L11  ANGULAR MOMENTUM

UCSC, PHYS 139A, SPRING 2008

GEY-HONG SAM GWEON

1.

Orbital Angular Momentum

From CM, we are familiar with the following denition of the angular momentum

L=rp
In QM, we dene angular momentum operators accordingly

Lx = y pz z py , Ly = z px xpz , Lz = xpy y px

(1.1)

The rst thing to note is the following canonical commutation relations for momentum and position
components:

pa = i

(1.2)
where

a = 1, 2, 3

standing for

, [a , pb ] = i ab , [a , pb ] = 0, [a , qb ] = 0
q
p
q
qa

x, y, z ,

and

q1 , q2 , q3 = x, y, z

respectively.

to QM, there is one thing that needs to be always checked.

When going from CM

Operators for observables must be

Hermitian. This is not always automatic because of operator algebra. In this case, we are ne since

x z

L = p y p z = pz y py z = y pz z py = Lx
y

where we made use of the fact that, because of

[a , pb ] = i ab , pa
q

and

qb

for

a=b

can be swapped

freely. Now we calculate the commutation relations for the components of the angular momentum:

[Lx , Ly ] = [pz z py , z px xpz ]
y

= [pz , z px xpz ] [py , z px xpz ]

y

= [pz , z px ] [pz , xpz ] [py , z px ] + [py , xpz ]

y
y
z
z
Using the rule

[AB, C] = A[B, C] + [A, C]B

and

[A, B C] = B[A, C] + [A, B]C

repeatedly, and using

Eq. 1.2, we get

(note that

[Lx , Ly ] = y [z , z ]x 0 0 + py [, pz ]
p p
z x
x, py commute) = i y px + i xpy

= i Lz

Other commutators need not be calculated but they can be inferred by cyclic permutation of indices

(xyz) (zxy) (yzx)

[Lx , Ly ] = i Lz , [Ly , Lz ] = i Lx , [Lz , Lx ] = i Ly

(1.3)
The

L2

operator is dened as

x y z
L2 = L2 + L2 + L2
1

UCSC, PHYS 139A, SPRING 2008 GEY-HONG SAM GWEON

La

and the commuation relation between it and

[L , Lx ] =
=

is easy to see

[L2 + L2 + L2 , Lx ]
x
y
z
2 , Lx ] + [L2 , Lx ]

[Ly
z

= Ly (i Lz ) i Lz Ly + Lz (i Ly ) + i Ly Lz
= 0
Thus

[L2 , Lx ] = [L2 , Ly ] = [L2 , Lz ] = 0
In the spherical coordinate system, what are the dierential operators for
the next section, only

Lz

and

L2

L?

As it will turn out in

are necessary to know. We already know, from L9, what they are:

(1.4)

Lz = i

(1.5)

L2 =

sin

sin

We have stated then that the spherical harmonics

eigenstates of

Lz

and

2
sin2 2

{|lm = Ylm (, )}

form the set of simulataneous

L2 :

Lz |lm

L2 |lm

= m |lm

= l(l + 1) 2 |lm
l = 0, 1, 2, ....
m = l, ..., 0, ..., l

We will see in the next section how this comes about.

2.

General Angular Momentum  Algebraic Approach

In this section, we take a very general approach to describe the angular momentum. This approach
is a little similar to the

approach,
approach.


or an

a , a+

operator
analytical

description of the simple harmonic oscillator in that it is an

algebraic approach,

dierential equation approach,

rather than a

or an

There is a fundamental dierence in the current case, though.

The operator, or algebraic, approach to describe angular momentum includes cases that simply
cannot be described within the analytical approach.

The reason for this is the spin angular momentum, which does not have any CM analogue, and so
we simply do not know how to set up a dierential equation for that!
So much for the motivation.

dened

Within the algebraic approach, the angular momentum operator is

as consisting of three observables

Jx , Jy , Jz

that satisfy the following fundamental relations

[Jx , Jy ] = i Jz , [Jy , Jz ] = i Jx , [Jz , Jx ] = i Jy

(2.1)

which are called

fundamental commutation relations for angular momentum.

Dene

J = Jx iJy

(2.2)
Note that

2 2

2 2


J+ J = Jx + Jy + i Jy Jx i Jx Jy = Jx + Jy +

2 J and so
z

2
J 2 = J+ J + Jz

2J .
z

J J+

can be treated similarly and we get

2
J 2 = J J + Jz

(2.3)

1Here,
and

we use

to denote general angular momentum.

for spin angular momentum.

Jz

is the standard notation for orbital angular momentum

is for the general or the total angular momentum.

L11  ANGULAR MOMENTUM

The following commutation relations can be derived easily (the rst line was already derived in the
previous section, since only the fundamental commutation relations are used there):

[J 2 , Jx ] = [J 2 , Jy ] = [J 2 , Jz ] = 0

[J , Jz ] = [Jx iJy , Jz ] = i Jy i(i )Jx =

(2.4)
(2.5)

J2

From the general formalism of QM, Eq. 2.4 means that

and any of

Jx , Jy , Jz

are compatible

observables and so simulatneous eigenstates of the two of them can be found to form a natural

J2

basis. By convention, we take simultaneous eigenstates of

just as well choose

Jx

or

Jy

and

as the second observable instead of

Jz ,

Jz .

while in principle we could

Two things are worthwhile to

note. (1) This example illustrates the basic point that just because an observable
with two observables

B, C

does not mean that

B, C

Jy ,

is compatible

are compatible with each other.

