Professional Documents
Culture Documents
and
ih
r
ih
x y z
xyz
(2)
(3)
ihr
ih y
(4)
ypz
zpy
Ly
zpx
xpz
Lz
xpy
ypx
ih z
x
ih x
y
x
z
y
x
(5)
(6)
(7)
First we note that , as bets an observable dynamical variable, L is a Hermitian operator because both r
and p are Hermitian,
Li Li
i 123
(8)
Next we study the commutation relations between the three components of the angular momentum operator using the canonical commutation relations. These state that coordinates commute with alien components of the momentum operator, but do not commute with their own, xi p j ihi j , and that coordinates
0, pi p j
0:
commute with coordinates and momenta with momenta, xi x j
1 This
topic is quite long and complicated. I recommend that you rst look at the summary section at the end, p.21, and
then work through the material several times, looking through the summary at regular intervals. Your primary aim should be
to understand the results given in the summary. The detailed derivations and discussion are to create that understanding by
showing where the results come from. There are, of course, several techniques and concepts of considerable importance in the
derivations, so they should not be skipped.
2 I remember the cross-product by drawing a circle in my mind with the components x y z arranged clockwise. The components of the cross-product, reading from left to right in the equation, always occur in cyclic order (ie. clockwise order) for the
rst (positive sign) term and anti-cyclic for the second (negative sign) term. Thus, for the z-component Lz , you start the cyclic
y, giving Lz xpy ; the anti-cyclic one is z y x, giving Lz xpy ypx . This is just
sequence with z, giving z x
another way of expressing the rule based on the determinant, but its easier to keep in your head without the need to write down
any intermediate steps.
[b
Lx ; b
Ly ]
[(y p
bz
z pby ); (z pbx
bz ; z p
bx ]
[y p
[y p
bz ; x p
bz ]
y pbx [ pbz ; z]
(x p
by
ih (x pby
[z p
by ; z p
bx ] + [z p
by ; x p
bz ]
[z; p
bz ] = ih
= ih b
Lz :
(9)
ih b
Lz ;
(10)
[b
Lz ; b
Lx ] =
ih b
Ly ;
(11)
[b
Ly ; b
Lz ] =
ih b
Lx :
(12)
Notice that the commutators give +ih when, reading from left to right, the components are in cyclic order
just as in a cross product. Indeed, these commutators can be written in an even more succinct form as a
cross product:
bL
b:
b = ih L
L
(13)
This form of the commutators shows how unintuitive non-commuting operators can be: the cross product
of two parallel ordinary vectors vanishes, but not when they are operators and their components dont
commute!
The physical consequences of this failure to commute are remarkable and follow from the generalised
Heisenberg uncertainty relation
6=
A B
1 b b
jh[A; B]ij;
2
b; B
b] 6= 0:
0
if [A
(14)
(15)
First,3
bz ]
[b
Lx ; L
2
b
bx; L
bz ]b
Lx [b
Lx ; b
Lz ] + [L
Lx
byb
ih(b
Lx b
Ly + L
Lx ) using
=
3 Using
bz ] =
[b
Lx ; L
bBbAb from
the identity obtained by adding and subtracting A
b2 Bb]
[A
=
=
=
=
b2Bb BbAb2
A
ihb
Ly ;
(16)
Similarly,
bz ]
[b
Ly ; L
2
by [b
L
Ly ; b
Lz ] + [b
Ly ; b
Lz ]b
Ly
bxL
by );
+ih(b
Ly b
Lx + L
using
[b
Ly ; b
Lz ] = ihb
Lx
(17)
(18)
b
by ] + [L
b2 ; b
Ly ] = [b
L2 ; b
Ly ] + [b
L2 ; L
Ly ] = 0
x
(19)
Similar calculations for the other components give the same result, which we can summarise as:
b
[L
b
L i ] = 0;
for all i = 1; 2; 3
(20)
b 2; L
bz ] = 0 (i = 3), which shows
Since we have chosen to measure b
Lz exactly, the relevant commutator is [L
2
b
that we can also measure L exactly: it is possible to measure exactly both the length and one component of the angular momentum vector, but at the price of knowing nothing whatever about the other two
components. This can be illustrated with Figure 1, which shows a vector of known length and projection
onto the z-axis, but which can have any position on the circle obtained by rotating the vector around the
b 2 , but not
bz are also eigenfunctions of L
z-axis. In formal terms our result tells us that eigenfunctions of L
bz rather than b
Lx nor of b
Ly . Of course the choice of L
Lx or b
Ly is an arbitrary convention; the point is that
of b
only one may be chosen from the three components because the corresponding operators do not commute.
6AKA
hbLzi
A
A
?
x
q
b2
L
A
A
A
h i
-y
Figure 1: Picturing what can be measured exactly. This quasi-classical picture is not to be taken too
b 2 i are the eigenvalues of the corresponding operators. L is a
bz i and hL
literally - for an eigenstate hL
quasi-classical vector representing the quantum mechanical angular momentum.
b IN SPHERICAL POLARS. 4
2. REPRESENTING L
In many physical applications, such as the quantum mechanics of the hydrogen atom, we either have
exact or approximate spherical symmetry. This, together with angular momentums association with
rotation, demands that we consider writing down the angular momentum operators in spherical polar
coordinates; this also leads to a great simplification and a recognition that the orbital angular momentum
operators only involve the angular variables and .
In cartesian coordinates the orthogonal x ; y , and z axes, are fixed and lie along the orthogonal
unit vectors ex = i; ey = j, and ez = k; but in spherical polar coordinates the unit vectors, er ; e , and
e , although orthogonal, vary in direction with the direction of the position vector r as may be seen
by inspecting Figure 2. Any vector may be written as an expansion in terms of these socalled basis
vectors:
V
(21)
er Vr + e V + e V
(22)
ix+jy +kz
(23)
er r
(24)
=
=
+j
+k
;
x
y
z
1
1
er + e
+ e
:
r
r
r sin
i
(25)
(26)
Of course you could just look up this last expression in a maths text book, but the following argument shows that it is almost obvious: The gradient involves differentials with respect to the 3 orthogonal infinitesimal distances: dx; dy; dz for cartesian coordinates and, as inspection of Figure 3 shows,
dr; rd; r sin d for spherical polars.
