipolar. O coeficiente δ
descreve o deslocamento de fase da onda eletromagnética transmitida através da nuvem atômica.
do feixe I ∼ |E|2 pode ser relacionada através do coeficiente de absorção α (também chamado
de densidade óptica) à densidade atômica:
Z
I(x, y)
− ln = α(x, y) = σ(∆) n(r)dz . (14.132)
I0
A sombra impressa pela nuvem atômica sobre o perfil transversal do feixe laser é registrada
numa câmera CCD.
Já notamos, que a absorção está acompanhada de pressão radiativa. Após alguns eventos
de espalhamento, os átomos tem acumulado momento por recuo fotônico suficiente para ficar
fora de ressonância devido ao efeito Doppler. Fótons chegando aos átomos subsequentemente
não são mais espalhados e só aumentam a iluminação da CCD sem trazer informação sobre a
distribuição dos átomos. Consequentemente, é vantajoso usar pulsos de laser bem curtos. Além
disso, a intensidade do feixe laser não deve saturar a transição a fim de garantir uma secção
transversal óptica independente da intensidade e garantir a validade da lei de Lambert-Beer.
Finalmente, a frequência do laser deve ser sintonizada perfeitamente para ressonância, ∆ = 0.
De outro modo, acontece dispersão levando à focalização ou defocalização do feixe por refração.
Isso distorce a imagem e vira impossı́vel a estimação do tamanho da nuvem.
A Fig. 14.22 mostra exemplos de imagens de absorção de uma nuvem atômica tomados
em fases diferentes do processo de evaporação. A Fig. 14.22(a,b) fui tomada na temper-
atura 320 nK; a nuvem é grande e isótropo, e portanto, puramente térmica. Em 250 nK [ver
Fig. 14.22(c,d)] uma parte de forma elı́ptica aparece no centro da nuvem térmica. E em 180 nK
[ver Fig. 14.22(e,f)] a nuvem térmica desapareceu quase completamente para o benefı́cio do con-
densado. Uma avaliação quantitativa da fração condensada é dada na Sec. 15.1.4. Resolva o
Exc. 14.7.6.1.
424 CHAPTER 14. MANIPULATION OF ATOMIC GASES
Figure 14.22: Imagens de absorção após tempo de vôo permitem identificar a presencia de um
condensado de Bose através da sua distribuição de momento caracterı́stica. Mostrado são uma
foto acima (a), ligeiramente abaixo (b), e muito abaixo da transição de fase (c) (figuras de [113]).
2 ∆2
α→0
I¯dg = 12 |E − E0 |2 = I0 e−α/2+iδ − 1 −→ I0 δ 2 = I0 α 2 . (14.133)
Γ
I(y, z) R
= e−σ n(r,z)dx = e−σf (y,z) . (14.135)
I0 (y, z)
• MW AbelTrafo4: Forward Abel transform, comparison with column integration and ana-
lytical formula.
• MW AbelTrafo5: Forward Abel transform, comparison with column integration and ana-
lytical formula in 2D.
14.7 Exercises
14.7.1 The atomic motion
14.7.1.1 Ex: Limite de recuo
Calcule o limite de Doppler, o limite de recuo e o limiar da degenerescência quântica para uma
densidade de n = 1014 cm−3 para a transição de sódio D2 (λ = 590 nm, Γ/2π = 10 MHz) e a
transição de rubı́dio D2 (λ = 780 nm, Γ/2π = 6 MHz).
Mostre que não é possı́vel criar máxima de intensidade com um potencial magnético.
Considere (a) uma armadilha quadrupolar, (b) uma armadilha harmônica isotrôpica. Qual é o
gradiente, respectivamente curvatura necessário para levantar uma nuvem de rubı́dio sujeito à
gravitação? Qual é a inclinação da nuvem dentro do potencial devido à gravitação?
14.7. EXERCISES 429
Como varia a temperatura quando está comprimida adiabaticamente (a) uma armadilha quadrupo-
lar, (b) uma armadilha harmônica. Como variam a densidade, a densidade do espaço de fase,
a taxa de colisões elásticas. Ajuda: Define a compressão para armadilha quadrupolar como
η ≡ ∂r Br,f inal /∂r Br,initial e para armadilha harmônica como η ≡ ωr,f inal /ωr,initial .
A armadilha TOP (time-orbiting potential trap) fui a primeira sucedida para criação de conden-
sados de Bose. Ela consiste da superposição de um campo magnético quadrupolar (gradiente
2∂r Bqua = ∂z Bqua ) e um campo magnético homogêneo rotatório no plano de simetria do campo
quadrupolar, Btop . Átomos oscilando com uma amplitude além de um dado raio rd , chamado
do ’circulo da morte’, fazem transições de Majorana e são expulsados da armadilha. Calcule o
raio da morte.
An interesting system is the 1D array of annular optical potentials realized in a standing wave
formed by red-detuned Gaussian beam and a counterpropagating blue-detuned doonat-mode. In
general, the tight longitudinal confinement freezes out the axial motion by quantum confinement.
It can be readily shown [257] that in the far-off resonance case and if the potential is approximate
by a harmonic potential around its minimum the eigenenergy spectrum is given by,
~2 p2
Epq = U0 + ~ω(q + 12 ) + .
2mR02
It thus reproduces the ro-vibrational spectrum of a 2D artificial molecule and gives rise to two
normal motions. In its ground state, we have the atom optical analog of a 2D rigid rotator.
Gravity plays formally the same role as static electric fields for molecules. Such systems might
be interesting for investigating the selection rules for transitions between ro-vibrational states
involving conservation of total angular momentum of light and atoms and yield insight into the
concept of orbital angular momentum of light fields.
Como descrever o amortecimento em misturas de espécies e como utilizar uma medida do tempo
de amortecimento para determinação do comprimento de espalhamento interspecies?
430 CHAPTER 14. MANIPULATION OF ATOMIC GASES
Figure 15.1: Illustration of atomic Broglie waves. From the top to the bottom the temperature
of the atoms is decreasing.
Motivated by the need to test the concept of condensation of composite particles in weakly
interacting systems, in 1962 Blatt et al. proposed the investigation of the BEC in gases of
excitons [28]. Excitons are bound electron-hole pairs that can form a weakly interacting gas
in certain non-metallic crystals. They are interesting because their small mass allows BEC at
high temperatures and gas density can be controlled over a wide range, by only modifying the
intensity of the optical excitation. Being quasi-particles, excitons can be created and annihilated,
that is their number is not conserved. Excitons were discovered in 1968, and the first evidence
for Bose-Einstein of biexciton molecules in a CuCl crystal dates back to 1979 [44].
The laser as coherence phenomenon between photons shares many analogies with conden-
431
432 CHAPTER 15. THERMODYNAMICS OF QUANTUM GASES
sates. However, photons are quasi-particles as well, and again their number is not conserved 1 .
Hence, there is no phase transition: When an optical cavity containing photonic modes is cooled,
the photons prefer to disappear in the walls of the cavity instead of condensing.
Hecht [117] suggested in 1959, followed by Stwalley and Nosanow [240] in 1976, that an
atomic hydrogen gas with polarized spins would be an appropriate candidate for BEC. The
advantage of this system is that interactions between atoms are weak and only give rise to a
negligible quantum depletion below 1%. In 1978 Greytak and Kleppner started at the MIT
intensive efforts to generate BECs in dilute hydrogen gases. In the 1990s, important advances in
the cooling of atoms using laser light allowed to reach very low temperatures, and the invention
of the magneto-optical trap (MOT) for neutral atoms permitted their spatial confinement and
the compression of their density. These successes boosted efforts to try to create BEC in alkaline
gases, which have electronic level schemes that lend themselves to optical cooling. Later, it was
discovered that the phase space density in MOTs is limited by radiation trapping effects. As
a solution to this problem, scientists had to learn how to trap atoms without the use of light
in conservative traps, e.g. by their magnetic dipole moment, and to replace optical cooling
with evaporative cooling. This was the crucial step that finally permitted to reach BEC in
alkaline gases in 1995. Later, the hydrogen experiment, which initially stimulated the alkaline
experiments, now taking advantage of their success, has been taken to BEC as well [89].
