0% found this document useful (0 votes)
295 views16 pages

Solution 11

1) The document is an exercise sheet for a physics course that examines relativistic corrections to the hydrogen atom energy levels and one-dimensional harmonic oscillator. 2) It shows that the lowest-order relativistic correction to the Hamiltonian is a term proportional to p^4, and uses perturbation theory to calculate the energy shift. 3) For the hydrogen atom, it expresses the relativistic correction to the energy levels in terms of the unperturbed energies and expectation values of 1/r and 1/r^2, lifting the degeneracy between states of different angular momentum l that previously had the same energy.

Uploaded by

denecity.y
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
295 views16 pages

Solution 11

1) The document is an exercise sheet for a physics course that examines relativistic corrections to the hydrogen atom energy levels and one-dimensional harmonic oscillator. 2) It shows that the lowest-order relativistic correction to the Hamiltonian is a term proportional to p^4, and uses perturbation theory to calculate the energy shift. 3) For the hydrogen atom, it expresses the relativistic correction to the energy levels in terms of the unperturbed energies and expectation values of 1/r and 1/r^2, lifting the degeneracy between states of different angular momentum l that previously had the same energy.

Uploaded by

denecity.y
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 16

Exercises Sheet 11

Prof. R. Grange Physics III, HS 2023


Handout: 4th December 2023
Submission: 14th December 2023

1 Relativistic corrections to hydrogen energy levels


Learning Objective: Hydrogen atom, perturbation theory, relativistic quantum mechanics

The basic Hamiltonian from the lecture,


p̂2
Ĥ0 = + V̂ , (1)
2m
reflects the non-relativistic kinetic energy. In this exercise you will examine the influence of an addi-
tional relativistic term on the lowest energy eigenstates of the hydrogen atom and one-dimensional
harmonic oscillator.

a) Show by expanding the relativistic kinetic energy Ekin , that the lowest-order relativistic
correction to the Hamiltonian is
p̂4
δ Ĥ = − 3 2 . (2)
8m c
√ 2
Hint: Use the expansion 1 + x ≈ 1 + x2 − x8 + . . .
b) In first-order perturbation theory, the correction δEn to the unperturbed energy En0 is given
by the expectation value ⟨δ Ĥ⟩ in the unperturbed state. Show the following general inter-
mediate result:
1 h 0 2 i
δEn = ⟨δ Ĥ⟩ = ψn0 δ Ĥ ψn0 = − 2
En − 2En0 ⟨V̂ ⟩ + ⟨V̂ 2 ⟩ , (3)
2mc
where Ĥ0 ψn0 = En0 ψn0 .
Hint: Note that p̂2 = 2m(Ĥ0 − V̂ ).
c) For the eigenstates of the non-relativistic (and spinless) Hydrogen atom,

p̂2 e2 −1
Ĥ0 = − r̂ , (4)
2m 4πϵ0
it is possible to derive the following relations,
1 1
⟨r̂−1 ⟩nlm = , ⟨r̂−2 ⟩nlm = . (5)
n2 a0 (l + 1/2)n3 a20

Use these to find the (lowest-order) relativistic correction δEnlm = ⟨δ Ĥ⟩nlm to the energy
levels of the hydrogen atom in terms of the unperturbed energies Enlm 0
. Formerly we had
that different eigenstates of hydrogen with the same principal quantum number n but with
different angular momentum l were degenerate, i.e., they had the same energy. What can
you say about the degeneracy of states when the relativistic correction is applied?
d) Calculate the (lowest-order) relativistic correction to the energy levels of the one-dimensional
harmonic oscillator, with Hamiltonian
p̂2 1
Ĥ0 = + mω 2 x̂2 . (6)
2m 2

e) Compute the relative correction δEn /En0 to the energy of the |Ψ210 ⟩ state of the hydrogen
atom and to the energy of the ground state of the one-dimensional harmonic oscillator to
justify the perturbation theory approach. For the harmonic oscillator consider an electron
oscillating at ω/(2π) = 500 MHz.

1
Exercises Sheet 11 Prof. R. Grange; HS 2023

Solution
a) The relativistic kinetic energy is given by the relativistic energy E 2 = m2 c4 + p2 c2 minus the
rest energy mc2 , thus p
Ekin = m2 c4 + c2 p2 − mc2 , (7)
where m is the particle’s rest mass. Since we know the expansion
√ x x2
1+x=1+ − + ... (8)
2 8
we can readily expand the kinetic energy to fourth order in p:
r !
p 2
Ekin = mc2 −1 + 1 + 2 2 , (9)
m c
p2 p4
 
2
≈ mc −1 + 1 + − + ... , (10)
2m2 c2 8m4 c4
p2 p4
= − + .... (11)
2m 8m3 c2
To get the corresponding quantum mechanical expression, we replace p → p̂. So the first
correction term to Ĥ0 is
p̂4
δ Ĥ = − 3 2 . (12)
8m c
b) First, note that the energy shift is given by using the relation
2 h  i2
p̂4 = p̂2 = 2m Ĥ0 − V̂ , (13)

and thus

δEn = ⟨δ Ĥ⟩ = ψn0 δ Ĥ ψn0 (14)