In spherical polars the momentum operator is therefore
b
p =
=
=
ih
(27)
er pbr + e pb + e pb ;
1
1
ih er + e
+ e
;
r
r
r sin
(28)
(29)
ih
1
;
pb = ih
;
pb =
r
r
ih
1
:
r sin
(30)
To find an expression for the orbital angular momentum operator we need the cross product of the vector
r = rer with the momentum operator. This brings in the cross products of the unit vector er with all three
4 You
may find this section somewhat intimidating at a first reading. First read through the summary Sections 5 at the end of
this chapter then work through the material several times, referring to the summary at the end of each reading. If you still feel
its beyond you then just skim the material and take the results on trust. As long as you understand the meaning of the results
in the summary you should be able to follow the subsequent developments.
orthogonal unit vectors. These can be found by inspecting Figure 2 and using the righthand rule for the
crossproduct:
er er = 0;
(31)
er e = e ;
er e =
(32)
e ;
(33)
making it easy to obtain the angular momentum operator in spherical polar coordinates,
b := r b
L
p
(rer )
r (e pb
e pb )
1
ih e
e
;
sin
(34)
Where we have used our previous results for the cross products and the momentum operators. This
expression immediately shows the important property alluded to above: the orbital angular momentum operator only acts on the angular variables. Thus for a potential not depending on direction,
ie. a central potential with V (r) V (r; ; ) = V (r), the angular momentum commutes with the
potential:
b ; V (r)] = 0
For a central potential:
[L
(35)
which means that each component and any power of the angular momentum commutes with the potential:
For a central potential:
and
[b
Li ; V (r)]
i = 1; 2; 3
b 2 ; V (r)] = 0
[L
(36)
(37)
by ; b
Lx ; L
Lz
We conclude this section by giving expressions for the orbital angular momentum operators b
2
b in spherical polars. For this we need to find the projections of the unit vectors er ; e and e onto
and L
the x ; y ; and z axes. The details are given in Appendix A and Figure 4.
er
(38)
(39)
k sin :
i sin + j cos :
(40)
Having established the projections of vectors er ; e and e onto the x ; y ; z axes we can simply rewrite
the angular momentum vector in eq. (34), using the above equations, as:
b
L
=
=
=
1
;
sin
sin
cos
cos
cos
sin
ih i
sin
+ j cos
+k
;
sin
sin
sin
ih
i sin + cot cos
+ j cos
cot sin
+k
;
(41)
ih e
b
Lx
b
Ly
b
Lz
ih cos
cot sin
ih
(42)
(43)
(44)
(lever
(r sin )( p
b )
(r sin )(
ih
y plane)(b
p projected onto x
ih
y plane)
1
)
r sin
2
1
sin
sin
1 2
+
sin2 2
(45)
This is a remarkable result: apart from the factor h2 , and a missing 1=r2 , this is just the angular part of
the delsquared operator in spherical polars,
=
=
1
1
1
1 2
2
+
+
:
r
sin
r2 r
r
r2 sin
sin2 2
2
2
1
1
1 2
+
sin
+
+
:
r2 r r
r2 sin
sin2 2
(46)
(47)
b
p2
(48)
2m
h2 2
2m
(49)
we immediately see that this can be written in terms of the squared angular momentum as
c =
KE
h2
2m
2
r2
2
+
r r
+
b2
L
2m r2
(50)
In the following section we explain how we might have guessed this simple result from a knowledge of
the classical kinetic energy.
(51)
(52)
Looking at Figure 5 we see that the lever arm of p? about the origin is r so that the magnitude of the
orbital angular momentum is
Lcl = r p? ;
p2? =
and therefore,
L2cl
;
r2
(53)
Hence we easily obtain the classical Kinetic energy as the sum of linear and rotational energies:
(KE )cl =
p2
2m
(pr + p? )2
2m
L2cl
p2r
+
2m 2mr2
(54)
where we used p2r = pr :pr = p2r . To obtain the QM form one might be tempted to simply replace pr and
Lcl by the quantum mechanical operators. This would be wrong: the quantum mechanical calculation is
altogether more tricky because the two momentum operators dont commute,
b
b? :p
br ;
pr :b
p? = p
with
b
pr :b
p? = 0
but
b
p? :b
pr =
2ih
pbr
r
6= 0
(55)
This shows itself through the fact that although all the unit vectors are independent of r, they all vary
p? :b
pr . Recalling
with the angles and . 6 In Appendix B we show how to calculate this additional term b
the expression for the r-component of the momentum operator,
pbr =
ih
and hence,
pb2r = h2
2
r2
(56)
=
=
=
b2
p
2m
b2
pb2r
pbr
L
ih +
2m
mr 2mr2
b2
h2
2
2
L
+
+
2m r2 r r
2m r2
h2 1
r2
2
2m r r
r
b2
L
2m r2
(57)
(58)
(59)
(60)
This expression shows a close resemblance to the classical expression for the KE, eq. (54): the additional
term, ih pbr =mr, is clearly of quantum origin because it has an explicit factor of h even before we use a
particular representation for the operator. It arises from the non-commuting operators and corresponds to
the extra term (2=r)=r which we failed to discover by unthinkingly translating the classical formula
the momentum vector along r is pr = er pr , and the one perpendicular to r is p? = e p + e p
Appendix A for a full discussion
5 Where
6 See
into operator form. This expression will be important in our study of the hydrogen atom. For the present
we note that the Hamiltonian now takes the form,
b=
H
h2
2m
2
r2
2
+
r r
+
b2
L
2m r2
+ V (r)
(61)
This form for the Hamiltonian in any central potential immediately demonstrates that it commutes
b 2 . This is because b
b 2 , (a) only act on the angular
Li and L
Li and L
with all the angular momentum operators b
b2
variables ; , and therefore do not effect any rdependent terms, and (b) commute with the operatorL
2
2
b =2mr and other r-dependent terms in the Hamiltonian. Hence we find
and hence they commute with L
the important results,
b; L
bi ] = 0
[H
i = 1; 2; 3
(62)
(63)
EE (r);
2
`(` +
(64)
1)E (r)
h mE (r) m =
`
`;
= 0; 1; 2; : : : ;
`+
1; : : : ; 0; : : : ; `
(65)
1; `:
(66)
(We will show later that there is also the possibility of ` values 1=2; 3=2; 5=2; : : : for angular momentum
in general, including spin, but not for orbital angular momentum.) From the above summary we see that
the electron wave function should not only carry the energy label E ( or equivalently the quantum number
n, as in the infinite square well, harmonic oscillator or the hydrogen atom), but also those corresponding
to the angular momentum, E m , or n m .