Why did it take so long to reach Bose-Einstein condensation, seven decades after its pre-
diction by Bose and Einstein? How can we see when we have a condensate? What are the
characteristics of a BEC accessible to observation and how to measure them? These are the
answers that we will answer in the following sections. Solve Exc. 15.3.1.1.
Figure 15.2: Scheme of the Broglie wave of cold atoms. From top to bottom the temperature
decreases.
well the behavior of weakly interacting hot atoms. Deviations from this law are insignificant
until, at low temperatures, quantum effects come into play. For this to happen the temperature
must be so low that the atomic Broglie wavelengths become comparable to the average distance
between particles. For a gas in thermal equilibrium the characteristic wavelength, called thermal
de Broglie wavelength, is,
s
2π~2
λth = , (15.2)
mkB T
where ~ = h/2π is Planck’s constant. In a gas of density n, the mean distance between particles
is n−1/3 . So, quantum effects are expected to emerge when n−1/3 ∼ λdB (T ), such that the limit
for this regime is defined by,
2π~2 2/3
kB T (n) = n . (15.3)
m
For example, an atomic gas with density n ∼ 1016 cm−3 and temperature 900 K is certainly
in the classical regime, since n−1/3 ∼ 106 cm λdB = 10−9 cm. To observe quantum effects,
we need relatively dense and cold clouds of atoms. In most gases, lowering the temperature or
increasing the density promotes the system to liquidity before the quantum regime is reached.
Well-known exceptions are spin-polarized hydrogen (H↑), which does not become liquid and
helium, which exhibits quantum degeneracy effects in the liquid phase, although these effects
are quite complex due to strong interparticle forces.
We have already seen that all particles in the quantum world are either bosons with integer
spin or fermions with semi-integer spin. Fermions do not share a quantum state, because they
must follow the Pauli’s exclusion principle. They obey a quantum statistical distribution called
Fermi-Dirac distribution (FD). In contrast, bosons enjoy to share a quantum state and even
encourage other bosons to join them in a process called bosonic stimulation. Bosons obey a
quantum statistical distribution called Bose-Einstein distribution (BE). The basic difference
between the MB-statistics on one hand and the BE- or FD-quantum statistics on the other is
that the former applies to identical particles which, however, are distinguishable from each other,
while the second describes identical indistinguishable particles. For the BE/FD statistics one
can derive [149] the occupancy number for a non-degenerate quantum state having the energy
ε when the system is kept at temperature T ,
1 1
wT,µ = 3 β(ε−µ)
, (15.4)
(2π) e ∓1
where we used the abbreviation β ≡ 1/kB T . The upper sign refers to the BE statistics, the
lower sign to the FD statistics. The chemical potential µ is an important system parameter,
which helps to normalize the distribution (15.4) to the total number of particles,
X
N= wT,µ (ε) . (15.5)
ε
A very remarkable effect occurs in a bosonic gas at a certain characteristic critical temperature
Tc : below this temperature a substantial fraction of the total number of particles occupies the
434 CHAPTER 15. THERMODYNAMICS OF QUANTUM GASES
lowest energy state, while all other states are occupied by a negligible number of particles.
Above the transition temperature the macroscopic observables of the gas, such as pressure, heat
capacity, etc., receive contributions of all states with a certain statistical weight, but without
favoring the state of lower energy. Below the transition temperature, the observables are altered
by a macroscopic occupation of the ground state, which results in dramatic changes of the
thermodynamic properties. The phase transition is named after Shandrasekar Bose [32] and
Albert Einstein [80] Bose-Einstein condensation (BEC).
Z ≡ eβµ , (15.7)
we may write, Z ' 1 − 1/N . It should be noted that the chemical potential in a bosonic system
must always be less than the ground state energy in order to guarantee non-negative occupation
wT,µ (ε) of any state. Z ' 1 denotes macroscopic occupation of the ground state. We define
the critical temperature for Bose-Einstein condensation via the occupation of the ground state.
Above this temperature the occupancy of ground state is not macroscopic, below it is.
For a Bose gas of N non-interacting particles with mass m confined inside a box potential of
volume V = L3 the critical temperature for BEC can be calculated from equation (15.3). The
boundary conditions require that the momenta satisfy pj = 2π~lj /L, where j = x, y or z and j
are integers. Each state is labeled by a set of three integers (lx , ly , lz ). In the thermodynamic
limit, the sum over all quantum states can be converted into an integral over a continuum of
states,
X N →∞ 1 Z Z
−→ 3 d3 rd3 p = d3 rd3 k . (15.8)
~
r,k
For a free gas with energy ε = p2 /2m we can derive, simplifying the calculation using
the density of states η defined by (14.19), which basically depends on the geometry of our
system 2 . Using the occupation number wT,µ (ε) for the Bose-Einstein distribution (15.4) in the
thermodynamic limit and the density-of-states in a box potential (14.22), let us calculate the
total number of particles,
ZZ Z ∞
3 3 3
N = N0 + wT,µ (ε(r, k))d rd k = N0 + (2π) wT,µ (ε)η(ε)dε (15.9)
0
√ 3 Z ∞
2m ε1/2 dε
= N0 + V ,
(2π)2 ~3 0 eβ(ε−µ) − 1
where the ground state population N0 is maintained explicitly. In the process of converting the
sum to an integral (15.8) the density of states disappears when we approach the ground state.
This error is corrected by adding a contribution of an explicit term N0 to the integral.
2
We must, however, keep in mind that the state density approach is an approximation not valid for experiments
with a limited number of atoms.
15.1. QUANTUM STATISTICS OF AN IDEAL BOSE GAS 435
where Γ(η) denotes the Gamma function. Analogically, we can define the Fermi function via,
∞
X Z ∞
(−Z)t 1 xξ−1 dx
fξ (Z) = − ξ = . (15.11)
t Γ(ξ) 0 Z −1 ex + 1
t=1
(a) 3 (b) 6
c2
g3 5
2 f3
g9 (z)
4
1(9)
g 3/2
f 3/2 3
1
2
0 1
0 0.2 0.4 0.6 0.8 1 1 2 3 4
z 9
Figure 15.3: (Code: AO T hermodynamics BoseF ermiF unction.m) (a) Bose and Fermi func-
tions for box potentials (g3 and f3 ) and for harmonic potentials (g3/2 and f3/2 ). Also shown is
the Boltzmann limit (15.12). (b) Riemann function.
With this definition and the definition of the thermal Broglie wavelength (15.2), Eq. (15.9)
becomes,
V
N = N0 + 3 g3/2 (eβµ ) . (15.14)
λth (T )
(3/2)
We can use Eq. (15.14) to calculate the critical temperature3 Tc defined for N0 % 0 and
(3/2)
µ → 0. Above the phase transition, T > Tc , the population is distributed over all states,
(3/2)
each individual state being weakly populated. Below Tc the chemical potential is fixed by
µ = 0 and the number of the particles occupying the excited states is,
!3/2
V T
Nth = 3 g3/2 (1) = N (3/2)
, (15.15)
λth (T ) Tc
3
The superscript (3/2) denotes the box potential shape of the trapping potential. See 15.3.1.3 for an explanation
of the notation.
436 CHAPTER 15. THERMODYNAMICS OF QUANTUM GASES
with g3/2 (1) = 2.612. Since N0 + Nth = N , the number of particles in the ground state is given
by,
!3/2 2/3
(3/2)
N0 min(T, Tc ) ~2 N
=1− (3/2)
with Tc(3/2) = , (15.16)
N Tc 2πm V g3/2 (1)
N0 /N is the fraction of the atomic cloud which is condensed in the ground state. The abrupt
(3/2)
occurrence of a finite occupation in a single quantum state at temperature below Tc indicates
a spontaneous change in the system and a thermodynamic phase transition. Solve Exc. 15.3.1.2.