4
p̂ 1
= ψn0 − ψ 0 = − 3 2 ψn0 p̂4 ψn0 , (15)
8m2 c3 n 8m c
1 h i2
= − 3 2 ψn0 2m(Ĥ0 − V̂ ) ψn0 , (16)
8m c
1
=− ψ 0 Ĥ 2 − Ĥ0 V̂ − V̂ Ĥ0 + V̂ 2 ψn0 . (17)
2mc2 n 0

Remember that while Ĥ0 commutes with itself, it does not necessarily commute with with
its component parts, namely V̂ . Since ψn0 are the unperturbed eigenstates of Ĥ0 with
eigenenergies En0 , we have Ĥ0 ψn0 = En0 ψn0 , and the hermitian conjugate ψn0 Ĥ0 =
ψn0 En0 , (Ĥ0 is hermitian), which allows us to replace Ĥ0 with En0 in the expression, which
as a simple number, does commute with V̂ . Thus we can simplify further:
1  
δEn = − ψn0 Ĥ02 ψn0 − ψn0 Ĥ0 V̂ ψn0 − ψn0 V̂ Ĥ0 ψn0 + ψn0 V̂ 2 ψn0 , (18)
2mc2
1  2 
=− ψn0 En0
ψn0 − ψn0 En0 V̂ ψn0 − ψn0 V̂ En0 ψn0 + ψn0 V̂ 2 ψn0 , (19)
2mc2
1 h 2 i
=− En0 − En0 ψn0 V̂ ψn0 − En0 ψn0 V̂ ψn0 + ψn0 V̂ 2 ψn0 , (20)
2mc2
1 h 2 i
=− En0 − 2En0 ⟨V̂ ⟩ + ⟨V̂ 2 ⟩ . (21)
2mc2
This result holds for any Hamiltonian where we have added the perturbation to account
for the relativistic kinetic energy. We will now examine how this perturbation affects two
systems we know well, the hydrogen atom and the harmonic oscillator.

2
Exercises Sheet 11 Prof. R. Grange; HS 2023

c) For hydrogen, V̂ = −(e2 /4πϵ0 )r̂−1 . The lowest-order relativistic correction is then
"  2   2 2 #
1 2 e e
δEnlm = − 0
Enlm + 2Enlm 0 −1
⟨r̂ ⟩ + −2
⟨r̂ ⟩ , (22)
2mc2 4πϵ0 4πϵ0
"  2   2 2 #
1 2 e 1 e 1
0 0
(23)

=− Enlm + 2Enlm + ,
2mc2 4πϵ0 n2 a0 4πϵ0 (l + 1/2)n3 a20
where we used both the intermediate result found in b) and the expressions given in the
problem formulation.
Now we recall the unperturbed energies Enlm
0
for hydrogen,
"  2 2 #
m e 1 ℏ2
0
Enlm =− = − , (24)
2ℏ2 4πϵ0 n2 2ma20 n2
 ℏ2
and the definition of the Bohr radius a0 = 4πϵ
e2
0
m , and simplify the expression to
0
 2  
Enlm 4n
δEnlm = − −3 . (25)
2mc2 l + 1/2
We see that while formerly the unperturbed eigenenergies Enlm 0
only had a dependence on the
principal quantum number, and all states with the same n and different angular momentum
l were degenerate, now the relativistic correction lifts this degeneracy, as states with different
n and l have different energies.
d) For the one-dimensional harmonic oscillator the potential is V̂ = 21 mω 2 x̂2 and the eigenen-
ergies are En = (n + 12 )ℏω for eigenstates |n⟩. The lowest-order relativistic correction is
"    2 #
1 0 2 1 1
0 2 2 2 4
(26)

δEn = − En − 2En mω ⟨n| x̂ |n⟩ + mω ⟨n| x̂ |n⟩ ,
2mc2 2 2
" 2      2 #
1 1 2 2 1 1 2 2 1 2 4
=− n+ ℏ ω −2 n+ ℏω mω ⟨n| x̂ |n⟩ + mω ⟨n| x̂ |n⟩ ,
2mc2 2 2 2 2
(27)
From script section 5.13.5 we know that

⟨n| x̂2 |n⟩ = (2n + 1) , (28)
2mω
and we see the first two terms of the relativistic correction exactly cancel, and it simplifies
to
 2
1 1
δEn = − mω 2
⟨n| x̂4 |n⟩ , (29)
2mc2 2
q
Writing n̂ = ↠â, and x̂ = 2mω

(↠+ â) we now evaluate ⟨n| x̂4 |n⟩. First we note that
ℏ ℏ
x̂2 = (↠+ â)2 = (↠)2 + 2↠â + 1 + â2 (30)

2mω 2mω
where the commutator of ↠and â was used (see also exercise sheet 9, problem 4 for more
details). Thus we can write the expectation value of x̂4 = (x̂2 )2 as
 2
 