J2

a natural basis is chosen as simultaneous eigenstates of

and

and

Jz ,

(2) Once

all other operators, e.g.

Jx

are completely desribed using those basis states by denition. Let us call a simulatneous

eigenstate

|jm

, speccied by some value of the total angular momentum squared (J ) and the
2

component of the angular momentum (Jz ). At the moment, we need to specify only the latter
value, while we just specify the former as some nite number:

Jz |jm = m |jm
Investigate the meaning of

J ,

by applying the commutator of Eq. 2.5 on

[J , Jz ]|jm

J m |jm Jz J |jm

|jm

J |jm

J |jm

=
=

Re-arranging the terms, we get

Jz J |jm

(2.6)
Thus,

J+

increases the

Jz

J |jm

= (m 1)

value by 1, while

decreased it by 1.

ladder operators for angular momentum, in analogy with

harmonic oscillator problem.

Namely, these operators are

that were ladder operators for the

There is an important dierence here, though:

values must be

bounded both from below and from above, given a nite total angular momentum value. In fact,
it is the convention that the symbol

stands for the maximum value of

m.

In other words,

J+ |jj = 0
Now, apply

J2

to

|jj

. From Eq. 2.3 (2nd form), we get

J 2 |jj = j(j + 1) 2 |jj

mmin such that J |jmmin = 0. Note two

2 |jmmin = mmin (mmin 1) 2 |jmmin . Second, since
things. First, due to Eq. 2.3 (1st form), J

J 2 and J commute with each other, as can be seen from Eq. 2.4, if | is an eigenstate of
2 with eigenvalue , then J 2 J | = J J 2 | = J | , i.e. J | is an eigenstate of J 2 with

J
2 eigenvalue, and especially,
the same eigenvalue. Namely, in our case, any |jm has the same J
2 =m
2
comparing the m = j and m = mmin cases, we have j(j + 1)
min (mmin 1) . The solution

is mmin = j . Note that by the nature of J (or J+ ), Eq. 2.6, the dierence between m = j and
m = j must be an integer, and so j can be any half integer. Since we dened j as the maximum

Jz value, it cannot be negative.

Now, consider operating

repeatedly until we reach

We summarize our ndings so far.

2Standard

notations for the

component of the angular momentum is

be used whenever necessary to avoid ambiguity.

mass.

Unfortunately,

m,ml ,ms ,mj ,

where the subscript may

is the same symbols as that for the electron

However, this is the standard practice and usually no confusion seems to arise.

UCSC, PHYS 139A, SPRING 2008 GEY-HONG SAM GWEON

Jz |jm

J 2 |jm

= m |jm

= j(j + 1) 2 |jm
jm|j m
= jj mm
m = j, ..., 0, ..., j
1
3
j = 0, , 1, , 2, ...
2
2
8

Note that the orthonormalization condition is imposed for

is a discrete index (Denition 4.9 of L8).

2j + 1,

values is

Denition 2.1.

For a given

which is often referred to as the

j,

|jm

jm
dierent m

, possible since the group index

the number of states with

multiplicity.

Spin, Fermion, Boson

These are the solutions for the angular momentum problem purely from the algebraic approach.
Note that there is a dierence from the solutions written for the

that

can be now

half integers

as well as integers, while

orbital

angular momentum in

was restricted to integers (previous

section and L9). There are two possibilities, (i) the algebraic solutions above are superuous and
really the half integer solutions are forbidden due to some reasons that we failed to consider,
and (ii) the algebraic solutions with half integer solutions correspond to some real physics that
we could not capture by dening QM angular momentum starting from CM angular momentum
as we did in Section 1.
analogue is called the

The correct one is (ii), and the angular momentum that has no CM

spin angular momentum.

Spin is an intrinsic property of elementary particles.

In fact, elementary particles can be grouped into two kinds, one having half integer spin values
and the other integer spin values.
bosons.

The former is a group of so-called fermions and the latter

The spin angular momentum for all elementary fermions (

e,, ,e , , ;

and their anti-particles),is

j = 1/2,

udcstb

quarks, and leptons

and so we call them spin 1/2 particles. The spin

for elementary bosons vary from 0 (Higgs boson  not yet observed), 1 (photon), and 2 (graviton 
not yet observed). The spin of composite particles is determined by the particular way the spins of
constituent particles are summed (see below for the topic of the angular momentum addition). For

instance

mesons consist of a quark and an anti-quark in such a way that their spin is 0. So,

mesons themselves are bosons, although they are made up of two fermions.

Note
The

. The Dimension of the Hilbert Space

solutions presented in the box are all we need
2.2

in order to compute any physical quantities.