2.0
1.5
E / NkBTo
0.0
1.0
-0.2
∆
0.5 -0.4
0.0
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
T / To(N)
(3)
Figure 15.4: Left: Condensed fraction Nc /N as a function of reduced temperature T /Tc . The
circles represent measured values. The upper solid line shows the theory for an ideal gas,
the intermediate solid line shows a curve fitted to the measurements, and the lower part is
a theoretical curve taking into account finite size effects and interatomic interactions. Right:
Measurement of the release energy [82].
For a confined gas, volume and temperature are interdependent, and the concept of pressure
is somewhat vague. In this case, we can not refer to the heat capacity at constant volume or
pressure. However, one can define the heat capacity for a fixed number of particles,
∂E(T )
C(T ) = . (15.21)
∂T
438 CHAPTER 15. THERMODYNAMICS OF QUANTUM GASES
×10 5
6 2
(μK)
0
4
Nc
-2
μ/kB
2
-4
0 -6
0 0.5 1 1.5 0 0.5 1 1.5
T/Tc T/Tc
10 15
(μK)
10
C/N kB
5
E/kB
0 0
0 0.5 1 1.5 0 0.5 1 1.5
T/Tc T/Tc
Calculating the second moments of the distributions obtained for the same density by time-
of-flight of absorption images, we obtain the kinetic energy,
Z
p2
U= n(p)d3 p . (15.24)
2m
For confined ideal gases, the virial theorem ensures Ekin + Epot = 2Ekin . For real gases, the
repulsive energy of the mean field adds to this energy, E = Ekin + Epot + Eself . The sudden
extinction of the trapping potential before time-of-flight takes away the potential energy Epot
non-adiabatically. The kinetic energy and the self-energy of the condensate are fully converted
into kinetic energy during ballistic expansion. It is this energy, p2 /2m = Ekin + Eself , which is
sometimes called release energy, which is measured after ballistic expansion 4 . Fig. 15.4(right)
shows a measurement of the release energy. Solve the Exc. 15.3.1.3.
4
It is interesting to measure the heat capacity of a partially condensed cloud near the critical point and analyze
the discontinuity, because it contains important information about interatomic interactions and finite-size effects
([50], Seç. 3.4). In addition, the classification of Bose-Einstein condensation as a phase transition depends very
much on the behavior of the thermodynamic potential near the critical point [157, 131].
15.1. QUANTUM STATISTICS OF AN IDEAL BOSE GAS 439
1
n(x) = g3/2 (e−β[U (x)−µ] )
λ3dB
bosonic distribution functions . (15.25)
a6trap 2
n(k) = 3 g3/2 (eβ(µ−p /2m) )
λdB
In the classical limit, we can calibrate the chemical potential by Eq. (15.14) for a box potential
or by (15.17) for a harmonic potential,
( N 3
βµ βµ βµ V λth 3
for a box potential
g3/2 (e ) → c3/2 (e ) = e = ~ω̄ . (15.26)
c3 (eβµ ) =N kB T for a harmonic potential
We see that we recover the Maxwell-Boltzmann velocity distribution, as seen in Fig. 15.6,
r 3
m3 m 2
n(v) = n(k) 3 = N e−βmv /2 . (15.29)
~ 2πkB T
440 CHAPTER 15. THERMODYNAMICS OF QUANTUM GASES
(cm!3 )
(cm!3 ) 6 1.5
4 1
n(p)
n(x)
2 0.5
0 0
-50 0 50 -50 0 50
x=atrp atrp p=7h
Figure 15.6: (a) Density and (b) momentum distribution of a Bose gas (red) and a Boltzmann
gas (green) at T = 1.1Tc (solid line) and at T = 2Tc (dotted line).
where we defined,
kB T ttof
rrms ≡ . (15.31)
m
This distribution does not directly depend on the potential U (r), that is, the expansion is
isotropic. However, the chemical potential does depend on the potential. For very long flight
times (usually several 10 ms) the density resembles a Gaussian distribution [50]. In Exc. 15.3.1.4(b)
we determine the time-of-flight density distribution of an ultracold Bose gas.
In a time-of-flight experiment, any deviation observed between the results (15.30) and (15.31)
points towards an impact of quantum statistics. However, absorption images only record pro-
jections of the time-of-flight distribution on a plane.
integer spin are fermions. At high phase space densities, atoms have to figure out how they
will organize their coexistence. Bosons encourage each other to occupy the same phase space
cell, in contrast to the reluctant fermions, which prefer to follow Pauli’s exclusion principle. The
different behavior is described by different quantum statistics that determine how the phase space
(i.e., the available energy levels) has to be filled by the atoms. The Bose-Einstein distribution is
valid for bosons, the distribution of Fermi-Dirac for fermions and both asymptotically approach
the Boltzmann distribution at high temperatures. We have seen that bosons undergo a phase
transition and condense in the ground state when the temperature is reduced below a critical
threshold. On the other hand, the fermions organize their phase space, so that their energy levels
are arranged like a ladder. The impact of fermionic quantum statistics on a cold cloud of atoms
were observed experimentally by DeMarco and Jin [69, 195]. They cooled a two-components
Fermi gas of 7 × 105 potassium atoms down to 300 nK, which corresponded to 60% of the atoms
populating energy levels below the Fermi energy. The measured density distribution was found
to deviate from the one expected for an ideal Boltzmann gas 5 .
In the last line, we inserted the density of the states into a harmonic potential (14.21) and used
the definition of the Fermi functions (15.11).
For low temperatures, x ≡ µ 1, we can use the Sommerfeld expansion which in first order
gives fη (ex ) ' xη /Γ(η + 1). From this we immediately obtain the energy, the Fermi radius and
the momentum of free particles,
5
We note that meanwhile ultracold two-components Fermi gas have been demonstrated to form bosonic Cooper-
pairs, similarly to the phenomena known as superconductivity in some metals and as superfluidity of the fermionic
3
He.
442 CHAPTER 15. THERMODYNAMICS OF QUANTUM GASES
Z→1
In the limit of high temperatures, fη (Z) −→ Z
3 3
kB T βµ kB T
N= e = (1 + βµ + ...) , (15.35)
~ω̄ ~ω̄
3 3
~ω̄ 1 EF
µ ≈ kB T ln N = kB T ln .
kB T 6 kB T
This means that we recover the Boltzmann gas, which satisfies an ideal gas equation similar to
that of particles in a box potential,
3
kB T
N= (Boltzmann) . (15.36)
~ω̄
15.2.2 Energy
R
The total energy E1 ≡ E/N ≡ N −1 wd3 xd3 k per particle is given by,
R
wT,µ (x, k)d3 xd3 k
E1 = R (15.38)
wT,µ (x, k)d3 xd3 k
R −1
η() eβ(−µ) + 1 dε f4 (Z)
= R −1 = 3kB T .
η() eβ(−µ) + 1 dε f3 (Z)
Again using the Sommerfeld approximation, we see that for T → 0 the energy is limited by
3
E1 = f4 (eβµ ) (15.39)
N β(β~ω̄)3
3µ4 2π 2 T →0 3
= 3 1+ 2
+ ... −→ EF (Fermi) .
4EF (βµ) 4
because the atoms are forced to adopt a state of larger dynamics in the outermost regions of
T →0
the trap. This implies, E1 /EF −→ 3/4.
In the limit of high temperatures, T → ∞, a classical gas has the energy per particle,
3
3 −1 (βEF )
E1 = f4 f3 ≈ 3kB T (Boltzmann) , (15.40)
N β(β~ω̄)3 6
Z→0
which is seen by taking the high temperature limit fη (Z) −→ Z and extrapolating to all Z.