There are two approaches to demonstrating the quantisation of angular momentum:
b 2 and b
Lz . I will motivate this approach
APPROACH (1), by seeking directly the eigenfunctions of L
physically by showing how the eigenvalue equations arise in seeking the energy eigenfunctions for a
particle in a central potential. We will apply this later to our study of the hydrogen atom.
APPROACH (2), by treating the angular momentum operators as abstract entities defined entirely by
their commutation relations. It is this approach which reveals the possibility of non-integer quantum
numbers ` and m and which enables us to handle the purely quantum mechanical phenomenon of spin.
;`;
;`;
ie.
b
Lz m () = mh m ();
(67)
m ()
(68)
ih
mh m ();
(69)
but there is nothing yet to tell us the value of m, the azimuthal quantum number; for this we need to
introduce some further physical information.
b 2 , eq. (45), operates on both and , so that its eigenfunctions depend on both
In contrast, the operator L
variables; we will also find that the eigenfunctions depend on m as well as the L2 eigenvalue `:
b 2 Ylm (; ) = h2 `(` + 1) Ylm (; )
L
(70)
Again, at this stage, there is nothing to tell us the value of the quantum number `.
To introduce the physics we consider a particle in a central potential V (r) and we proceed to solve
the TDSE in the obviously appropriate spherical coordinate system:
b (r; ; ) E (r; ; ) = E E (r; ; );
H
(71)
h2
2m
2
2
+
2
r
r r
b2
L
+
+ V (r)
2m r2
E (r; ; ); = E E (r; ; )
(72)
h2
2m
2
2
+
r2 r r
h2
1
sin
2m r2 sin
+
1 2
sin2 2
+ V (r )
= E E
(73)
Separation of variables: To solve this equation we first notice that the angular dependent terms are
very nearly separated from the r dependent ones; all we have to do is multiply by r2 to achieve the full
separation. This suggests that a way to find the solutions is to try to find one in the form
E (r; ; ) = R(r) Y (; )
(74)
r2
h2
R(r)
d 2 R(r) 2 dR(r)
+
dr2
r dr
V (r))2m r2 =
(E
h2
Y (; )
1 b2
L Y (; )
Y (; )
Y (; )
1
sin
sin
1 2Y (; )
+
sin2 2
:
This equation must hold for all values of the three independent variables r; and ; but the the left side
is only a function of r, the right only a function of and . They can only possibly be equal for arbitrary
9
choices of r; and if they dont vary at all, ie. they are constant. We call this constant, which has the
dimensions of h2
h2 `(` + 1)
for convenience, where ` is an unknown dimensionless constant.7 Thus we obtain the radial equation
whose solution depends on the precise form of the potential,
h2
2m
d 2 R(r) 2 dR(r)
+
dr2
r dr
h2 `(` + 1)
V (r ) +
2m r2
R(r) = E R(r);
(75)
1
sin
sin
1 2
+
Y (; ) = h2 `(` + 1) Y (; ):
sin2 2
(76)
Something remarkable has happened: the differential operator which operates on the function Y (; ) is
b 2 , and this equation is nothing but the statement that Y (; )
none other than our recent acquaintance, L
b 2 , with the corresponding eigenvalue being
is an eigenfunction of the angular momentum operator L
2
the constant h `(` + 1):
b 2 Y (; ) = h2 `(` + 1) Y (; )
L
(77)
Remember that at this point we know nothing about the numerical value of `. Incidentally, its not
b 2 commutes with the
too surprising that this equation occurs in this problem: for a central potential L
Hamiltonian, and therefore eigenstates of energy - in this case E = R(r)Y (; ) - are also eigenstates
b 2 . We now anticipate that the wave function is also an eigenstate of b
Lz because this operator also
of L
b
commutes with H. To see that this is so we now separate the eigenvalue equation for Y (; ) by trying a
solution of the form:
Y (; ) = () ():
(78)
Substituting into the equation, multiplying by sin2 = and isolating the dependent terms on the
right we obtain the separated form,
d
d()
1
sin
sin
()
d
d
2
+ `(` + 1) sin
1 d 2 ()
:
() d2
(79)
Once again we see that this equation must hold for all values of the independent variables and , so
that the left and right sides must be a constant. Calling this as yet unknown dimensionless constant m2 ,
we obtain for the equation:
d()
1 d
sin
sin d
d
+
`(` +
1)
m2
() = 0:
sin2
(80)
This is the associated Legendre equation whose finite solutions, known as the associated Legendre
functions, are well-studied and are labelled by the values of the two constants ` and m:
() = Pm (cos ):
`
d 2 ()
d2
m2 ():
(81)
(82)
(83)
7 At this stage we have no idea whatever what the value of ` is and we are just anticipating the form that will come out of
subsequent discussion. The discussion would go exactly the same way if we called this constant A, , or anything else.
10
(84)
d 2 ()
d2
= h
m2 ();
(85)
(86)
(87)
Lz has quanThus do we obtain the quantisation of the z component of orbital angular momentum: b
tised eigenvalues hm
; m = 0; 1; 2; 3; : : : .