{|jm |m = j, ..., 0, ..., j}

2j + 1. This is the case of spin, which is xed for a given particle. For
the electron, spin is 1/2, and so the Hilbert space is two dimensional, i.e. there are two basis
vectors corresponding to m = 1/2. For the orbital angular momentum problem for the Hydrogen
problem, j is not xed, but covers all integers 0, 1, 2, ..... In that case, the dimension of the Hilbert

If

is xed for a given problem, then the Hilbert space is spanned by

and so its dimension is

space is innite dimensional, which comes from the basic fact that in the real space, the Hilbert
space is a space of functions, and thus innite dimensional.

Note

. Wave Function

2.3

You may ask what is the wave function for

and

|lm

|jm ? In

Lz are

2
, we have already noted how L and

and how

|lm

is the spherical harmonic function,

the case of the orbital angular momentum,

represented by dierential operators in

Ylm (, ).

and

This is all well, but actually we do

not need to know what this function looks like in order to gure out all physical quantities. It is
quite the same situation as in the harmonic oscillator problem. Life was much easier when we used
the

algebra on eigenstates of

with the explicit functions of

easier to use the

(or

a+ a for the

p). Similarly,

simple harmonic oscillator problem than dealing

in the case of the angular momentum, it is much

algebra on simulataneous eigenstates of

L2 , Lz

rather than dealing with the

L11  ANGULAR MOMENTUM

, . Besides, in the case of spin, we do not even know 3 how to write the wave
terms of , . This is related to the fact that we do not know what the microscopic
for the spin 1/2 angular momentum is and we know the existence of the spin only

explicit functions
function in
mechanism

through its observed properties.

states
Let

|jm

us

In that case, the best we can do is to deal with the

abstract

. However, the structure of science is such that that is all we need for now.

close

this

section

by

expressing

Cjm,

Eq. 2.6, we can dene a number

jm|J = j, m 1|Cjm, ,

the

action

of

ladder

operators

J |jm = Cjm, |jm 1

by

more

denitely.

Noting that

4
we have

jm|J J |jm

jm|J

J = J

From
and

J |jm

= |Cjm, |2
On the other hand, from Eq. 2.3, we have

jm|J J |jm

j(j + 1) m2

j 2 m2 + j

= (j
Equating these two results, we can determine

|jm

m)(j m + 1)

Cjm,

m
2

up to an arbitrary phase.

But, the states

themselves are arbitrary up to a phase and so we have some freedom. By convention, relative

phases of

|jm

are chosen so that

Cjm,

J |jm =

(2.7)

3.

values are real and positive. Thus we get

m)(j m + 1)|j, m 1

(j

Conservation of Angular Momentum

The angular momentum is conserved when the system is invariant under rotation. For a spherically
symmetric system, the angular momentum is conserved since by denition the system is isotropic.
A less symmetric problem can occur, e.g. if a

eld is applied along the

the system becomes only cylindrically symmetric.

Lz

direction. In that case,

is conserved, but not

Lx , Ly .

When the angular momentum is conserved, according to Denition 8.3 of L8, it should be possible

to take simultaneous eigenstates of

and the angular momentum.

In the previous section, we

saw that the eigenstates of the angular momentum is characterized by

J 2 , Jz

eigenvalues.

Thus,

it is no surprise that in the spherically symmetric potential problem we found that eigenstates
can be written as
momentum.

|nlm = Rnl (r)|lm

, where

|lm

(Ylm ) are the eigenstates of the orbital angular

This remains true when the problem becomes only cylindrically symmetric.

The

dierence is that in the cylindrically symmetric problem the degeneracy between states with dierent

values is lifted, and that

Lx , Ly

are no longer conserved.

The following table summarizes this situation in general.

3This

might sound impossibly ridiculous at rst.

momentum of spin as
equation for it?

Also,

that if you solve those

in

terms

of

and

not

You may ask, what is wrong if I just

dened

this

way,

then

you

get

results

for

the

two

wave

functions,

|1/2, 1/2

, that do

problems.

the angular

In short, this way of naively justifying the spin angular momentum is futile.

corresponding

to

obey the general angular momentum properties, Eq. 2.6. There are also other mathematical

Modern Quantum Mechanics

4A

dene

Sz = i /s , where s is the angle in the spin space and just solve the eigenvalue
2
similarly for S , why can't we just use Eq. 1.5 with s and s ? The problem is
3 2

2
two dierential equations Sz |1/2, 1/2 = 2 |1/2, 1/2 and S |1/2, 1/2 = 4 |1/2, 1/2

See more discussions in

by J. J. Sakurai.

comment on the notation.

Griths uses expressions such as

O|

and

|O

These expressions are not

standard, while it is perhaps understandable why they were chosen for pedagogy. In the standard notation, these

two are written as,

|O and O| . As we convered in the formalism, (O| ) = |O . In my opinion, the
standard notation is better, since it is more powerful and cleaner.