T →∞
This implies, E1 /EF −→ 3kB T /EF .
Comparing with bosons,
g4 (1)
E1 = 3kB T ≈ 2.7 (Bose) . (15.41)
g3 (1)
15.2. QUANTUM STATISTICS OF AN IDEAL FERMI GAS 443
f4 (Z) µ 4E1 µ
− =
S1 = 4kB − . (15.42)
f3 (Z) T 3T T
1
The heat capacity per particle C1 = ∂E 0
∂T is easily calculated using Zfη (Z) = fη−1 (Z),
N
f4 (Z) 3µ f4 (Z)f2 (Z) E1 3µ f4 (Z)f2 (Z)
C1 = 3kB − 1− = − 1− . (15.43)
f3 (Z) T f3 (Z)2 T T f3 (Z)2
For fermions well below the Fermi temperature, T → 0, using the Sommerfeld approximation,
we calculate,
2
T →0 3π kB T
C1 −→ (Fermi) . (15.44)
2 TF
For high temperature T
(a) (b) 6
2
5
(7K)
(7K)
4
0
3
E=kB
7=kB
-2
2
1
-4
0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
T=TF T=TF
#10 6
(c) 10 (d) 10
8 8
C=N kB
6 6
C=kB
4 4
2 2
0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
T=TF T=TF
such that,
n(x) = λ−3
dB f3/2 (e
−β[U (x)−µ]
) (Fermi) . (15.47)
At low temperatures, T → 0, we can apply the Sommerfeld expansion [39], which to first order
gives µ → EF ,
3/2
1 Γ(3/2) 2m
n(x) ≈ [µ − U (x)] (15.48)
(2π)2 Γ(5/2) ~2
3/2
1 2 2m 3/2 m 2 2 3/2 8λ N ρ2
= EF − ωr ρ = 2 3 1− 2 .
(2π)2 3 ~2 2 π RF RF
At high temperatures, T → ∞, we should recover the Boltzmann gas situation,
n(x) = λ−3
dB f3/2 (e
−β[U (x)−µ]
) (15.49)
2 3/2
3 −βU (x) mβ ω̄ 2 2 +ω 2 y 2 +ω 2 z 2 )/2
≈ λ−3
dB N (β~ω̄) e = N e−βm(ωx x y z .
2π
R
It’s easy to check, n(x)d3 x = N . Introducing the peak density n0 , we obtain,
2 ρ2 /2k
n(x) = n0 e−mω BT
(Boltzmann) . (15.50)
q
The rms-radius of the distribution is σj = kB T /mωj2 , which seems contrary to the above
D E
results, m ω
2 j
2 x2 = k T . In comparison,
j B
h i
n(x) = λ−3 g
dB 3/2 e β(µ−U (x))
(Bose gas above Tc ) . (15.51)
p p
where λdB = 2π~2 /mkB T e atr = ~/mω̄.
such that,
ñ(k) = λ−3 6
dB atr f3/2 (e
β(µ−ε)
) (Fermi) . (15.54)
At low temperatures, T → 0,
3/2
1 2 Γ(3/2)
ñ(k) ≈ (β [µ − ε])3/2 (15.55)
(2π)2 βmωr2 Γ(5/2)
3/2 3/2 3/2
1 2 2 ~2 k 2 8 N k2
≈ EF − = 1 − .
(2π)2 mωr2 3 2m π 2 KF3 KF2
Z ∞ Z ∞ Z Z 3/2
8 N k2
ñT →0 (kz ) = ñcl (k)dkx dky = 2 3 1− 2 dkx dky (15.56)
−∞ −∞ π KF |k|≤KF KF
Z 2π Z √K 2 −kz2 !3/2 5/2
8 N F kz2 kρ2 16 N kz2
= 2 3 1− 2 − 2 kρ dkρ dφ = 1− 2 .
π KF 0 0 KF KF 5π KF KF
R∞
It is easy to check −∞ ñT →0 dkz = N , with Maple.
At high temperatures, T → ∞, we should recover the Boltzmann gas situation,
3/2
~2 ω̄ 2
ñ(k) ≈ N e−βε (Boltzmann) . (15.57)
2πmωr2
√
Since ε is the kinetic energy, the rms-radius k 2 of this distribution is β~2 hk 2 i = m. In
comparison,
h i
β (µ−p2 /2m)
ñB (k) = λ−3
dB a6
tr g3/2 e (Bose gas above Tc ) . (15.58)
3/2 Z ∞
1 2 2m 1 ∞ 2 1
dr̃
= 2 2
r̃ ln −β~ 2 k 2 /2m+βµ−r̃ 2 −k 2
π βmω̃tr β~ 2 0 1+e z ρ
0
1/2 Z ∞
2 2 2 2 2
= 2 2 r̃2 ln 1 + eβµ−β~ kz /2m−r̃ dr̃ .
π (β~ω̃tr ) βmω̃ tr 0
446 CHAPTER 15. THERMODYNAMICS OF QUANTUM GASES
~2 kj2 /2m + 21 mωj2 (xj + ~kj t/m)2 = ~2 kj2 /2m(1 + ωj2 t2 ) + ωj2 txj ~kj + 21 mωj2 x2j (15.61)
s √ 2
~2 kj2 ωj2 txj 2m mωj2 x2j
= 2
(1 + ωj t2 ) + q +
2m 2 1 + ωj2 t2 2(1 + ωj2 t2 )
m
= ξj + ω̌j2 x2j .
2
r
−1/2 2
where we define ω̌i ≡ ωi 1 + ωi t 2 2 . With the substitution dξj = dkj 2~ m j ξ 1 + ω 2 t2 we
j
obtain
3/2 Z
1 mkB T 1 β 3/2 (ξx ξy ξz )−1/2 dξx dξy dξz
nT oF (x, t) = Q 2 2
(15.62)
(2π)3 2~2 i 1 + ωi t eβΣj [ξj + 2 ω̌j xj ] /Z + 1
m 2 2
Z
1 1 ω̃ 3 β 3/2 ξ −3/2 4πξ 2 dξ
= ,
23 π 3/2 λ3dB ω̄ 3 eβΣj [ξ+ 2 ω̌j xj ] /Z + 1
m 2 2
1 ω̌ 3
βµ− 12 βmΣj ω̌j2 x2j
nT oF (x, t) = f3/2 e . (15.63)
λ3dB ω̄ 3
At low temperatures,
m 3 N 8 (mx/~t)2
3/2
nT oF (x, t) = 1− (15.65)
~t KF3 π 2 KF2
!2 3/2
m 3 R 3 R mx/~t
F F
= 1−
~t 6π 2 λ (48N λ)1/3
15.2. QUANTUM STATISTICS OF AN IDEAL FERMI GAS 447
# 1014
2
1.5
(cm!3 )
1
n
0.5
0
0 50 100 150 200
r (7m)
At high temperatures,
1 1 2 2
nT oF (x, t) = f3/2 (eβ(µ−mx /2t ) ) (15.66)
λ3dB ω̄ 2 t2
1 1 β(µ−mx2 /2t2 )
≈ 3 e
λdB ω̄ 2 t2
3/2
mkB T 3/2 1 ~ω̄ 3 −βmx2 /2t2 ω̄ m 2 2
≈ 2 2 2
N e ≈N 2 e−βmx /2t .
2π~ ω̄ t kB T t 2πkB T
A rms-width is,
Z
2
rT oF = r2 nT oF (x, t)d3 x (15.67)
Z
1 ω̌ 3 2 βµ− 12 βmΣj ω̌j2 x2j
= 3 3
r f3/2 e d3 x
λdB ω̄
Z
2 εg(ε)dε kB T g4 (Z)
= 2
= .
mω̌r N e β(ε−µ) +1 mω̌r2 g3 (Z)
This shows that the width of the flight-of-time
√ distribution can simply be obtained from the spa-
tial distribution by substituting ω → ω/ 1 + ω 2 t2 . Of course this does not hold for condensed
gases Bose.