We now return to the equation which can be simplified by using the variable w = cos :9
(1
d2
w) 2
dw
2
d
2w
+ `(` + 1)
dw
m2
jmj
P ( w) = 0
1 w2
`
(88)
This can be solved by the series method: as with the harmonic oscillator, it turns out that a physically
acceptable wave function, finite at cos = 1, only results if the series terminates after a finite number
of terms. This imposes a relationship between ` and m:
`
0; 1; 2; : : :
`;
`+
1; `:
(89)
p
11
We have now shown that the wave function depends on the energy and on the quantum numbers `; m;
in fact if we look at the equations satisfied by R(r) and Y (; ) we see that the former depends only on
E and `, R(r) = RE (r), while the latter does not involve E at all, but does depend on both ` and m,
Y (; ) = Y m (; ). Thus we should write the energy eigenstates not as E (r) but as:
`
jE
j2 d 3x = 1
m (r)
R E (r ) Y
m (r) =
m (r)
m (; ):
(90)
(91)
r2 dr
Z 2
d
0
sin d jE
j2 = 1
m (r)
(92)
A convenient shorthand for the angular integration variable is the solid angle notation, d sin d d,
with the accompanying shorthand for the integration itself:
Z
d
Z 2
d
0
sin d
(93)
jY m ( )j2 d = 1
;
(94)
m (r)
(95)
jRE (r)j2 r2 dr = 1
(96)
As expected for eigenfunctions of Hermitian operators, the spherical harmonics are orthonormal:
Z
Y m (; ) Y m (; ) d = m m
0
` `
(97)
Moreover the most general solution to the TISE equation is a linear combination of spherical harmonics:
they constitute a complete set of functions for which the expansion theorem holds. Thus the angular part
L2 and b
Lz :
of any wave function may be expanded in terms of eigenfunctions of b
(r; ; ) =
`=
11 The
+`
0 m=
m E ` (r )Y`m (; );
`;
(98)
(2
`+
1)(` m)!
4(` + m)!
1=2
12 This is important for the physical interpretation of plots of the radial wave function: one normally plots jr R (r )j2 because
E`
this represents the physical probability density at a distance r from the origin. Of course we must remember that this is
then modulated by the direction-dependent angular wave function jY`m (; )j2 factor to give the full spatial dependence of the
probability density.
12
where the coefficients c m play the same role as the cn in our general discussion of the expansion theorem.
Normalisation of this state requires,
`;
+`
jc m j2 = 1
`;
`=
0 m=
(99)
which tells us that the jc m j2 are probabilities. Thus, for a system in the normalised state Y (; ) a
measurement of both the square and z component of orbital angular momentum yields, with probability
jc mj2 ; one of the eigenvalues h2`(` + 1) of Lb 2 and one of the eigenvalues, hm
; m m = `; ` +
b
1; : : : ; ` 1; `, of Lz corresponding to that ` value; immediately after the measurement which gave the
values (`; m) the wave function will be Y m (; ). If only the squared angular momentum is measured,
yielding h2 `(` + 1), then the probability for this outcome is the sum over the possible, but unmeasured
2
m values, +
m= jc m j .
If we are concerned with a state of definite total orbital angular momentum, then it must be an
b 2 , fixing the value of the quantum number `. This is a frequently used special case of the
eigenstate of L
more general expansion theorem in eq. (98). The angular part of the wave function would then be:
`;
`;
`;
Y(
`) (; ) =
+`
m=
cm Y m (; );
`
+`
where
jcm j2 = 1
m=
(100)
13
jY m ( )j2
;
jmj j2 jeim j2
N 2m jP
N 2m P
jmj i2 ;
(101)
we see that all systems with a central potential have rotational symmetry about the z axis. Put another
way, a plot of the angular probability distribution can always be pictured as a 2-dimensional figure rotated
about the z axis; examples of the first few are shown in Figures 8a and 8b. The first three angular
momentum eigenfunctions are also tabulated in Table 1; notice that the states have definite parity which,
being ( 1) , alternates in a nearly familiar manner.13 Of course this is another example of our theorem
that eigenstates of energy are also eigenstates of parity if the potential is symmetric, V ( r) = V (r). This
is trivially satisfied for a central potential because under a mirror reflection the length r of the vector r
doesnt change.
`
(r; ; )
leading to,
! (r
;
; + )
giving
sin ! sin ;
cos !
14
cos ;
Table 1.
Quantum
Numbers
Parity = ( 1)
`
m
Associated Legendre
Function
P
m (r) =
R E (r ) Y
`
Azimuthal
Function
Spherical Harmonic
eim
Y m (; ) = N m P eim
Y0 0 =
1
4
cos
Y1 0 =
3
4
1
sin
ei
Y1 1 =
Y2 0 =
m (; ):
jmj (cos )
1
2
2 (3 cos
1)
3 sin cos
ei
Y2 1 =
2
3 sin2
e2i
Y2 2 =
Y3 0 =
ei
Y3 1 =
1
3
2 (5 cos
3 cos )
1
7
16
3
2
2 sin (5 cos
2
15 sin2 cos
e2i
Y3 2 =
3
15 sin3
e3i
Y3 3 =
12
15
8
1
2
sin ei
(3 cos2
12
12
12
1)
sin2 e2i
(5 cos3
21
64
105
32
cos
3
8
15
32
1
1)
5
16
1
1
jmj
12
35
64
12
3 cos )
sin (5 cos2
1) ei
sin3 e3i
11
Notes: (a) The mysterious in the last column are conventional phase factors - see
p footnote .
(b) The associated Legendre functions are indeed functions of cos since sin = (1 cos2 ).
(c) The normalised components of the wave function for a given ` are:
`m (cos ) =
2N m P
`
jmj (cos )
+1
m0 (w)`m (w) dw = m0 m ;
`m () m ().