UCSC, PHYS 139A, SPRING 2008 GEY-HONG SAM GWEON

Hamiltonian

Symmetry

Conserved Angular Momentum

Spherical

J 2 , Jx , Jy , Jz
d
d 2
= 0 for a = x, y, z
dt Ja = dt J
can be L or S or the total

J

J 2 , Jz only
d

Jz = d J 2 = 0 but d Jx,y = 0

Stationary States
(angular momentum part)

p2

2m

p2

2m

+ V ()
r

+ V () + AJz
r

Cylindrical

dt

Note

dt

dt

|jm
Ejm

m
2j + 1

is independent of

degeneracy at least

|jm

Ejm

is

dependent

on

. Orbital and Spin Subspaces

3.1

In the above table, the nature of

is left ambiguous on purpose. Generally, a particle's angular

mometum comprises of the orbital part and the spin part. In the above table,
angular momentum

L,

the spin angular momentum

S,

can be the orbital

or the sum of the two. When we learn how

to add angular momenta, we will update the above table so that stationary states are written
more completely in terms of all relevant quantum numbers. For now, consider
only

one

of

L, S

above as meaning

and the sum of the two.

[H, Ja ] = 0 for a = x, y, z
Jx,y ] = 0. For any given

[H,

The above discussion makes it clear that in the spherical symmetric case
and in the cylindrically symmetric case

[H, J 2 ] = [H, Jz ] = 0

but

Hamiltonian, this can be explicitly checked. The following two examples illustrate this point, and
along the way, we derive some useful algebra, such as Eqs. 3.1,3.2, which are actually specic
examples of very general properties of operator algebra in the angular momentum space.

Other

than that, how much satisfaction one draws from the following two examples depends on how much
one appreciates the rst sentence of this section that it is the rotational invariance that leads to
the consevation of the angular momentum.

Example3.2. SphericallySymmetricPotentialandOrbitalAngularMomentumConservation
First of all, in this and the next example, the Einstein summation convention (any repeating indices

imply summation) and the Levi-Civita symbol

x, y, z for a = 1, 2, 3
a = abc qb pc .
L

means
as

abc will be used for the sake of economy. Also,

respectively. The orbital angular momentum (r

p)

qa

is then expressed

Take the case

p2

+ V ()
r
H=
2m
(Reminder:

r is

the operator for

r,

not the unit vector along the radial direction!) From Eqs. 1.1,1.2,

the following can be seen

[La , qb ] = [ acd qc pd , qb ]

= acd qc [d , qb ]
p
= acd qc db (i )

= i abc qc

(3.1)
Similarly,

[La , pb ] = [ acd qc pd , pb ]

= acd [c , pb ]d
q p
= acd pd cb (i )

= i abc pc

(3.2)

5The
abc

denition of this symbol is as follows.

bac (anti-symmetry).

two indices are the same,

(iv)

abc

=0

if

(i)

a = b.

123

= 1.

(ii)

abc is invariant under a cyclic permutation.

From this denition it follows that

(iii)

abc is zero if any of the

abc acquires a minus sign if any of the two indices are swapped.

L11  ANGULAR MOMENTUM

where in the last step,

c=b

is set rst using

cb

and then the dummy index

c.

is changed to

These imply

[La , r2 ] = [La , qb qb ]
q
= qb [La , qb ] + [La , qb ]b

bc

are dummy summation indices and

anti-symmetry of

= qb abc qc + abc qc qb

can be redened as cb = ( abc + acb )b qc

q
the Levi-Civita symbol = 0

The proof that

[La , p2 ] = 0
is totally unnecessary, since it is precisely the same as the one just given if one substitutes

q.

Thus,

La

Hamiltonian,

commutes with any function of

p2

2m

p2

and

r2

for

and thus with a spherically symmetric

+ V ().
r


[H, L2 ] = [H, Lx,y,z ] = 0

Example 3.3. Cylindrically Symmetric Potential and Orbital Angular Momentum

Conservation
Let us say that the symmetry is broken so that the system is symmetric on rotation around the

axis, but not symmetric on other rotations. We can write

p2

H=
+ V (, )
r
2m
r, are operators not unit vectors.) In this case, the Hamiltonian commutes with


z = i /, since H is independent of , but not with Lx , Ly , since Lx , Ly necessarily involve

pz , which involves a dierentiation on . Another way to see this latter point is that H is not

x drives the
invariant when changes: when = 0 and changes that is the process in which L

motion, when = /2 it is Ly , so just by considering these two cases it is obvious that Lx,y cannot

be conserved in general if V is a function of . You can do an explicit evaluation of Lx , Ly in
2
terms of and as well (Section 4.3.2 of Griths), if you are unsatised at this point. Lastly, L

2
2 z
2
commutes with H , since [L , cos()] = [L , ] = 0 because L commutes with both z and r . Thus,

2 commutes with any function of .

L



[H, L2 ] = [H, Lz ] = 0, [H, Lx,y ] = 0

(Reminder:

4.

Spin 1/2

Who Has Seen the Wind?