Z
16 u4 v 2 dudv
=
πN β 4 (~ω)3 eu2 +v2 /Z + 1
Z R 2 2
1 ~2 k 2 d3 xd3 k ~ k wT,µ (x, k)d3 xd3 k
= 2 2 2 2 = R = Ekin,1 .
(2π)3 N 2m eβ[~ k /2m+mω r /2−µ] + 1 2m wT,µ (x, k)d3 xd3 k
448 CHAPTER 15. THERMODYNAMICS OF QUANTUM GASES
This confirms the equipartition theorem for confined particles, which postulates,
3/2 3 s 3
1 ~ωtr ~2 2 2
ñT →∞ (k) = 2 N e−βε = N e−~ k /2mkB T . (15.86)
2πβmωtr kB T 2πmkB T
hx̃2tof i
N f4 (Z̃)
CT (N ) = = (15.89)
hx2tof i
Nnorm f4 (Z)
N f4 f3−1 (Nnorm (β~ω̄)3 )
= .
Nnorm f4 f3−1 (N (β~ω̄)3 )
Whether an atom is a fermion or a boson uniquely depends on its total spin. Halfinteger spin
particles are fermions, integer spin particles are bosons. E.g. Rb atoms have in the ground state
J = 1/2, I = 7/2, integer F , and are therefore bosons. Ca+ ions have J = 1/2 and no hyperfine
structure so that F is halfinteger, and are therefore fermions. 6 Li has half-integer F and is a
boson.
For a composite particle the quantum statistical nature may depend on the interaction
strength of the partners. For weak interaction, e.g. Feshbach the total spins of the partners
will couple to a total total spin, which determines the nature of the composite particle. A
fermion pairing with a fermion or a boson pairing with a boson will be bosons. A fermion
pairing with a boson will be a fermion. Composite trimers may be either bosonic or fermionic
depending on the coupling scheme. Can the quantum nature change with the tightness of the
binding? What is the total spin of a deeply bound molecule? [213, 33, 94], [?, 127, 206, ?]
kF3
nloc = (15.90)
3π 2
Z Z RF 3/2
8σ N x2 + y 2 z2
σndy = 2 3 1− − 2 dy (15.91)
π RF −RF RF2 ZF
3/2 Z RF 3/2
8σ N x2 z2 y2
= 2 3 1− 2 − 2 1− 2 dy .
π RF RF ZF −RF RF − x2 − RF2 z 2 /ZF2
452 CHAPTER 15. THERMODYNAMICS OF QUANTUM GASES
q
Writing a = RF / RF2 − x2 − RF2 z 2 /ZF2 ,
Z Z a
8σ N
σndy = 2 2 4 (1 − ỹ 2 )3/2 dỹ (15.92)
π RF a −a
2σ N p p
= 2 2 4 9a 1 − a2 − 2a3 1 − a2 + 3 arcsin a .
π RF a
In the center, a = 1,
Z
3N σ 9mω 2 N
σndy = 2 = 2 r , (15.93)
πRF kL EF
such that for EF ' 1µK we expect nloc ' 4 × 1012 cm−3 .
is independent on the atomic mass. This means that a Li and a Rb cloud in the same harmonic
trap at the same temperature have the same radius. This ensures good overlap. E.g. at T =
10 µK assuming the Rb secular frequencies ωr ' 2π × 300 Hz and ωz ' 2π × 30 Hz, we expect
rrms = 16 µm and zrms = 160 µm. However below the temperature 0.5TF , which is TF ' 1 µK
for NF = 104 , the quantum pressure stops the reduction of the fermion cloud while cooling.
This evtl. reduces the overlap with the boson cloud, disconnects the two clouds and stops the
evaporative cooling. On the other hand, the interaction energy of the boson cloud also increases
its size, when the Rb cloud approaches the critical temperature Tc ' 0.6 µK for NB = 106 .
The Pauli blocking of sympathetic cooling is a signature for the advent of quantum statistics
[70, 102, 196]. It is due to a reduced mobility (or better reduced available phase space at
collisions) of the atoms and not to be confused with the prohibition of s-wave collisions due to
the Pauli exlusion principle. Furthermore, elastic collisions are suppressed [69], because atoms
cannot be scattered into occupied trap levels [130, 244, 108, 109].
Example 59 (Estimations
q for 1D): Let qus consider a Fermi gas in a very elongated
87 87 5
microtrap: ωr = 7 2π × 1.4 kHz and ωz = 7 2π × 15 Hz for Rb. With NLi = 10 the
Fermi temperature is as high as TF ' 5 µK. However we need N 100 to see 1D features.
2
1 2
Assume ε = 2m p + m4 b4 r4 ,
n
1 Γ +1
N= 4
n
(kB T )3n/4 f3n/4 (z)
(~b)n Γ 2 +1
!4/3n
Γ n2 + 1 Γ( 3n4 + 1)
EF ≈ (~b)4/3 N .
Γ n4 + 1
In 1D,
1.02
N= (kB T )3/4 f3/4 (z)
~b
EF ≈ 0.87(N ~b)4/3 .
15.3 Exercises
15.3.1 Quantum statistics of Bose-condensates
15.3.1.1 Ex: Boson or fermion?
Whether an atom is a fermion or boson solely depends on its total spin. Half-integer spin
particles are fermions, integer spin particles are bosons. For example, Rb atoms have in the
ground state J = 1/2, I = 7/2 and F integer, and therefore are bosons. Ca+ ions have J = 1/2
and no hyperfine structure, and thus are fermions. 6 Li has the half-integer F and is a boson.
Decide on the bosonic or fermionic nature of the following atoms/molecules:
85 Rb with I = 3/2 in the state 2 S
1/2
88 Sr with I = 0 in the state 1 S
0
88 Sr with I = 0 in the state 3 P
2
87 Sr with I = 9/2 in the state 1 S
0
172 Yb+ with I = 0 in the state 2 S
1/2
171 Yb+ with I = 1/2 in the state 2 S
1/2
15.3. EXERCISES 455
[2] Y. Aharonov and J. Anandan, Phase change during cyclic evolution, Phys. Rev. Lett. 58
(1987), 1593, .
[3] R. Alicki, The markov master equations and the fermi golden rule, International Journal
of Theoretical Physics (1977).
[5] B. P. Anderson and M. A. Kasevich, Macroscopic quantum interference from atomic tunnel
arrays, Science 282 (1998), 1686, .
[10] N. W. Ashkroft and D. N. Mermin, Solid state physics, Hartcourt College Publishers, 1976.
[11] P. W. Atkins and R. S. Friedman, Molecular quantum mechanics, Oxford University Press,
1997.
[12] R. Bachelard, N. Piovella, and Ph. W. Courteille, Cooperative scattering and radiation
pressure force in dense atomic clouds, Phys. Rev. A 84 (2011), 013821.
[15] Vanderlei S. Bagnato, Simple calculation of the spatial evolution of atoms in a trap during
occurance of bose-einstein condensation, Phys. Rev. A 54 (1996), 1726.
457
458 BIBLIOGRAPHY
[17] A. Bambini and S. Geltman, Feshbach resonances in cold collisions of potassium atoms,
Phys. Rev. A 65 (2002), 062704, .
[18] A. Barchielli, L. Lanz, and G. M. Prosperi, A model for the macroscopic description and
continual observations in quantum mechanics, Il Nuovo Cimento 72 (1982), 79.
[22] A. Beige and G. C. Hegerfeldt, Projection postulate and atomic quantum zeno effect, Phys.
Rev. A 53 (1996), 53.
[25] C. M. Bender, Brody, and P. N. Jones, Complex extension of quantum mechanics, Phys.
Rev. Lett. 89 (2002), 270401, .