Z 2
m ()m ()d = m m ;
0
Y m (; ) Y m (; ) d = m m :
`
15
(103)
b2 ; B
b[A
b; B
b; B
b to obtain:
b] = A
b] + [A
b]A
For each term we use the identity [A
[Jbx ; Jbz ]
2
Similarly,
2
[Jby ; Jbz ]
(104)
(105)
while, finally and obviously, Jbz commutes with its own square, [Jbz2 ; Jbz ] = 0. Thus the two non-vanising
J2 indeed commutes with Jbz . Similar proofs go through for the other two
terms cancel, showing that b
components, giving finally:
[b
J2 ; Jbi ] = 0; for all i = 1; 2; 3
(106)
J2 and Jbz commute they have a common eigenfunction , with as yet unknown eigenvalues h2
Since b
and h
z respectively:
b
J2 = h2
(107)
Jbz = h
z
(108)
We now proceed to show that the commutation relations alone imply the quantisation rules obtained
before for orbital angular momentum, but with a bonus: half-integral angular momentum is also possible.
The analysis is divided into several distinct steps:
16
Step 1:
jzj
(109)
jzj
(110)
Notice that z can have any sign, but is positive because its the eigenvalue of the square of an operator.
This result is not surprising when we refer to the quasi-classical Figure 10, since the projection of the
angular momentum vector onto the z axis can never be larger than the length of the vector.15
z
6
b
J2 = h2
jzj
p
bz = h
L
z
h
z
h
?
x
-y
p
b is just the expectation value of that operator when the system is in the
that the eigenvalue of an operator O
corresponding eigenstate:
Z
Z
= ;
hObi = Ob dx = dx for an eigenstate, O
14 Remember
15 Deeper
thought reveals that the result is more subtle: the quasi-classical picture does no more than depict the content of
the eigenvalue equations. That the eigenvalues do indeed behave this way is true but not trivial to prove. In the case of spin
one has to prove that the expectation value of a squared operator is positive without knowing anything about the nature of the
coordinates upon which the wave function depends - indeed such coordinates may not even exist for intrinsic properties like
spin. However if the wave functions are dependent on the spatial variables, as in the case of orbital angular momentum, then
its clear why the square of a Hermitian operator has positive expectation value:
Z
Z
b ) Ob dx since Ob is Hermitian
hOb2 i = Ob2 dx = (O
Z
=
jOb j2 dx 0; since its the integral of a positive quantity.
The matrix or Dirac formalism is needed to prove this result when, as for spin angular momentum, we do not know what is the
analogue of the coordinates x; y; z or r; ; . That is the great merit of Heisenbergs matrix form of QM and why we will use it
later for dealing with spin.
17
Jb .
As we discovered in solving the SHO, the introduction of raising and lowering operators is the key
to the analysis. They are simple linear combinations of Jbx and Jby :
Jb Jbx iJby
Jb+ = Jb :
with
(111)
b
J2
Jbz2 +hJbz
(112)
Jb Jb+
b2
(113)
Jb Jb
b
J2
Jbz2
Jbz2
Jb+ Jb
or, more succinctly,
leading to:
and
[Jb ; Jb ] =
2hJbz
[Jbz ; Jb ] =
hJb
hJbz
hJbz
(114)
(115)
(116)
Jb ] = 0:
(117)
=
=
b
J2 (Jb )
(118)
The proof shows why Jb are called raising and lowering operators. We use their commutators with
(119)
to find
Jbz (Jb )
=
=
16 Thus,
h
z (Jb ) hJb since Jbz = h
z :
(120)
for example,
Jb+ Jb
=
=
b b b iJby )
2
b
Jx + Jby2 i(Jbx Jby Jby Jbx )
(Jx + iJy )(Jx
18
Collecting the (Jb ) terms on the right together gives the result:
Jbz (Jb ) = h(z 1) (Jb )
Jb+ max = 0
6
Jb+
Jb
?
Jb min = 0
Jbz Eigenvalue
Eigenfunction
(121)
b
J2 Eigenvalue
h max
z
max
h2
h(z + 2)
(Jb+ )2
h2 )
h(z + 1)
Jb+
h2
h z
h2
h(z
1)
Jb
h2
h(z
2)
(Jb )2
h2
min
h2
h min
z
Figure 11: Illustrating the action of the raising and lowering operators on the eigenstates of Jbz
and b
J2 . All are eigenstates of b
J2 with eigenvalue h2 .
Figure 11 depicts the action of these ladder operators: a single application of Jb+ to an eigenstate raises
repeated applications simply create eigenit up the ladder, increasing its Jbz eigenvalue by one unit of h;
b
states progressively higher up the ladder. J performs the same operations down the ladder, decreasing
the eigenvalue by one unit of h at each application. Can we continue the process
p indefinitely? No, beb
cause Step 1 showed us that the magnitude of the Jz eigenvalue cannot exceed h .
Step 5: Hence, there must exist on the ladder both a highest eigenstate, max
.
and a lowest one, min .
Just as for the ground state of the SHO, so these states must be annihilated by the action of the
appropriate ladder operator, thereby blocking any further progress on the ladder:
Jb+ max
0;
(122)
Jb min
0:
(123)
These states, being eigenstates of Jbz , have the maximum and minimum allowed eigenvalues:
Jbz max
Jbz min
max
z max ;
(124)
min
z min :
(125)
19
Jb (Jb+ max ) = 0
= (b
J2
=
Jb2
z
max 2
( z )
h (
2
(126)
hJbz )max
(127)
max
z )max
(128)
max
max
+ 1)
z (z
(129)
min
min
z ( z
1)
(130)
max
z
j;
(131)
where, because of its pivotal role in the argument we introduce a more succinct notation for the maximum
eigenvalue of the hJbz operator,
max
j
(132)
z
J2 operator is:
so that the eigenvalue of the hb
h2 = h2 j( j + 1)
(133)
(
j
+
n
)
.