⟨n| x̂4 |n⟩ = ⟨n| (↠)2 + 2n̂ + 1 + â2 |n⟩ , (31)
2mω
ℏ2
= ⟨n| (↠)2 â2 + (2n̂ + 1)2 + â2 (↠)2 |n⟩ (32)
4m2 ω 2
ℏ2
+ ⟨n| (↠)4 + ((↠)2 + â2 )(2n̂ + 1) + (2n̂ + 1)((↠)2 + â2 ) + â4 |n⟩,
4m2 ω 2 | {z }
=0
(33)

3
Exercises Sheet 11 Prof. R. Grange; HS 2023

Note that the second term of the sum must be zero, because expectation values of eigenstates
|n⟩ must contain a equal powers of raising and lowering operators, otherwise they vanish due
to the orthogonality of eigenstates.
√ We now use the definition
√ of the raising and lowering
operators acting on |n⟩, â |n⟩ = n |n − 1⟩ and ↠|n⟩ = n + 1 |n + 1⟩, to evaluate the
remaining expression,

ℏ2  p p p p 
⟨n| x4 |n⟩ = n(n − 1) (n − 1)n + (2n + 1) 2
+ (n + 1)(n + 2) (n + 2)(n + 1) ,
4m2 ω 2
(34)
2

(n2 − n) + (4n2 + 4n + 1) + (n2 + 3n + 2) , (35)

= 2 2
4m ω
 2
ℏω
6n2 + 6n + 3 . (36)

=
2mω 2

Finally, we see the relativistic correction to the energy for the one-dimensional harmonic
oscillator is
2
ℏ2 ω 2

1 1 2
⟨n| x̂4 |n⟩ = − 6n2 + 6n + 3 . (37)

δEn = − 2
mω 2
2mc 2 32mc

e) The relative correction to the energy of the hydrogen eigenstates is

0
2  
δEnlm 1 Enlm 4n
0 = − 0 − 3 (38)
Enlm Enlm 2mc2 l + 1/2
0
 
E 3 2n
= nlm − , (39)
mc2 2 l + 1/2

which for the eigenstate |Ψ210 ⟩ is


0
δE210 7 E210
0 =− (40)
E210 6 mc2
4 eV
7 − 13.6
=− ≈ 8 × 10−6 ≪ 1. (41)
6 0.51 MeV
The relative correction to the energy of an electron in the ground state of the one-dimensional
harmonic oscillator with ω = 2π · 500 MHz is
δEn 1 ℏ2 ω 2
6n2 + 6n + 3 (42)

0
=− 2
En (n + 1/2)ℏω 32mc
ℏω 6n2 + 6n + 3
 
=− 2 , (43)
mc 32n + 16

which for the ground state (n = 0) is

δE0 3 ℏω
0 =− (44)
E0 16 mc2
3 h(500 MHz)
=− ≈ −7.6 · 10−13 ≪ 1. (45)
16 0.51 MeV
These ratios are both small, so we this perturbative approach is justified.

4
Exercises Sheet 11 Prof. R. Grange; HS 2023

2 Coupled harmonic oscillators (Exam problem FS2021)


Learning Objective: Quantum harmonic oscillator, perturbation theory, commutators, eigen-
states, solving exam problems

Two harmonic oscillators which are at the same frequency ω can be described by the Hamiltonian

Ĥ0 = Ĥa + Ĥb (46)

where
 
1
Ĥa = ℏω â â +†
(47)
2
 
1

Ĥb = ℏω b̂ b̂ + (48)
2

with
1 1
â = √ (mωx̂ + ip̂x ), b̂ = √ (mω ŷ + ip̂y ), (49)
2mℏω 2mℏω
where x̂, p̂x and ŷ, p̂y are the operators for position and momentum of two different oscillators
respectively.

a) Using the explicit forms for the raising and lowering operators written in terms of the variables
describing the position of the oscillators, show that

â, ↠= 1 (50)


 

Hint: Remember that [x̂, p̂x , =] iℏ.

A spring is now introduced that couples the two oscillators together. The Hamiltonian is now

Ĥ = Ĥ0 + Ĥc (51)

with the spring represented by the coupling term

Ĥc = ℏg(↠b̂ + b̂† â) (52)

with g ≪ ω

b) Evaluate the commutation relations between Ĥc and the operators ↠â, b̂† b̂ and ↠â + b̂† b̂.
What does the result tell you about the eigenstates of Ĥ0 and Ĥc ?
Hint: Recall that for operators Â, B̂ and Ĉ the following commutator relation holds:
h i h i h i
Â, B̂ Ĉ = Â, B̂ Ĉ + B̂ Â, Ĉ (53)

For any state |na , nb ⟩ = |na ⟩a |nb ⟩b , where |na ⟩a , |nb ⟩b are the eigenstates of Ĥa and Ĥb , respec-
tively, the number of excitations in the system is N = na + nb . N is consered because each term in
Ĥ0 and Ĥc has the same number of raising and lowering operators. For the rest of this problem,
we will restrict ourselves to the degenerate subspace where N = 1, which is spanned by the basis
states |1, 0⟩ and |0, 1⟩.

c) Evaluate the energy shift in first-order perturbation theory due to Ĥc for the states |1, 0⟩
and |0, 1⟩.