[26] H. Bender, Ph.W. Courteille, C. Marzok, C. Zimmermann, and S. Slama, Direct mea-
surement of intermediate-range casimir-polder potentials, Phys. Rev. Lett. 104 (2010),
083201.
[28] M. Blatt, K.W. Boër, and W. Brandt, Bose-einstein condensation of excitons, Phys. Rev.
126 (1962), 1691.
[30] I. Bloch, T. W. Hänsch, and T. Esslinger, Atom laser with a cw output coupler, Phys. Rev.
Lett. 82 (1999), 3008, .
[31] C. A. Blockley, D. F. Walls, and H. Risken, Quantum collapses and revivals in a quantized
trap, Europhys. Lett. 17 (1992), 509.
[32] Satyendra N. Bose, Plancks gesetz und lichtquantenhypothese, Z. Phys. 26 (1924), 178.
[37] Léon Brillouin, La mécanique ondulatoire de schrödinger: une méthode générale de reso-
lution par approximations successives, Comptes Rendus de l’Academie des Sciences 183
(1926), 24.
[38] I. N. Bronstein and K. A. Semandjajew, Taschenbuch der mathematik, Thun und Franfurt,
Main, 1978.
[39] D. A. Butts and D. S. Rokhsar, Trapped fermi gases, Phys. Rev. A 55 (1997), 4346, .
[40] S. Bux, Ch. Gnahm, R. A. Maier, C. Zimmermann, and Ph. W. Courteille, Cavity-
controlled matter wave superradiance at the recoil limit, Phys. Rev. Lett. 106 (2011),
203601.
[41] Y. Castin and R. Dum, Bose-einstein condensates in time dependent traps, Phys. Rev.
Lett. 77 (1996), 5315, .
[45] Steve Chu, J. E. Bjorkholm, A. Ashkin, and A. Cable, Experimental observation of opti-
cally trapped atoms, Phys. Rev. Lett. 57 (1986), 314.
[46] J. I. Cirac, R. Blatt, P. Zoller, and W. D. Phillips, Laser cooling of trapped ions in a
standing wave, Phys. Rev. A 46 (1992), 2668, .
[47] J. I. Cirac, M. Lewenstein, and P. Zoller, Quantum dynamics of a laser-cooled ideal gas,
Phys. Rev. A 50 (1994), 3409, .
[48] C. Cohen-Tannoudji, B. Diu, and F. Lao, Quantum mechanics, 1, vol. 1 & 2, 1977, Dedalus:
530.12 C678q v.1 e.1.
[51] Ph. W. Courteille, S. Bux, E. Lucioni, K. Lauber, T. Bienaimé, R. Kaiser, and N. Piovella,
Modification of radiation pressure due to cooperative scattering of light, Euro. Phys. J. D
58 (2010), 69.
[52] Ph. W. Courteille, B. Deh, J. Fortágh, A. Günther, S. Kraft, C. Marzok, S. Slama, and
C. Zimmermann, Highly versatile atomic micro traps generated by multifrequency magnetic
field modulation, J. Phys. B 39 (2006), 1055.
[54] Ph. W. Courteille, D. J. Han, R. H. Wynar, and D. J. Heinzen, New observation of bose-
einstein condensation of 87 rb atoms in a magnetic top trap, Proc. of SPIE 3270 (1998),
116.
[57] M. Ben Dahan, E. Peik, Y. Castin, and C. Salomon, Bloch oscillations of atoms in an
optical potential, Phys. Rev. Lett. 76 (1996), 4508, .
[61] , Laser cooling below the doppler limit by polarization gradients: Simple theoretical
studies, J. Opt. Soc. Am. B 6 (1989), 2023.
[62] N. Davidson, H. J. Lee, C. S. Adams, M. Kasevich, and S. Chu, Long atomic coherence
times in an optical dipole trap, Phys. Rev. Lett. 74 (1995).
[63] 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 an gas of sodium atoms, Phys. Rev. Lett.
75 (1995), 3969, .
[65] K. B. Davis, M.-O. Mewes, and W. Ketterle, An analytical model for evaporative cooling
of atoms, Appl. Phys. B 60 (1995), 155, .
[66] S. R. de Groot, G. J. Hooyman, and C. A. Ten Deldam, Proc. R. Soc. Lond. A 203 (1950),
266.
BIBLIOGRAPHY 461
[69] B. DeMarco and D. S. Jin, Onset of fermi-degeneracy in a trapped atomic gas, Science
285 (1999), 1703, .
[70] B. DeMarco, S. B. Papp, and D. S. Jin, Pauli blocking of collisions in a quantum degenerate
atomic fermi gas, Phys. Rev. Lett. 86 (2001), 5409, .
[71] W. Demtröder, Atoms, molecules and photons: An introduction to atomic- molecular- and
quantum physics, Springer, 2006.
[74] R. H. Dicke, Coherence in spontaneous radiation processes, Phys. Rev. 93 (1954), 99.
[75] F. Diedrich, J.C. Bergquist, W.I. Itano, and D.J. Wineland, Laser cooling to the zero-point
energy of motion, Phys. Rev. Lett. 62 (1989), 403, .
[76] P. A. M. Dirac, Quantized singularities in the electromagnetic field, Proc. Roy. Soc. A 133
(1931), 60.
[78] J. M. Doyle, B. Friederich, J. Kim, and D. Patterson, Buffer-gas loading of atoms and
molecules into a magnetic trap, Phys. Rev. A 52 (1995), .
[80] A. Einstein, Quantentheorie des einatomigen idealen gases, S. B. Kgl. Preuss. Akad. Wiss.
35 (1924).
[84] W. Ertmer, R. Blatt, J. L. Hall, and M. Zhu, Laser manipulation of atomic beam velocities:
Demonstration of stopped atoms and velocity reversal, Phys. Rev. Lett. 54 (1985), 996.
462 BIBLIOGRAPHY
[88] P. T. Fisk, M. A. Lawn, and C. Coles, Laser cooling of 171yb+ ions a linear paul trap,
Appl. Phys. B 57 (1993), 287.
[94] S. Giovanazzi, A. Görlitz, and T. Pfau, Tuning the dipolar interaction in quantum gases,
Phys. Rev. Lett. 89 (2002), 130401, .
[95] R. J. Glauber, Coherent and incoherent states of the radiation field, Phys. Rev. 131 (1963),
2766, .
[96] , The quantum theory of optical coherence, Phys. Rev. 130 (1963), 2529.
[97] J. P. Gordon and A. Ashkin, Motion of atoms in a radiation trap, Phys. Rev. A 21 (1980),
1606, .
[98] R. Graham and D. Walls, Spectrum of light scattered from a weakly interacting bose-
einstein condensed gas, Phys. Rev. Lett. 76 (1996), 1774.
[99] M. Greiner, O. Mandel, T. Esslinger, T. W. Hänsch, and I. Bloch, Quantum phase transi-
tion from a superfluid to a mott insulator in a gas of ultracold atoms, Nature 415 (2002),
39, .
[101] , Probing the excitation spectrum of a fermi gas in the BCS-BEC crossover regime,
Phys. Rev. Lett. 94 (2005), 070403, .
[103] David J. Griffiths, Introduction to quantum mechanics, Pearson Prentice Hall, New Jersey,
2005.
BIBLIOGRAPHY 463
[104] R. Grimm, M. Weidemüller, and Y. B. Ovchinnikov, Optical dipole traps for neutral atoms,
Adv. At. Mol. Opt. Phys. 42 (2000), 95, .
[105] S. Grossmann and M. Holthaus, On bose-einstein condensation in harmonic traps, Phys.
Lett. A 208 (1995), 188, .
[106] G. Grynberg, B. Lounis, P. Verkerk, J.-Y. Courtois, and C. Salomon, Quantized motion
of cold cesium atoms in two- and three-dimensional optical potentials, Phys. Rev. Lett. 70
(1993), 2249, .