But
we
know
steps we have reached the state (Jb+ )n min which has Jbz eigenvalue h(min
z
that this climb must end, and that the end must be the state of maximum eigenvalue max because this is
the defining property of max .18 Thus there is some n for which the resulting eigenvalue is h j, giving
n
j+n = j
or, j = ;
n = 0; 1; 2; : : :
(134)
2
We therefore discover two possibilities, integral (for n even) and half-integral (for n odd) angular momentum (in units of h),
whereas for orbital angular momentum only integral values were possible. Writing
for the eigenvalues z m j we then have the quantisation rules for general angular momentum:
Integral angular momentum: j = 0; 1; 2; : : : mj =
Half-integral angular momentum: j =
j;
1 3 5
; ;
;::: mj =
2 2 2
j + 1:::0:::; j
j;
j +1:::; j
1; j
1; j
(135)
(136)
20
h2 j( j + 1) j m j
(137)
Jbz j m j
hm
j j mj
(138)
The angular momentum quantum numbers j and mj labelling an eigenstate are often loosely referred to
as the angular momentum and its z component, or azimuthal projection, although strictly the angular
momentum is h j and hm
j . A special case of integral angular momentum is obviously the already encounb which
b we shall see later that spin angular momentum, b
J = L;
J = S,
tered orbital angular momentum b
cannot be pictured classically as a microscopic spinning top, can have both integral and half-integral
values.
Jbz Eigenvalue
m jh
j mj
+ j
h
max = j
+2
h
Jbz Eigenvalue
Eigenfunctions
j mj
max = j
+h
+ j
h
j 3
+ 32h
j 1
+ 2h
1=2
1
2h
3=2
3
2h
j 1
6
Jb+
; =
j 0
0
Jb
?
j 2
; =
6
Jb+
m jh
1h
j
2h
jh
Jb
?
min = j
min = j
Angular Momentum j = 0; 1; 2; : : :
jh
Angular Momentum j =
1 3 5
2; 2; 2;:::
21
5. SUMMARY.
b & ITS COMMUTATORS.
5.1 ORBITAL ANGULAR MOMENTUM L
ih r ;
(139)
bx
L
y pbz
z pby =
b
Ly
z pbx
x pbz =
b
Lz
x pby
y pbx =
ih y
z
ih z
x
ih x
y
;
z
y
x
;
z
y
:
x
(140)
(141)
(142)
These components do not commute, but obey the angular momentum algebra. This can be expressed
succinctly as:
8
bx ; L
by ] = ih b
Lz ;
< [L
b
b
b
b
b
b
L L = ih L
or, in fully explicit form,
(144)
[L ; L ] = ih Ly ;
: z x
by ; L
bz ] = ih b
[L
Lx :
Hence, by the generalised Heisenberg uncertainty relation, only one component, conventionally taken to
Lz , can be measured with perfect precision (Lz = 0). Once this measurement has been performed,
be b
we cannot determine the other two components precisely (Lx 6= 0 and Ly 6= 0) 19 because the operator
b
Lz commutes with neither b
Lx nor b
Ly . Formally we can say that an eigenstate of b
Lz cannot also be an
b
b
eigenstate of either Lx or Ly .
However, there is a further twist to the story: the length of the angular momentum vector can also be
measured with perfect precision along with its z-component because the square of the angular momentum, L2 ,
b2 = b
b2 + b
L
L2x + L
L2z :
(145)
y
Lz :
commutes with b
b
[L
b
L i ] = 0;
for all i = 1; 2; 3
(146)
Figure 1 illustrates succinctly the physical consequences of these results: the length and the z component of the angular momentum can be measured exactly, but the angular momentum vector can lie at any
Lz are
orientation on the cone indicated by the circle. In terms of eigenstates it means that eigenstates ofb
19 The
hbLx i = hLbyi = 0
and that
hbL2x i = hbL2y i = h2 f (
2
` `+
ph
1)
m2 g
`(` + 1)
m2
2
These are consistent with Heisenbergs generalised uncertainty relation. The proof uses the raising and lowering operators
bL = (bLx Lby )=2 and the orthornormality of the eigenstates.
22
In spherical polars we write the unit vectors in the orthogonal (r; ; ) directions as (er ; e ; e ); these
are the analogues of the cartesian unit vectors (ex ; ey ; ez ) = (i; j; k). See Figures 2, 3 & 4 for the notation.
Then since orbital angular momentum is always perpendicular to the r-vector, it has no component along
r,
b
L
(er pbr + e pb + e pb )
(er r)
r (e pb
e pb )
1
ih e
e
;
sin
(147)
(148)
(149)
b ]=0
[V (r); L
2
and
(150)
b2
p
2m
b2
pb2r
pbr
L
ih +
2m
mr 2mr2
2 2
b2
h
2
L
+
+
2m r2 r r
2m r2
=
=
=
h2 1
r2
2m r2 r
r
b2
L
2m r2
(152)
(153)
(154)
(155)
i = 1; 2; 3
(156)
(157)
p2
2m
( pr + p? )2
2m
L2cl
p2r
+
2m 2mr2
The only difference is the additional quantum correction due to non-commuting operators:
ih pbr =mr =
h2
(2=r )=r
2m
23
(151)
b
Lx
b
Ly
b
Lz
b2
L
ih cos
cot sin
ih
1
1 2
2
h
sin
+
:
sin
sin2 2
(158)
(159)
(160)
(161)
24
0; 1; 2; : : :
`;
`+
1; : : : ; 0; : : : ; `
1; `:
(162)
The fact that these quantum numbers are integers comes from the physical requirement that the wave
functions be finite and single-valued. The eigenfunctions Y m (; ), known as Spherical Harmonics, are
b 2 operator,
eigenfunctions of both the L
`
b2 Y
L
Lz operator,
and the b
m (; ) = h
`(` +
1) Y
m (; )
(163)
b
Lz Y` m (; ) = h mY` m (; )
(164)
(165)
jmj
(the N m are normalisation constants, see footnote11 ) where the associated Legendre Function, P (cos ),
are well-studied functions of cos . As expected for eigenfunctions of Hermitian operators the spherical
harmonics are orthonormal:
Z
Y m (; ) Y m (; ) d = m m
(166)