d) Evaluate the energy shift in first-oder perturbation theory due to Ĥc for the states |+⟩ =
√1 (|1, 0⟩ + |0, 1⟩) and |−⟩ = √1 (|1, 0⟩ − |0, 1⟩)
2 2

5
Exercises Sheet 11 Prof. R. Grange; HS 2023

e) The energy shifts you found in parts c) and d) should give you different results. Which pair
of states spanning the degenerate subspace is the correct one to consider and why? With the
correct pair of basis states, what are the higher order corrections?
f) Ĥc is a simplification of a real coupling Hamiltonian which looks like

∆Ĥ = ℏg(↠+ â)(b̂† + b̂). (54)

The time-dependence under the effect of this coupling leads to time-dependet equations for
the probability amplitudes cna ,nb of energy eigenstates of the combined system |na , nb ⟩ =
|na ⟩ |nb ⟩ of the form
dcna ,nb X −i
= ⟨na , nb | δ Ĥ |ka , kb ⟩ cka ,kb (t)eiω(na +nb −ka −kb ) (55)
dt ℏ
ka ,kb

By considering the behavior of terms coupled by a ↠b̂† and âb̂ versus those of âb̂† and ↠b̂,
describe why the former produce a negligible effect compared to the latter when the coupling
Hamiltonian is applied?

Solution
a) Using [x̂, p̂x ] = − [p̂x , x̂] = iℏ and [x̂, x̂] = [p̂x , p̂x ] = 0 and the definitions of â and ↠we can
compute the commutator directly:
1
â, ↠= (56)
 
[(mωx̂ + ip̂x ) , (mωx̂ − ip̂x )]
2mℏω
1
= ([mωx̂, mωx̂] + [mωx̂, −ip̂x ] + [ip̂x , mωx̂] + [ip̂x , −ip̂x ]) (57)
2mℏω
1
= (m2 ω 2 [x̂, x̂] −imω [x̂, p̂x ] +imω [p̂x , x̂] + [p̂x , p̂x ]) (58)
2mℏω | {z } | {z } | {z } | {z }
=0 =iℏ =−iℏ =0
1
= (ℏmω + ℏmω) (59)
2mℏω
= 1. (60)

b) By using the relation given in the hint we find


h i h i h i
Ĥc , ↠â = ℏg ↠b̂, ↠â + b̂† â, ↠â (61)
  
= ℏg b̂ ↠, ↠â + b̂† â, ↠â (62)
 

= ℏg(b̂ ↠, ↠â + b̂† ↠, â +b̂† â, ↠â + b̂† ↠[â, â]) (63)
     
| {z } | {z } | {z } | {z }
=0 =−1 =1 =0
 

= ℏg b̂ â − â b̂ †
(64)

Analogolously one finds


h i  
Ĥc , b̂† b̂ = ℏg ↠b̂ − b̂† â (65)
h i
= − Ĥc , ↠â . (66)

From these two results we can directly deduce


h i h i h i
Ĥc , ↠â + b̂† b̂ = Ĥc , ↠â + Ĥc , b̂† b̂ (67)
| {z }
=−[Ĥc ,↠â]

=0 (68)

6
Exercises Sheet 11 Prof. R. Grange; HS 2023

h i
The operator ↠â + b̂† b̂ is basicall Ĥ0 up to a constant factor and offset, so Ĥc , Ĥ0 = 0.
This means, that Ĥ0 and Ĥc share a set of eigenstates.
c) From first order perturbation theory we know that the energy shift is given by the expectation
value of the perturbing energy operator, in this case Ĥc . Hence we can compute the energy
shift for the |0, 1⟩ state:
D E
δE1,0 = 1, 0 Ĥc 1, 0 (69)
D   E
= ℏg 1, 0 ↠b̂ + b̂† â 1, 0 (70)
D E D E
= ℏg 1, 0 ↠b̂ 1, 0 + 1, 0 âb̂† 1, 0 (71)
D E
= ℏg( 1 ↠1 a ⟨0|b̂|0⟩b + ⟨1|â|1⟩a 0 b̂† 0 ) (72)
| {z } | {z } | {z } | {z }b
=0 =0 =0
=0
=0 (73)

where we used the orthogonality of the basis states. We find exactly analogously (we will
again end up with single ladder operators sandwiched between the same state which will
cancel)
D E
δE0,1 = 0, 1 Ĥc 0, 1 (74)
= 0. (75)

d) For |+⟩ and |−⟩ the principle is the same, but the situation is different because of the
emergence of cross-terms:
D E
δE+ = + Ĥc + (76)
ℏg D  
= 1, 0| + ⟨0, 1|) ↠b̂ + b̂† â (|1, 0⟩ + |0, 1⟩) (77)
2
ℏg D E D E D E D E
= ( 1, 0 ↠b̂ 1, 0 + 1, 0 ↠b̂ 0, 1 + 1, 0 âb̂† 1, 0 + 1, 0 âb̂† 0, 1 (78)
2 | {z } | {z } | {z } | {z }
=0 =1 =0 =0
D E D E D E D E
† † † †
+ 0, 1 â b̂ 1, 0 + 0, 1 â b̂ 0, 1 + 0, 1 âb̂ 1, 0 + 0, 1 âb̂ 0, 1 ) (79)
| {z } | {z } | {z } | {z }
=0 =0 =1 =0
= ℏg, (80)

and similarly for |−⟩:


D E
δE− = − Ĥc − (81)
ℏg D  
= 1, 0| − ⟨0, 1|) ↠b̂ + b̂† â (|1, 0⟩ − |0, 1⟩) (82)
2
ℏg D E D E D E D E
= ( 1, 0 ↠b̂ 1, 0 − 1, 0 ↠b̂ 0, 1 + 1, 0 âb̂† 1, 0 − 1, 0 âb̂† 0, 1 (83)
2 | {z } | {z } | {z } | {z }
=0 =1 =0 =0
D E D E D E D E
− 0, 1 ↠b̂ 1, 0 + 0, 1 ↠b̂ 0, 1 − 0, 1 âb̂† 1, 0 + 0, 1 âb̂† 0, 1 ) (84)
| {z } | {z } | {z } | {z }
=0 =0 =1 =0
= −ℏg (85)

e) One might have noticed in parts c) and d) that unlike |1, 0⟩ and |0, 1⟩, |+⟩ and |−⟩ are
actually eigenstates of Ĥc with eigenvalues ±ℏg:

7
Exercises Sheet 11 Prof. R. Grange; HS 2023

ℏg  
Ĥc |+⟩ = √ ↠b̂ + âb̂† (|1, 0⟩ + |0, 1⟩) (86)
2
ℏg
= √ (|0, 1⟩ + |1, 0⟩) = ℏg|+⟩ (87)
2
ℏg  † 
Ĥc |−⟩ = √ â b̂ + âb̂† (|1, 0⟩ − |0, 1⟩) (88)
2
ℏg ℏg
= √ (|0, 1⟩ − |1, 0⟩) = − √ (|1, 0⟩ − |0, 1⟩) = −ℏg|−⟩ (89)
2 2

This means that they give the correct energy shift due to the coupling Hamiltonian. Follow-
ing, there are no higher order corrections since the eigenvalue is already the correct result.
f) The coupling terms ↠b̂† and âb̂ only produce non-zero transition matrix elements when
ka,b = na,b ± 1. Mathematically, this means that the phase in the exponential factor is non-
zero because na + nb − ka − kb ̸= 0. So each of these terms is oscillating rapidly with respect
to the coupling rate g and thus quickly averages out over the timescales of consideration.
Physically, this means that even though the terms seem to result in the number of excitations
and energy not being conserved, their effects are canceled out on the timescale of the coupling
Hamiltonian.

8
Exercises Sheet 11 Prof. R. Grange; HS 2023

3 The Zeeman effect and dipole allowed transitions


Learning Objective: Zeeman effect, optical transitions

Consider the situation depicted in the graphic below, where electromagnetic radiation at wave-
length λ can drive transitions between energy eigenstates |n, l, m⟩, with the lower energy levels
given by |n1 , 1, m1 ⟩ and the higher energy levels by |n2 , 2, m2 ⟩ where n2 > n1 .

l=2

l=1

without magnetic with magnetic


field field B

a) Complete the qualitative diagram with the Zeeman sublevels for the case where there is a
magnetic field.
b) Draw all the electric dipole allowed optical transitions and indicate the polarization of the
electromagnetic radiation required to drive them.

Solution
The solutions for both parts a) and b) are shown below:
circ. lin. circ.
E pol. pol. pol.
m=2
}

}
}

m=1
l=2 m=0
m=-1
m=-2

m=1
l=1 m=0
m=-1
σ+ σ-
}
}

π
Δm=+1 Δm=0 Δm=-1
without magnetic with magnetic
field field B

The magnetic field lifts the degeneracy between the sublevels with different m. The electric dipole
allowed transitions are those which fulfill the selection rules
• ∆l = ±1 and ∆m = 0 for linearly (π) polarized light.
• ∆l = ±1 and ∆m = +1 for right handed (σ + ) circularly polarized light.
• ∆l = ±1 and ∆m = −1 for left handed (σ − ) circularly polarized light.

Comment
Be careful! The labels ’left’ and ’right’-handed rely on a convention, make sure you know which
one you are using.

9
Exercises Sheet 11 Prof. R. Grange; HS 2023

4 Time-dependent perturbation theory


Learning Objective: Time-dependent perturbation theory, two-level system, atomic transitions

Consider two eigenstates of an atom, |ϕa ⟩, |ϕb ⟩ (for example hydrogen) with a time-dependent per-
turbation term given by the interaction of the atom’s dipole moment p⃗0 = −e⃗r with an oscillating
˜
⃗ = E0 cos(ωt)⃗k,
electric field E

˜
Ĥ1 (t) = −p⃗ˆ0 · E
⃗ = eE0 cos(ωt)⃗rˆ · ⃗k. (90)

For a general state given by a superposition of these two eigenstates |ψ⟩ = ca |ϕa ⟩ + cb |ϕb ⟩, first-
order time-dependent perturbation theory approximates these coefficients as
(1)
∂cm (t) 1 X (0) −i(ωn −ωm )t
= c e ⟨ϕm | Ĥ1 (t) |ϕn ⟩ . (91)
∂t iℏ n n