[107] D. Guéry-Odelin, J. Söding, P. Desbiolles, and J. Dalibard, Is bose-einstein condensatiopn
of atomic cesium possible?, Europhys. Lett. 44 (1998), 25.
[108] S. Gupta, Z. Hadzibabic, M. W. Zwierlein, B. J. Verhaar, and W. Ketterle, Radio-frequency
spectroscopy of ultracold fermions, Sciencexpress (2003), 1, .
[109] Z. Hadzibabic, S. Gupta, C. A. Stan, C. H. Schunck, M. W. Zwierlein, K. Dieckmann,
and W. Ketterle, Fiftyfold improvement in the number of quantum degenerate fermionic
atoms, Phys. Rev. Lett. 91 (2003), 160401, .
[110] E. W. Hagley, L. Deng, M. Kozuma, J. Wen, K. Helmerson, S. L. Rolston, and W. D.
Phillips, A well-collimated quasi-continuous atom laser, Science 283 (1999), 1706, .
[111] D. S. Hall, J. R. Ensher, D. S. Jin, and et al., Recent experiments with bose-condensed
gases at JILA, Proc. SPIE 3270 (1998), 98, cond-mat/9903459.
[112] D. S. Hall, M. R. Matthews, J. R. Ensher, C. E. Wieman, and E. A. Cornell, Dynamics of
component separation in a binary mixture of bose-einstein condensates, Phys. Rev. Lett.
81 (1998), 1539, .
[113] D. J. Han, R. H. Wynar, Ph. W. Courteille, and D. J. Heinzen, Bose-einstein condensation
of large numbers of atoms in a magnetic time-averaged orbiting potential trap, Phys. Rev.
A 57 (1998), R4114.
[114] S. Haroche, J. C. Gay, and G. Grynberg (eds.), Atom traps, World Scientific, 1989.
[115] T. Hartmann, F. Keck, H. J. Korsch, and S. Mossmann, Dynamics of bloch oscillations,
New J. Phys. 6 (2004), 2.
[116] L. V. Hau, B. D. Busch, Ch. Liu, Z. Dutton, M. M. Burns, and J. A. Golovchenko, Near-
resonant spatial images of a confined bose-einstein condensates in a 4-dee magnetic bottle,
Phys. Rev. A 58 (1998), R54, .
[117] C. E. Hecht, The possible superfluid behaviour of hydrogen atom gases and liquids, Physica
25 (1959), 1159.
[118] K. Helmerson, A. Martin, and D. E. Pritchard, Laser and rf spectroscopy of magnetically
trapped neutral atoms, J. Opt. Soc. Am. B 9 (1992), 483.
[119] A. Hemmerich, Phys. Rev. Lett. 68 (1992).
[120] , Phys. Rev. Lett. 72 (1994).
[121] A. Hemmerich and T. W. Hänsch, Two-dimensional atomic crystal bound by light, Phys.
Rev. Lett. 70 (1993).
464 BIBLIOGRAPHY
[123] J. Herbig, T. Kraemer, M. Mark, T. Weber C. Chin, H.-C. Nägerl, and R. Grimm, Prepa-
ration of a pure molecular quantum gas, Science 301 (2003), 1510, .
[124] C. Herzog and M. Olshanii, Trapped bose gas: The canonical versus grand canonical statis-
tics, Phys. Rev. A 55 (1997), 3254, .
[125] H. F. Hess, Evaporative cooling of magnetically trapped spin-polarized hydrogen, Phys. Rev.
B 34 (1986), 3476.
[126] T. W. Hijmans, Yu. Kagan, and G. V. Shlyapnikov, Bose condensation and relaxation
explosion in magnetically trapped atomic hydrogen, Phys. Rev. B 48 (1993), 12886, .
[127] Tin-Lun Ho and E. J. Mueller, High temperature expansion applied to fermions near fes-
hbach resonance, Phys. Rev. Lett. 92 (2004), 160404, .
[132] R. Huesmann, Ch. Balzer, Ph. W. Courteille, W. Neuhauser, and P. E. Toschek, Single-
atom interferometry, Phys. Rev. Lett. 82 (1999), 1611.
[137] W. M. Itano, J. J. Bollinger, and D. J. Wineland, Quantum zeno effect, Phys. Rev. A 41
(1990), 2295, .
[138] W. M. Itano and D. J. Wineland, Laser cooling of ions stored in harmonic and penning
traps, Phys. Rev. A 25 (1982), 35, .
BIBLIOGRAPHY 465
[140] J. Javanainen and J. Ruostekoski, Off-resonance light scattering from low-temperature bose
and fermi gases, Phys. Rev. A 52 (1995), 3033, .
[145] , Evaporative cooling of trapped atoms, Adv. At. Mol. Opt. Phys. 37 (1996), 181.
[153] M. Kozuma, Y. Suzuki, Y. Torii, T. Saguira, T. Kaga, E. W. Hagley, and L. Deng, Phase-
coherent amplification of matter-waves, Science 286 (1999), 2309, .
[155] Hendrik A. Kramers, Wellenmechanik und halbzählige quantisierung, Zeitschrift für Physik
39 (1926), 828.
[156] A. Lakhtakia, Positive and negative goos-hänchen shifts and negative phase-velocity medi-
ums (alias left-handed materials), ePrints:Physics0305133 (2003).
[159] H. J. Lee, C. S. Adams, M. Kasevich, and S. Chu, Raman cooling of atoms in an optical
dipole trap, Phys. Rev. Lett. 76 (1996), 2658, .
[160] U. Leonhardt and H. Paul, Measuring the quantum state of light, Prog. Quant. Electr. 19
(1995), 89.
[161] Mingzhe Li, Zijun Yan, Jincan Chen, Lixuan Chen, and Chuanhong Chen, Thermodynamic
properties of an ideal fermi gas in an external potential with U = brt in any dimensional
space, Phys. Rev. A 58 (1998), 1445, .
[162] Jia Ling Lin and J. P. Wolfe, Bose-einstein condensation of paraexcitons in stressed Cu2O,
Phys. Rev. Lett. 71 (1993), 1222, .
[164] R. Loudon, The quantum theory of light, Clarendon Press Oxford, 1982.
[173] J. Meixner and F.W. Schäfke, Mathieu’sche funktionen und sphäroidfunktionen, Springer-
Verlag Berlin (1954).
[175] M.-O. Mewes, M. R. Andrews, N. J. van Druten, D. M. Kurn, D. S. Durfee, and W. Ket-
terle, Bose-einstein condensation in a tightly confining dc magnetic trap, Phys. Rev. Lett.
77 (1996), 416, .
[176] M.-O. Mewes, M. R. Andrews, N. J. van Druten, D. M. Kurn, D. S. Durfee, C. G.
Townsend, and W. Ketterle, Collective excitations of a bose-einstein condensation in a
magnetic trap, Phys. Rev. Lett. 77 (1996), 988, .
[177] A. L. Migdall, J. V. Prodan, W. D. Phillips, T. H. Bergman, and H. J. Metcalf, First
observation of magnetically trapped neutral atoms, Phys. Rev. Lett. 54 (1985), 2596.
[178] G. Milburn, Intrinsic decoherence in quantum mechanics, Phys. Rev. A 39 (1991), 5401,
.
[179] J. D. Miller, R. A. Cline, and D. J. Heinzen, Far-off-resonance optical trapping of atoms,
Phys. Rev. A 47 (1993), R4567, .
[180] Randell L. Mills, The grand unified theory of classical quantum mechanics.
[181] P. W. Milonni and R. W. Boyd, Momentum of light in a dielectric medium, Adv. Opt.
Phot. 2 (2010), 519, .
[182] B. Misra and E. C. G. Sudarshan, The zeno paradox in quantum theory, J. Math. Phys.
18 (1977), 756.
[183] B. R. Mollow, Pure-state analysis of resonant light scattering: Radiative damping, satura-
tion, and multiphoton effects, Phys. Rev. A 12 (1975), 1919.