Christina Rossetti
Who has seen the wind?
Neither I nor you:
But when the leaves hang trembling
The wind is passing thro'.
Who has seen the wind?
Neither you nor I:
But when the trees bow down their heads
The wind is passing by.
The spin 1/2 angular momentum is fundamental in descrbing elementary particles, since all elementary
fermions have spin 1/2. It is dened as a dynamical variable in a two dimensional Hilbert space
satisfying the general angular momentum algebra, Eq. 2.1.

Unfortunately, [in the standard view]

we do not really yet understand how spin 1/2 arises microscopically (for instance, you are welcome
to do Prob. 4.25 and see the hint of why this is), so this is as concrete as we can get.

UCSC, PHYS 139A, SPRING 2008 GEY-HONG SAM GWEON

Then,

how

do we ever know what the spin is? The angular momentum, orbital or spin, gives rise to

the magnetic moment. We know the existence of spin 1/2, because when we apply magnetic eld to
align magnetic moments in atoms, we discover that some magnetic moment splits into two dierent
energy levels, which is impossible to understand if we use orbital angular momentum alone. In the
same vein, the Stern-Gerlach experiment  Example 4.4 of Griths  on certain atomic beams, like
an Ag beam, splits the incoming beam into

two

distinct outgoing beams, which can be understood

only if we invoke an angular momentum with degeneracy 2, that is spin 1/2. Thus, the above poem.
The Stern-Gerlach set up is shown below (http://www.modelofreality.org).

The essence of the

B = xi + (B0 + z)k , where i, k are unit

gure). Then, F =
( B) = z k x i.

Stern-Gerlach set up is an inhomogeneous magnetic eld,

vectors along the

axis and

Note that this classical thinking

axis, respectivley (see

6 does not include the quantum eect,the Larmor precession which

will be discussed later in this note. Here it suces to note that this precession can result in some
net spin along the

direction and this may result in some force in the

direction. The result is a

lip like image on the lm, where the central part corresponds to net force along the

direction

averaging out to zero, while the edge of the lip corresponding to the case when the net force
along the

direction is nite [Question for readers: why is the vertical separation reduced to zero

at the far, left or right, edge of the lip?].

Then, spin 1/2 is, by denition,

Sa =
where

a , a = x, y, z

or equivalently

x =
Note that I do not use

for

a = 1, 2, 3,

0 1
1 0

, y =

are the Pauli matrices.

0 i
i 0

Pauli matrices, but they

, z =

are

1 0
0 1

obviously QM operators, being matrices.

Since they cannot be confused with numbers, I will take hats from them to be ecient. I will still

Here, the classical approximation is justied since we are dealing with atoms. For Ag atoms, the typical energy

when it is hot is say, 100 meV, which translates to the momentum p =

2mE 2 108 940 106 0.1 eV/c
5
1.4 10 eV/c, where c is the speed of light. In contrast, if the uncertainty in position is say, 1 micron, then
p /1 micron 2 101 eV/c. And so, we see that p
p, justifying the classical approximation. So, that is
that for Ag atoms.

How about if one tries Stern-Gerlach for an electron?

that case as well, if the energy is high enough.

( B),

due to

qv B . Since there is a small component of B along the x

Fz evx x due to the uncertainty of the wave packet
be smaller than z due to
( B), for the Stern-Gerlach eect to be
Plugging in z = e /(2m) (see Eq. 4.3) and vx px /m, one sees that

but also is there the Lorenz force

direction, there will be a Lorenz force in the

along the
observable.

direction.

This force better

The ratio is

R 2xpx / 1!
oven?

This argument alone remains valid for

However, an electron is charged, and so, not only there is force

R = evx x/z .

direction,

So, the Stern-Gerlach eect is not possible for an electron. What if Ag atom is ionized in the
4
That is perfectly ne, due to the fact that vx px /MAg while z = e /(2m), and so R m/M 10
.

L11  ANGULAR MOMENTUM

keep

on spin though.

Many goody propertis of spin physics have been explored in Homework

7.For instance, they are Hermitian operators, as they should be, and also satisfy the commutation
relations

[a , b ] = 2i

abc c , so that

[Sa , Sb ] = i

abc Sc

satisfying the fundamental angular momentum commutation relations. Also, note that if we take

|z |1/2, 1/2 =

1
0

, |z |1/2, 1/2 =

0
1

then the phase convention of Eq. 2.7 is satised, since

S+ =

(x + iy )

0 1
0 0

S =

0 0
1 0

S |1/2, 1/2

|1/2, 1/2

S+ |1/2, 1/2

|1/2, 1/2

In the above, the notation is such that

|z = |+ = + = |1/2, 1/2

(4.1)
and

|z = | = = |1/2, 1/2

(4.2)

The spin wave functions, in whatever symbol they are represented by, mean column vectors in the
spin Hilbert space, and are generally referred to as

spinors.

The notable things about the spin 1/2 space is that wave functions do not return to itself on
rotation by

2 ,

but only on rotation by

4 .