(1)
a) Integrate equation (91) to find an expression for cb (t). Use the definition Vab ≡ ⟨ϕa | eE0⃗rˆ ·
⃗k˜ |ϕb ⟩.
Hint: You do not have to fully solve the integral, it will be simplified in subsequent parts of
the problem.
(0)
b) Assuming the system starts in the initial pure state |ϕa ⟩ (where ca = 1). Show that the
probability to make a transition from |ϕa ⟩ → |ϕb ⟩ is

|Vab | sin2 ω02−ω t


2

2
(1)
Pa→b (t) ≈ cb (t) = , (92)
ℏ2 (ω0 − ω)2

where ω0 = ωa − ωb is the frequency of the atomic transition. Take the approximation that
ω0 + ω ≫ |ω0 − ω|.

c) What is the period of the oscillations of Pa→b (t) and what is its maximum value?
d) For what parameters is the approximation in equation (92) valid? What is unphysical about
this result?
e) Instead of a applying an oscillating electric field for all time, consider a pulse with a charac-
˜
⃗ = E0 cos(ωt)e−t2 /τ 2 ⃗k.
teristic time τ , E Calculate Pa→b (t) at a time much later than when
this pulse is applied. When do you get the maximum probability for a transition?

Solution
(1)
a) We find the coefficient cb (t) of |ϕb ⟩ under first order perturbation theory by integrating
˜
equation (91), which we can first rewrite using Vab ≡ ⟨ϕa | eE0⃗rˆ · ⃗k |ϕb ⟩:
(1)
∂cb (t) 1 X (0) −i(ωn −ωb )t
= c e Vbn cos(ωt), (93)
∂t iℏ n n

Next, we explicitly write the sum over n ∈ {a, b} and integrate:


(1)
∂cb (t) 1 1 (0)
= c(0) e−i(ωa −ωb )t Vba cos(ωt) + cb Vbb cos(ωt), (94)
∂t iℏ a iℏ
Z t
(1) 1 −i(ωa −ωb )t′ 1 (0)
⇔ cb (t) = dt′ c(0)
a e Vba cos(ωt′ ) + cb Vbb cos(ωt′ ) (95)
0 iℏ iℏ

10
Exercises Sheet 11 Prof. R. Grange; HS 2023

(0) (0)
b) We will use the initial conditions ca = 1 and cb = 0, and plug them into the resulting
integral we found in a):
Vba t ′ −iω0 t′
Z
(1)
cb (t) = dt e cos(ωt′ ), (96)
iℏ 0
Vba t ′ −iω0 t′  iωt′
Z 

= dt e e + e−iωt , (97)
2iℏ 0
Vba t ′  −i(ω0 −ω)t
Z 
= dt e + e−i(ω0 +ω)t (98)
2iℏ 0
 −i(ω0 −ω)t
− 1 e−i(ω0 +ω)t − 1

Vba e
= + . (99)
2iℏ −i(ω0 − ω) −i(ω0 + ω)
Rt ′
We used 0 dt′ eat = a1 (eat − 1) for both integrals. We assume that the driving field is near
resonant, i.e., ω is close to the value of ω0 . Then ω0 + ω ≫ |ω0 − ω|, so we can neglect the
second term, leading to

(1) Vba e−i(ω0 −ω)t − 1


cb (t) = , (100)
2ℏ (ω0 − ω)
 −i(ω0 −ω)t/2
− ei(ω0 −ω)t/2

Vba e
= ie−i(ω0 −ω)t/2 , (101)
ℏ(ω0 − ω) 2i
 
Vba ω0 − ω
=− ie−i(ω0 −ω)t/2 sin t . (102)
ℏ(ω0 − ω) 2
The probability for the transition a → b is given by
|Vba | sin2 ω02−ω t
2

2
(1)
⇒ Pa→b (t) ≈ cb (t) = . (103)
ℏ2 (ω0 − ω)2

c) The period of oscillation of Pa→b (t) is the period of sin2 (x), which is π, so it is
T = 2π/(ω0 − ω), (104)
which is the inverse of the detuning of the drive from ω0 . The maximum value of sin2 (x)
|2
is 1, so the maximum value of Pa→b (t) is (ℏω|V0 ba
−ℏω)2 , which is the ratio of the energy scale
of the interaction between the two states |ϕa ⟩ and |ϕb ⟩ and the energy scale of the detuning
ℏ(ω0 − ω).
d) From the previous answer we can immediately notice something unphysical, which is that the
probability is not limited to be Pa→b (t) < 1. This gives us a regime where our perturbation
theory is valid, which is actually when we not near detuned, |Vba | ≪ ℏ(ω0 − ω).
2 2
e) If we instead consider a field applied with a Gaussian envelope e−t /τ the integration from
b) needs to be adjusted:
Vba ∞ ′ −iω0 t′
Z
(1) ′2 2
cb (t) = dt e cos(ωt′ )e−t /τ , (105)
iℏ −∞
Z ∞
Vba  ′ ′
 ′2 2
= dt′ e−i(ω0 −ω)t + e−i(ω0 +ω)t e−t /τ . (106)
2iℏ −∞
We will again neglect the second term, this time due to the rotating wave approximation,
which states that compared to the dynamics of the first term,
R ∞the second term p
oscillates very
2 2
fast and averages out. We can use the Gaussian integral ∞ dx e−ax +bx = πa eb /(4a) to
obtain
Vba ∞ ′ −i(ω0 −ω)t′ −t′2 /τ 2
Z
(1)
cb (t) = dt e e , (107)
2iℏ −∞
Vba √ − 1 (ω0 −ω)2 τ 2
= τ πe 4 . (108)
2iℏ