[184] K. Mølmer, Y. Castin, and J. Dalibard, Monte-carlo wave-function method in quantum
optics, J. Opt. Soc. Am. B 10 (1993), 524, .
[185] C. Monroe, E. A. Cornell, C. A. Sackett, C. J. Myatt, and C. E. Wieman, Measurement
of cs-cs elastic scattering at t=30uk, Phys. Rev. Lett. 70 (1993), 414, .
[186] H. Moritz, T. Stöferle, K. Günter, M. Köhl, and T. Esslinger, Confinement induced
molecules in a 1d fermi gas, Phys. Rev. Lett. 94 (2005), 210401, .
[187] A. Mosk, S. Jochim, H. Moritz, Th. Elsässer, and M. Weidemüller, Resonator-enhanced
optical dipol trap for fermionic lithium atoms, Opt. Lett. 26 (2001), 1837.
[188] G. M. Moy, J. J. Hope, and C. M. Savage, Born and markov approximations for atom
lasers, Phys. Rev. A 59 (1999), 667.
[189] M. Mudrich, S. Kraft, K. Singer, R. Grimm, A. Mosk, and M. Weidemüller, Sympathetic
cooling with two atomic species in an optical trap, Phys. Rev. Lett. 88 (2002), 253001.
[190] C. J. Myatt, E. A. Burt, R. W. Ghrist, E. A. Cornell, and C. E. Wieman, Production
of two overlapping bose-einstein condensates by sympathetic cooling, Phys. Rev. Lett. 78
(1997), 586, .
[191] W. Nagourney, J. Sandberg, and H. G. Dehmelt, Shelved optical electron amplifier: Ob-
servation of quantum jumps, Phys. Rev. Lett. 56 (1986), 2797.
[192] R. Napolitano, J. De Luca, V. S. Bagnato, and G. C. Marques, Effect of finite numbers in
the bose-einstein condensation of a trapped gas, Phys. Rev. A 55 (1997), R3954, .
468 BIBLIOGRAPHY
[211] D. E. Pritchard, Cooling neutral atoms in a magnetic trap for precision spectroscopy, Phys.
Rev. Lett. 51 (1983), 1336, .
[212] Ravinder Puri, Mathematical methods of quantum optics, Springer Series in Optical Sci-
ences (2000).
[214] M. Reich, U. Sterr, and W. Ertmer, Scheme for measuring a berry-phase in an atom
interferometer, Phys. Rev. A 47 (1993), 2518, .
[221] Th. Sauter, R. Blatt, W. Neuhauser, and P.E. Toschek, Observation of quantum jumps,
Phys. Rev. Lett. 57 (1986), 1696, .
[222] , Quantum jumps observed in the fluorescence of a single ion, Opt. Comm. 60
(1986), 287, .
[224] A. Schenzle, R. G. DeVoe, and R. G. Brewer, Possibility of quantum jumps, Phys. Rev. A
33 (1986), 2127, .
[225] C. A. Schrama, E. Peik, W. W. Smith, and H. Walther, Novel miniature ion traps, Opt.
Comm. 101 (1993), 32.
470 BIBLIOGRAPHY
[226] M. Schubert, I. Siemers, and R. Blatt, Line shape of three-level ions in paul traps, Phys.
Rev. A 39 (1989), 5098.
[227] Jian Qi Shen, Negatively refracting atomic vapour, J. Mod. Opt. 53 (2006), 2195, .
[228] Jian Qi Shen, J. Almlöf, and S. He, Negative permeability in a λ-type three level atomic
vapor, Appl. Phys. A 87 (2006), 291, DOI.
[229] Jian-Qi Shen, Zhi-Chao Ruan, and Sailing He, How to realize a negative refractive index
material at the atomic level in an optical frequency range, Journal of Zhejiang University
SCIENCE 5 (2004), 1322, .
[231] S. Slama, S. Bux, G. Krenz, C. Zimmermann, and Ph. W. Courteille, Superradiant rayleigh
scattering and collective atomic recoil lasing in a ring cavity, Phys. Rev. Lett. 98 (2007),
053603.
[233] M. D. Srinivas and E. B. Davies, Photon counting probabilities in quantum optics, Optica
Acta 28 (1981), 981.
[234] D. M. Stamper-Kurn and W. Ketterle, Spinor condensates and light scattering from bose-
einstein condensates, Proc. Les Houches Summer School, Session LXXII (2000), .
[236] S. Stenholm, The semiclassical theory of laser cooling, Rev. Mod. Phys. 58 (1986), 699,
.
[237] D. Stoler, Equivalence class of minimum uncertainty wavepackets, Phys. Rev. D 1 (2006),
3217, .
[238] , Photon antibunching and possible ways to observe it, Phys. Rev. Lett. 33 (2006),
1397, .
[239] H. T. C. Stoof, Atomic bose gas with a negative scattering length, Phys. Rev. A 49 (1995),
3824, .
[240] William C. Stwalley and L. H. Nosanow, Possible new quantum systems, Phys. Rev. Lett.
36 (1976), 910.
[243] U. M. Titulaer and R. J. Glauber, Correlation functions for coherent fields, Phys. Rev.
140 (1999).
BIBLIOGRAPHY 471
[262] Deshui Yu, Photonic band structure of the three-dimensional 88 sr atomic lattice, Phys.
Rev. A 84 (2011), 043833, .
[263] P. Zoller, M. Marte, and D. F. Walls, Quantum jumps in atomic systems, Phys. Rev. A
35 (1987), 198, .
473
474 INDEX
intensity, 23 Landau-Zener
intensive parameter, 450 fórmula de, 413
interaction picture, 60, 143, 246, 247 Langevin equation, 252
intermediate coupling, 231 large component, 169
interval factor, 185 Larmor
interval rule, 186 frequência de, 408
inverse Abel transform, 425 Larmor frequency, 19
Ioffe laser, 307
potencial de, 408 Legendre
irreducible matrix element, 204 Adrien-Marie, 104
isotropy Legendre operator, 103
spatial, 68 Legendre polynomials, 104
lepton, 188
Jaynes-Cummings Leucippus, 11
modelo de, 287, 300 Lie algebra, 47
Josephson junction, 159 light shift, 345
kernel, 52 light-shift, 255, 263
ket, 43 Lindbladt
Klein-Gordon equation, 38, 168 operador de, 345, 365
Koopman’s theorem, 225 linear momentum space, 52
Kramers Liouville equation, 245
Hendrik Anthony, 135 Liouville operator, 245, 262
Kramers-Heisenberg formula, 145 localization energy, 76
Kraus Lorentz
operador de, 363 calibre de, 375
Kronig-Penney model, 162 Hendrik Antoon, 177
modelo de, 342
Laguerre Lorentz boost, 66
Edmond, 109 Lorentz transform, 66
Laguerre polynomials, 109 lowering operador, 113
Laguerre’s associated differential equation, LS-coupling, 231
109 Luttinger liquid, 454
Lamb
Willis Eugene, Jr., 183 Mößbauer
Lamb shift, 183 efeito, 399
Lamb-Dicke efeito de, 350, 351
parâmetro de, 350 macromovimento, 417
regime de, 350 magnética
Lamb-Dicke parameter, 398 armadilha, 340
Lamb-dip, 256 magnetic quantum number, 104
Lambert-Beer magneto-óptica
lei de, 422, 423 armadilha, 400
Lambert-Beer law, 33 main quantum number, 108
Landé Majorana
fator de, 339 spin-flip de, 408
Landé factor, 197, 199 Markov
Landau gauge, 202 aproximação de, 303
Landau level, 203 Markoviano
478 INDEX
vacuum fluctuation, 89
variational method, 134
vector space, 43, 47
vestido
estado, 283, 287
virial theorem, 111
Voigt profile, 259
von Neumann
equação de, 300
John, 46