(Homework 07)

The exact relationship between the spin 1/2 and the magnetic moment is

= S
where the

gyromagnetic ratio

the electron, where

(more standard notation is

B =

(4.3)

B = 9.27 1024

gB ),

is pretty close to

2B /

, for

is the Bohr magneton

J/T(SI) or

5.79 109

e
2m

eV/gauss. So, for an electron in a

eld, the Hamiltonian

is

H = B
= B B
(dene z axis as pararllel to

L B Bz

B) = B Bz z
= L z

is the freqeuncy at which the spin precesses in this problem, and this is called the

Larmor precession.

as we have seen in Homework 7. Note an important fact.

10

UCSC, PHYS 139A, SPRING 2008 GEY-HONG SAM GWEON

9
In the Larmor precession problem for spin 1/2, expectation values such as
with an angular velocity

L .

When

L t = 2 ,

the wave function does

not

Sx

and

Sy

6
rotate

return to itself. Rather

|(t = 2/L ) = |(t = 0)

This is a fundamental property of spin 1/2  you rotate the particle by 360
acquires the negative sign.
space as only

and the wave function

It is as though the spin space is double the size of the ordinary

rotation brings the wave function to itself !

Note

. The Complete Wave Function for the Hydrogen Problem

4.1

When we discussed the periodic table, we mentioned the spin angular momentum already. Now we
are at a potision to write down the total wave function, including the spin.

nlml ms (x) = Rnl (r)Ylml (, )sgn(ms )

= |n, l |l, ml |ms
|n, l, ml , ms
is dened as in Eqs. 4.1,4.2, |n, l is the ket dened in the radial wave
|l, ml is the ket dened in angular function space, representing Ylml , and |ms is
for |s = 1/2, ms used in Eqs. 4.1,4.2. A general wave function is then
where

function space,
the short-hand

Cnlml ms Rnl (r)Ylml (, )sgn(ms )

(x) =
n,l,ml ,ms

Cnlml ms |n, l, ml , ms

=
n,l,ml ,ms

In the pure-Coulomb problem that we have been dealing with so far, energy eigen-values depend
only on

and so we have

|(t) =

exp(iEn t/ )Cnlml ms |n, l, ml , ms

n,l,ml ,ms

How can one view this complete wave function which is an amalgam of the spatial part and the
spin part?

One can think of

as a

dependent function whose value at each

is a column

vector (spinor). Or, one can think of it as a spinor, whose component is a function of

x.

These

two views are totally equivalent.

5.

Addition of Angular momenta

Let us consider two angular momenta

J1

and

J2 .

Let us be very general and treat these two as

any angular momenta. By denition, they are assumed to describe completely dierent degrees of
freedom, which means

[J1,a , J2,b ] = 0

(5.1)
for any

a = x, y, z

and

b = x, y, z .

An example of such two dierent angular momenta can be

orbital angular momentum and spin angular momentum of a given particle.

Or, they could be

spin angular momenta of two particles of the same kind or dierent kinds. We consider the sum

J J1 + J2 .

The

component is

(5.2)

Jz = J1,z + J2,z

and the raising/lowering operators are

(5.3)

J = J1, + J2,

L11  ANGULAR MOMENTUM

and the length squared operator is, noting that

J1

and

J2

11

represent separate degrees of freedom

and so they commute,

2
2
J 2 = J1 + J2 + 2J1 J2


2
2

= J1 + J2 + 2J1,x J2,x + 2J1,y J2,y + 2J1,z J2,z

(5.4)
(5.5)
Noting that

J1,x =

1
2

J1,+ + J1,

and

J1,y =

J1,+ J1,

1
2i

and so on, we see that

2J1,x J2,x + 2J1,y J2,y = J1,+ J2, + J1, J2,+
and so

2
2



J 2 = J1 + J2 + J1,+ J2, + J1, J2,+ + 2J1,z J2,z

(5.6)

Questions arise. When we form the sum of two angular momenta, does the sum also satisfy the
fundamental commutation relations of the angular momentum, Eq. 2.1? The answer is yes. Let us
see:

[Ja , Jb ] = [J1,a + J2,a , J1,b + J2,b ]

(Eq. 5.1) = [J1,a , J1,b ] + [J2,a , J2,b ]

abc J1,c

abc J2,c

abc Jc

and thus

[Ja , Jb ] =
which is the same as Eq. 2.1 (the notation

abc Jc

abc is explained in Example 3.2). Given these fundamental

Jz commutes with J 2 . How about J1,z ? From

commutation relations, it follows immediately that

Eq. 5.5, it can be noted that the two terms in the middle has a nite commutator with

J1,z

and

that is

[J 2 , J1,z ] = [2J1,x J2,x + 2J1,y J2,y , J1,z ]

= 2[J1,x , J1,z ]J2,x + 2[J1,y , J1,z ]J2,y


= 2J1,y J2,x + 2J1,x J2,y

J1,z does not commute with J 2 and nor does J2,z by symmetry. Only the total component

Jz = J1,z + J2,z commutes with J 2 . How about J1 ? Applying Eq. 2.4 to J1 , one notices easily
2 as expressed by Eq. 5.4 or Eq. 5.5 commute with J 2 . By symmetry, it also commutes

that J
1
2 . Let us summarize. Before we do it, just one note on notation. Before the angular
with J2
momentum is added, one can consider all possible combinations of m1 and m2 and write the state
as |j1 , m1 , j2 , m2 |j1 , m1 |j2 , m2 . This juxtaposition of two states corresponding to two dierent
Thus,

degrees of freedom technically means that we are considering a tensor product, or a direct product,
Hilber space [cf. Denition 4.21 of L8].