11
Exercises Sheet 11 Prof. R. Grange; HS 2023

Again, the probability is obtained by taking the absolute value squared:


2
|Vba | 2 − 1 (ω0 −ω)2 τ 2
Pa→b (t) ≈ τ πe 2 . (109)
4ℏ2
Here we get the maximum probability for transition when we are resonant, ω = ω0 . We see
that we have a condition on the length of our pulse, it must be shorter than the timescale
given by the strength of the interaction, τ ≪ ℏ/ |Vba |.

12
Exercises Sheet 11 Prof. R. Grange; HS 2023

5 Rabi oscillations
Learning Objective: Two-level system, Rabi oscillations, solving differential equations, check
validity of approximations

⃗ = E0 cos(ωt)⃗k˜ we can solve the dynamics of the two-level system


For the oscillating electric field E
exactly, without using perturbation theory. First write the system of coupled differential equations
for the coefficients cn (t),

∂cm (t)) 1 X
≡ ċm (t) = cn (t)e−i(ωn −ωm )t ⟨ϕm | Ĥ1 (t) |ϕn ⟩ (110)
∂t iℏ n

We also introduce the shorthand


˜
Vab ≡ ⟨ϕa | E0⃗k |ϕb ⟩ (111)

a) To simplify these equations, use the fact that Ĥ1 (t) contains only off-diagonal elements when
the basis states are those of the hydrogen atom, i.e., Vaa = Vbb = 0, and show that
1
ċa (t) = cb (t)eiω0 t Vab cos(ωt) (112)
iℏ
1
ċb (t) = ca (t)e−iω0 t Vba cos(ωt). (113)
iℏ

Combining e±iω0 t with cos(ωt) yields terms like e±i(ω0 −ω)t and e±i(ω0 +ω)t . For transitions you can
drive with optical light, ω0 and ω are in the hundreds of THz regime, and for the resonant case
ω0 ≈ ω the frequency sum is much greater than the difference ω0 + ω ≫ ω0 − ω. Take the rotating
wave approximation, where we neglect terms like e±i(ω0 +ω)t as rotating so fast that their effect is
averaged out over the relevant timescale of the problem.
b) You now have two coupled first-order differential equations, which maybe you don’t know
how to solve yet. It works as follows: Take the time derivative of the equation for ċb , and
plug in the expression for ċa to arrive at a second order differential equation of the form
ac̈b + bċb + c = 0.
c) The general solution for such a differential equation is given by cb = Aeλ1 t + Beλ2 t where
λ1/2 are the solutions to the equation aλ2 + bλ + c = 0. Using this and the initial condition
cb (t = 0) = 0, show that
 
2 Ωt 1 1
|cb (t)| ∝ sin2 = − cos(Ωt), (114)
2 2 2
q
2
with Ω = (ω0 − ω)2 + |Ω0 | , where Ω0 = Vℏba is the so-called Rabi frequency. Ω is sometimes
called the generalized Rabi frequency.
d) Find the normalization constant of cb (t) by taking the time derivative of your solution,
plugging it into eq. (113) to find ca (t) and employing the initial condition ca (t = 0) = 1.
What is now the maximal transition probability, and at what time is it achieved?
2 2
e) As sanity check, confirm that |ca (t)| + |cb (t)| = 1.
f) Confirm that for the parameter regime you defined in problem 4, the exact solution for cb (t)
reduces to the one we found using the perturbation theory approach.

Solution
a) First, we note that
Z 
ˆ ˆ

2
ϕa/b eE0⃗r · ⃗k ϕa/b = eE0 ϕa/b ⃗r · ⃗k dV (115)
V

13
Exercises Sheet 11 Prof. R. Grange; HS 2023

is an infinite volume integral over an odd function of space, so it has to equal zero. Incidently
this implies that Vaa = Vbb = 0. This simplifies equation (110) to
1
ċa = cb eiω0 t Vab cos(ωt) (116)
iℏ
1
ċb = ca e−iω0 t Vba cos(ωt) (117)
iℏ

b) We rewrite the equations we found in part a) using cos(x) = 1


e−ix + eix . We find

2

1 −ix  1 1
e±iω0 t e + eix = e±i(ω0 −ω)t + e±i(ω0 +ω) (118)
2 2 2
Now we neglected the term with e±i(ω0 +ω) in the rotating wave approximation. This gives
us the starting point for solving the coupled differential equation:
1 Vab
ċa = cb ei(ω0 −ω)t (119)
iℏ 2
1 −i(ω0 −ω)t Vba
ċb = ca e . (120)
iℏ 2
Taking the time derivative of the second equation with the product rule yields
Vba Vba
c̈b = ċa e−i(ω0 −ω)t − i(ω0 − ω) ca e−i(ω0 −ω)t (121)
2iℏ 2iℏ
2
|Vba |
= cb − i(ω0 − ω)ċb , (122)
(2iℏ)2