`
` `
where we use the solid angle notation as a convenient shorthand for the angular integration variable,
d sin d d, with the accompanying shorthand for the integration itself:
Z
d
Z
Z 2
d
0
sin d
jY m ( )j2 d = 1
`
(167)
0
;
(168)
Finally note that the expansion theorem allows us topwrite the most general wave function for a particle
whose magnitude of orbital angular momentum is h `(` + 1), ie. orbital angular momentum quantum
number `, as a linear combination of eigenstates with definite projections along the z-axis:
Y(
`)
(; ) =
+`
m=
cm Y m (; );
`
+`
where
jcm j2 = 1
m=
(169)
25
EE (r);
2
`(` +
(170)
1)E (r)
h mE (r) m =
`
`;
= 0; 1; 2; : : : ;
`+
(171)
1; : : : ; 0; : : : ; `
1; `:
(172)
b 2 and b
Lz are quantised as above is shown by writing down the TISE for a central
That the eigenvalues of L
potential,
b (r; ; ) E (r; ; ) = E E (r; ; );
H
(173)
h2
2m
2
2
+
r2 r r
b2
L
+ V (r)
+
2m r2
E (r; ; ); = E E (r; ; )
(174)
h2
2m
2
2
+
r2 r r
h2
1
sin
2m r2 sin
1 2
+
sin2 2
+ V (r)
= E E
(175)
This equation is solved by separation of variables, writing the wave function as a product:
E (r; ; ) = R(r) Y (; )
(176)
Thus we obtain the radial equation whose solution depends on the precise form of the potential,
h2
2m
d 2 R(r) 2 dR(r)
+
dr2
r dr
+
h2 `(` + 1)
V (r ) +
2m r2
R(r) = E R(r);
(177)
Y (; )
1
sin
sin
1 2Y (; )
+
Y (; ) = h2 `(` + 1) Y (; ):
sin2 2
(178)
b 2 operator,
which we recognise as the equation for the eigenfunctions and eigenvalues of theL
b 2 Y (; ) = h2 `(` + 1) Y (; )
L
(179)
At this stage the eigenvalue h2 `(` + 1) is an unknown separation constant: we only know that it has the
dimensions of h2 ; but its value comes from solving this equation. This is accomplished by another step
of variable separation,
(180)
Y (; ) = () ():
leading to the very simple equation,
d 2 ()
d2
26
m2 ():
(181)
(183)
and the physical assumption that the wave function be single-valued in , ie. that ( + 2n) = (),
yields the eigenvalues
m = 0; 1; 2; 3 : : :
(184)
The equation is altogether more complicated,
d()
1 d
sin
sin d
d
+
`(` +
m2
() = 0;
sin2
1)
(185)
This is the associated Legendre equation whose finite solutions, known as the associated Legendre
functions, are well-studied and are labelled by the values of the two constants ` and m:
() = P
jmj (cos ):
(186)
The requirement that the solutions be finite imposes a relationship between ` and m,
`
0; 1; 2; : : :
`;
`+
1; : : : ; 0; : : : ; `
1; `:
(187)
E (r) E
m (r) =
R E (r ) Y
`
m (; ):
(188)
(189)
(the N m are normalisation constants, see footnote11 ) are orthonormal, as expected for eigenfunctions of
Hermitian operators:
Z
`
Y m (; ) Y m (; ) d = m m
0
` `
(190)
where we use the solid angle notation as a convenient shorthand for the angular integration variable,
d sin d d, with the accompanying shorthand for the integration itself:
Z
d
Z
Z 2
d
0
sin d
jY m ( )j2 d = 1
`
(191)
0
;
27
(192)
m (r)
jRE (r)j2 r2 dr = 1
(193)
J: QUANTISATION
5.3.2 GENERAL ANGULAR MOMENTUM OPERATORS b
.
USING ONLY ABSTRACT COMMUTATION RELATIONS.
The foregoing discussion showed that the eigenfunctions of orbital angular momentum, the spherical
harmonics Y m (; ), do not depend on the detailed form of the potential: the eigenvalue equations and
the quantisation do not involve the potential. This suggests that all their properties are a consequence
only of the fact that they are eigenstates of the angular momentum operators, whose properties are in
turn entirely encapsulated in the commutation relations. This is the point of view we take here. The
consequence will be the discovery that there is more to angular momentum in quantum mechanics that
just orbital angular momentum: there is in addition spin angular momentum. Reserving the symbol
b for orbital angular momentum, we shall use the notation b
L
J for angular momentum in general. We
now postulate that in general these abstract angular momentum operators obey the same commutation
relations as those of orbital angular momentum, and are defined through their commutation relations:
`
b
J
8
< [Jbx ; Jby ] = ih Jbz ;
[Jb; Jb ] = ih Jby ;
: z x
[Jby ; Jbz ] = ih Jbx :
bJ = ih bJ
(194)
(195)
Jbi ] = 0;
for all i = 1; 2; 3
(196)
J2 and Jbz commute they have a common eigenfunction j m j , with as yet unknown quantum numSince b
bers j and m j respectively (see Figure 10):
;
b
J2 j;m j = h2 j( j + 1) j;m j
Jbz j m j
;
= hm
j j;m j
(197)
(198)
with
Jb+ = Jb :
(199)
which are shown to have properties analogous to the raising and lowering operators of the SHO, ab and
ab: they move us up and down the ladder of eigenstates,
Jb j m j
;
= c j;m j 1
(200)
the physical interpretation of plots of the radial wave function: one normally plots jr RE ` (r)j2 because
this represents the physical probability density at a distance r from the origin. Of course we must remember that this is
then modulated by the direction-dependent angular wave function jY`m (; )j2 factor to give the full spatial dependence of the
probability density. See Table 1 and Figures 8a,b.