J1

When one adds two angular momenta

characterized by

J2

separate components

|j1 , m1 , j2 , m2 add

Note that there are

and

2
2
Jz . J1 and J2 are
along x or y ) are not.

and

(2j1 + 1)(2j2 + 1)

J2 ,

the total angular momentum

compatible with

J 2,

but

J1,z

and

J2,z

J J1 + J2

2
is

(or any other

In terms of states labeled with quantum numbers

the two angular momenta

|j, m, j1 , j2

distinct states, before angular momenta are added up. After

they are added, there should be the same number of distinct states. It is easy to notice that there

m = j1 + j2 . This is the
j = j1 + j2 . For this j value, we

can keep applying J and nd all states |j = j1 + j2 , m, j1 , j2 . Now consider m1 + m2 = j1 + j2 1

states on the left hand side. There are two such states, since m1 or m2 , but not both, can
be lowered by 1 from their respective maximum values to give this m1 + m2 value. Notice that

is one state on the left hand side with

maximum value that

m1 = j1

and

m2 = j2 ,

giving

can ever have, and therefore correspond to

12

UCSC, PHYS 139A, SPRING 2008 GEY-HONG SAM GWEON

j = j1 + j2

state on the left hand side, and so there is one state left.
state for the
minimum

m1 + m2 = j1 + j2 1
must be the maximum m
all j values. What is the

state on the right hand side consumed one linear combination of

j = j1 + j2 1

state.

This state

Continuing this way, we can nd

value? This is determined by the condition that the number of distinct states should

be the same before and after angular momenta are added:

j1 +j2

(2j1 + 1)(2j2 + 1) =

(2j + 1)
j=jmin

= (j1 + j2 )(j1 + j2 + 1) jmin (jmin 1) + (j1 + j2 jmin + 1)

which gives a quadratic equation

2
jmin = (j1 + j2 )2 + 2(j1 + j2 ) 4jj2 2(j1 + j2 )

= (j1 j2 )2
which means

jmin = |j1 j2 |
So, here is another important nding.


(5.7)

j1 , j2 add

the two angular momenta

j = |j1 j2 |, |j1 j2 | + 1, ..., j1 + j2 1, j1 + j2

The discussion above that leads to the identication of

jmin

actually indicates how

|j, m, j1 , j2

can

be constructed. When we are done with that construction, we get a general relation such as

j1 j2 j
Cm1 m2 m |j1 , m1 , j2 , m2

|j, m, j1 , j2 =

(5.8)

m1 +m2 =m
The coecients

j1 j2 j
Cm1 m2 m

Clebsh-Gordan coecients.

are called

Some simple cases are tabulated in

Table 4.8 of Gritths, as a column vector. The reverse relationship also exists

|j1 , m1 , j2 , m2

j1 j2 j
Cm1 m2 m |j, m = m1 + m2 , j1 , j2

=
j

These coecients can be read o from Table 4.8 of Griths as a row vector.
of vectors are both natural basis sets
following identities hold within

that

for the subspace

where

j1

and

j2

These two sets

are xed, and thus the

subspace.

|j1 , m1 , j2 , m2 j1 , m1 , j2 , m2 | = 1
m1 ,m2

j1 , m1 , j2 , m2 |j1 , m1 , j2 , m2

= m1 m1 m2 m2

|j, m, j1 , j2 j, m, j1 , j2 | = 1
j,m

j, m, j1 , j2 |j , m , j1 , j2

= jj mm

Example 5.1. Sum of Two 1/2 Spins

Two electrons. One electron plus a proton. A quark plus an anti-quark to form a meson. All of
these examples correspond to this important example. Let us use the following notation, which is
quite standard.

|
|
|
|

|j1
|j1
|j1
|j1

= 1/2,
= 1/2,
= 1/2,
= 1/2,

m1
m1
m1
m1

= 1/2, j2 = 1/2, m2 = 1/2

= 1/2, j2 = 1/2, m2 = 1/2
= 1/2, j2 = 1/2, m2 = 1/2
= 1/2, j2 = 1/2, m2 = 1/2

L11  ANGULAR MOMENTUM

Also, it is standard, when it is not ambiguous, to leave out

j1 , j2

13
in

|j, m, j1 , j2

. Simply by looking

up Table 4.8, we can write

|11
|10
|1 1

= |
1
= (| + | )
2
= |

j1 = 1/2 and
j = 1 and j = 0. In the spin context, we say s1 = 1/2 and s2 = 1/2
The s = 0 state is a singlet (again from Table 4.8)
1
|00 = (| | )
2
that s = 1 state is where the two spins are parallel, while s = 0 state is where

These form the triplet states for the total spin 1 case. According to Eq. 5.7, adding

j2 = 1/2 should result

give s = 1 and s = 0.