Where we replaced ċa with the first differential equation. Finally we obtain
2
|Vba |
c̈b + i(ω0 − ω)ċb + cb = 0. (123)
4ℏ2

c) The characteristic polynomial of this differential equation is


2
|Vba |
λ2 + i(ω0 − ω)λ + =0 (124)
4ℏ2
which can be obtained by plugging in the ansatz cb (t) = eλ into the equation. The solutions
to this quadratic equation is
 s 
2
1 |V ba |
λ1,2 = −i(ω0 − ω) ± −(ω0 − ω)2 −  (125)
2 ℏ2
1
= (−i(ω0 − ω) ± iΩ) , (126)
2
where Ω is the generalized Rabi frequency as defined in the problem text. The general
solution for cb is thus
 
cb (t) = e−i(ω0 −ω)t/2 AeiΩt/2 + Be−iΩt/2 (127)
        
Ωt Ωt Ωt Ωt
= e−i(ω0 −ω)t/2 A cos + Ai sin + B cos − + Bi sin − (128)
2 2 2 2
    
Ωt Ωt
= e−i(ω0 −ω)t/2 C cos + D sin , (129)
2 2

where A, B, C, D are constants, and combined via C = A + B and D = i(A − B), where we
used the (anti-)symmetry of cos(sin). Since in the beginning, the atom is in state |ϕa ⟩, the

14
Exercises Sheet 11 Prof. R. Grange; HS 2023

boundary condition is cb (t = 0) = 0 so that C = 0 and


 
Ωt
cb (t) = De−i(ω0 −ω)t/2
sin (130)
2
   
2 2 Ωt Ωt 1 1
⇒ |cb (t)| = |D| sin 2
∝ sin2
= − cos(Ωt), (131)
2 2 2 2

as we were to show. So we see that the probability for the atom to be in either |ϕa ⟩ or |ϕb ⟩
oscillates at the generalized Rabi frequency Ω.
d) We take the time derivative of eq. (130) and set it equal to eq. (120):
    
ω0 − ω −i(ω0 −ω)t/2 Ωt Ω −i(ω0 −ω)t/2 Ωt
ċb = D −i e sin + e cos (132)
2 2 2 2
(120) 1 Vba
= ca e−i(ω0 −ω)t (133)
iℏ 2    
2 i(ω0 −ω)t/2 ω0 − ω Ωt Ω Ωt
⇔ ca (t) = Diℏ e −i sin + cos . (134)
Vba 2 2 2 2

From the initial condition ca (t = 0) = 1 follows

2 Ω
1 = Diℏ (135)
Vba 2
Vba Ω0
⇔ D= = (136)
iℏΩ iΩ
So we find for ca and cb :
    
Ωt ω0 − ω Ωt
i(ω0 −ω)t/2
ca (t) = e cos −i sin (137)
2 Ω 2
 
Ω0 −i(ω0 −ω)t/2 Ωt
cb (t) = e sin (138)
iΩ 2
2  
2 Ω0 Ωt
⇒ |cb (t)| = sin2 . (139)
Ω 2

Since Ω ≥ Ω0 and Ω = Ω0 when the light is resonant with the atomic transition (ω = ω0 ), the
2
maximum transition probability is |cb | = 1, as we would expect. The complete transition
is achieved when Ωt/2 = π/2, i.e. ta→b = π/Ω. This result tells us that by applying a well-
timed pulse of monochromatic light (e.g. a laser) on resonance with an atomic transition,
we can deterministically transfer the atom from one state into another. As you can imagine,
this is the basis for many experiments involving atoms or other systems with discrete energy
levels which we can approximate as two-level systems.
e) We find

(ω0 − ω)2
   
2 Ωt Ωt
|ca (t)| = cos2 + sin2 (140)
2 Ω2 2
2
(ω0 − ω)2
     
2 2 Ωt |Ω0 | Ωt Ωt
⇒ |ca (t)| + |cb (t)| = cos 2
+ sin2
+ sin2 (141)
2 Ω2 Ω2 2 2
2
(ω0 − ω)2 + |Ω0 |
   
Ωt Ωt
= cos2 + sin2
(142)
2 Ω2 2
   
Ωt Ωt
= cos2 + sin2 = 1. (143)
2 2
2
where we used Ω2 = (ω0 − ω)2 + |Ω0 | .

15
Exercises Sheet 11 Prof. R. Grange; HS 2023

f) The parameter regime we considered is when Ω0 ≪ ω0 − ω so that cb (t) ≪ 1. Then,


q
2
Ω = (ω0 − ω)2 + |Ω0 | ≈ ω0 − ω (144)

This implies
 
Ω0 (ω0 − ω)t
cb (t) ≈ e−i(ω0 −ω)t/2 sin . (145)
i(ω0 − ω) 2

which is equivalent to the solution we found in problem 4.

16

You might also like