One qualitatively says

in total

the two spins are anti-parallel.

As you can see from the wave functions, there is more-than-half

truth to this statement, and so this qualitative statement is made quite often.
Even if one did not look up Table 4.8, this case is simple enough so that we can follow the
procedure described just before Eq. 5.7. First, we can identify that

|11 = |

S = S1, + S2, to this. First note that, from what we obtained in Section 4, S1, | 1 = | 1
1,+ | 1 = | 1 , and similarly for S2, . Thus, we get

Apply
and

This should be by denition

S.

S |11 = (| + | )

(1 + 1)(1 1 + 1)|10 =
2|10

, using Eq. 2.7 for the total spin

Therefore, we get

Applying

S on this, we get

1
|10 = (| + | )
2
2 |

, which must be equal to

(1 + 0)(1 0 + 1)|1 1 =

2 |1 1

So we get nally

|1 1 = |
For the
to

|11

|00

state, we need to make it a linear combination of

| , |

and also make it orthogonal

. Thus we get (with a sign convention),

1
|00 = (| | )
2
For your condence, you can apply Eq. 5.6 to these states and double check that they are what
you think they are in terms of

S 2 |10

=
=
=
=

S.

For instance,

1
2 2



S1 + S2 + 2S1,z S2,z + S1,+ S2, + S1, S2,+ (| + | )
2
1
1
2
2

S1 + S2 + 2S1,z S2,z (| + | ) + S1,+ S2, + S1, S2,+ (| + | )
2
2
1
1
3 3
2
+ 2
|10 + 2 (| + | )
4 4
4
2
2
2|10

= 1(1 + 1) 2 |10

Example 5.2. Strong Zeeman Eect

Cylindrical symmetry results if an external

eld is applied. In this case, (see Eq. 6.71 of Griths)

HZ =
(L + 2S) B
2m

14

UCSC, PHYS 139A, SPRING 2008 GEY-HONG SAM GWEON

must be added to the Hamiltonian.

Jz = Lz + Sz .

HZ (J + S) B .

Note that

2
Also, HZ commutes with L and

HZ commutes with Lz and Sz but not with

2 but not with J 2 , since it is in the form of

So the following table can be obtained, which is an update of the table of Section

1.
Hamiltonian

Symmetry

p2

2m

+ V ()
r

H0

Conserved Angular Momentum

Spherical

(vanilla Hydrogen)

H0 + HZ

HZ (L + 2S) B

L, S : cylindrical

J : none

J:

Stationary State
(Dm : minimum degeneracy)

spherical

d
dt

(spin orbit)

I2

|jmj

|ml ms
mj , ml , ms
Dm = (2l + 1)(2s + 1)
|ml ms but not |jmj
or

is independent of

d 2
d
=0
dt Iz = dt I

for I = L, S only


J 2 , Jz , L2 , S 2 but not Lz , Sz

(strong Zeeman)

H0 + Hso

Hso L S

Angular Momentum Part of

(J L + S )

L: orbital, S : spin, J : total

J 2 , L2 , S 2 , Jz , Lz , Sz
d 2
d
=0
dt Ix,y,z = dt I
= L, S, J

for any I


L2 , S 2 , Lz , Sz but not J 2 , Jz

is

dependent

on

ml , ms

|jmj

Jx,y,z = 0

= 0 for any I = L, S, J
d
dt

is independent of

mj

Dm = 2j + 1

Example 5.3. Spin Orbit Interaction

The spin orbit interaction occurs for an electron with non-zero angular momentum. Qualitatively, it
is easy to understand why it occurs. From the electron's reference frame, the nucleus is going around
the electron, with an angular momentum which is opposite to the electron's angular momentum
around the nucleus.

So, the nucleus moves in a current loop and thus has a magnetic dipole

moment. The electron at rest at origin of this reference frame has a magnetic dipole moment as
well due to spin,

= 2B S .

Two dipole moments interact via dipole-dipole interaction. So, the

spin orbit interaction has the following form:

Hso = C L S
where

is a positive constant, and

L, S

momentum of the electron. [For details of

easy to gure out that

are the orbital angular momentum and the spin angular

C,

look it up in any QM book, including Griths. It is

gets large for small

and large

Z,

the atomic number.]

note that the Hamiltonian breaks the spherical symmetry for both
of these vectors, aligning each other, are
symmetry for
either

Jz

Lz

or

Sz .

and

S.

dynamic,

In this case,

Moreover, since both

the problem does not even have a cylindrical

Hso

LS =

Another way of saying/seeing this is that

On the other hand, as we can see from Eq. 5.4,

does not commute with

1
2

J 2 L2 S 2

and so

is a good quantum number, since

1 2 2 2
[Jz , J L S ]
2
1 2


=
[Jz , J ] [Jz , L2 ] [Jz , S 2 ]
2
1


=
0 [Lz , L2 ] [Sz , S 2 ]
2
= 0

[Jz , L S] =

(S

and

S.

and

commute as they belong in dierent Hilbert spaces)

This situation is summarized in the last row of the above table.