21 This is important for
28
where c is a constant, c = hf j( j + 1) m(m 1)g1 2 . It is then possible to show that there is a lowest
eigenstate (as for the SHO) and also a highest eigenstate (unlike the SHO where the ladder of eigenstates
goes on forever); see Figure 11. Because the raising and lowering operators change the quantum number
m j by one unit this immediately leads to the identification of mmax
= j, mmin
j. Thus if we start from
j
j =
the lowest eigenstate, m j = j and repeatedly apply the raising operator we must eventually reach the
highest state, m j = j, and be able to go no further. If n is the the number of times the raising operator
must be applied to reach the highest state, then we have
=
j + n = mmax
= j
j
.
j=
or,
n
;
2
n = 0; 1; 2; : : :
(201)
2 j = INTEGER
ie.,
We therefore discover two possibilities: integral and half-integral angular momentum (in units of h),
whereas for orbital angular momentum only integral values were possible. We then have the quantisation
rules for general angular momentum:
Integral angular momentum: j = 0; 1; 2; : : : mj =
j;
j + 1:::0:::; j
1; j
(202)
1 3 5
; ;
;::: mj =
j ; j + 1 : : : ; j 1; j
(203)
2 2 2
Note the absence of the m j = 0 state for half-integral j. Thus, for an eigenstate of general angular
momentum we have,
Half-integral angular momentum: j =
b
J2 j;m
h2 j( j + 1) j m j
(204)
Jbz j m j
hm
j j mj
(205)
The angular momentum quantum numbers j and mj labelling an eigenstate are often referred to as the
angular momentum and its z component, or azimuthal projection, although strictly the angular momentum is h j and hm
j . Figure 12 illustrates the properties of the laddder of angular momentum eigenstates. A
special case of integral angular momentum is obviously the already encountered orbital angular momenb which cannot be pictured classically
b we shall see later that spin angular momentum, b
J = L;
J = S,
tum b
as microscopic spinning top, can have both integral and half-integral values.
29
The spherical polar coordinate system is depicted in Figures 2, 3 and the first diagram of Figure 4:
the unit vectors er ; e and e corresponding to the three chosen orthogonal directions are always mutually
perpendicular but vary in orientation with the vector r. This is why care must be taken when working
in non-cartesian coordinates. The most elementary way to see this is to just look at Figures 2, 3 (or the
first diagram of Figure 4), noting that as the vector r changes direction, so does the set of unit vectors
attached rigidly to it. More formally, we can express these unit vectors in terms of the fixed ones, i, j, k
of the cartesian system:
er
(206)
(207)
k sin :
(208)
To prove this we need to find the projections of the unit vectors er ; e and e onto the x ; y ; and z axes.
This is not as difficult as it may seem. First refer to Figure 2 or 3 (the first diagram of Figure 4 may also
help your 3-dimensional visualisation) and notice that the two unit vectors er and e lie in the plane of the
triangle OPQ; but e is perpendicular to it, so also perpendicular to the line OQ and parallel to the x y
plane, ie. e is perpendicular to the z axis. The following proofs are illustrated graphically in Figure 4.
(i) Take the unit vector er : it lies along the direction of the vector r, so that its projection in the z direction
is just 22 1 cos ; along the x and y axes we first project down onto the x y plane, giving a vector of
length 1 sin pointing along OQ at an angle with the x axis; then we see that its projections onto the
x and y axes are simply sin cos and sin sin respectively. This can be summarised in the vector
equation:
(209)
er = i sin cos + j sin sin + k cos :
Another way you may know this result is that the vector r has cartesian components,
r = (x; y; z) = (r sin cos ; r sin sin ; r cos );
(210)
and dividing by r gives the unit vector along the vector r, which is er .
er
r
r
(211)
(ii) e is rather special because it is perpendicular to the z axis and it also lies parallel to the x y plane;
therefore, when projected down onto that plane it still has unit length and lies perpendicular to the line
OQ which is at an angle with the x axis. Projection onto the x and y axes now gives sin and
cos respectively,
(212)
e = i sin + j cos :
(iii) The most difficult one is e . Since it is perpendicular to r its projection on the z axis is just sin
and its projection down onto the x y plane is cos . The latter projection points along OQ since the
original vector e is in the plane of the triangle OPQ. Hence the projection onto the x and y axes are
cos cos and cos sin respectively,
e = i cos cos + j cos sin
k sin :
(213)
22 The notation denotes simple multiplication in this section and in Figure 13. It has nothing to do with the vector cross
product; its only purpose is to emphasize that we are projecting unit vectors of length 1.
30
These equations allow us to discover how the unit vectors change with the angles by differentiating them,
remembering that the cartesian unit vectors i, j, k are constant:
er
er
i sin cos
j sin sin
er
k cos
(214)
e cos
(215)
(216)
i cos
e cos
j sin
er sin
(217)
=
=
k sin
(218)
(219)
Incidentally, since the unit vectors do not depend on the length r of the vector r, ei =r = 0: increasing
the length r does not change any unit vectors direction.
We can also get these geometrically by the sequences illustrated in Figure 13. The simplest is the
variation of er with . As shown in Figure 13a, we ask how er changes when is increased by d:
(i) increasing by d rotates er by d to the new unit vector e0r . The change, der = e0r er , as given
by the triangle of unit vectors illustrated, is in the direction of +e by an amount der =(the length of
er ) d = 1 d; remember that both er and e0r are unit vectors, its only the direction which has
changed. (iii) a similar argument holds for an increase d, but now the change requires projection from
the x y plane up to the er direction, giving an extra factor of sin , Figure 13b.
In the text we saw that the kinetic energy term in the classical Hamiltonian is:
(KE )cl =
p2
2m
(pr + p? )2
2m
L2cl
p2r
+
2m 2mr2
(220)
(221)
let us work out each term in the sum using the commutation relations:
[xi ; p
bj ] = ih
i j;
[xi ; x j ] = 0;
31
(222)
My plan is to get all the x; y; z factors to the left and all the momentum operators to the right:
b2
L
x
(y p
bz
(y p
bz )(y p
bz )
z pby )
(y p
bz )(z p
by )
(z p
by )(y p
bz ) + (z p
by )(z p
by )
In the first and last terms all factors commute and so can be placed in any order; in the second and third
terms there are two factors which dont commute, but which can be interchanged by using the known
for the last, pby y = y pby ih.
The terms
commutation relations. For the second term they are pbz z = z pbz ih;
can therefore be written with all momentum operators to the right and then rearranged into a suggestive
form:
b2
L
x
y2 pb2z
y2 pb2z + z2 pb2y
y(z pbz
ih) pby
z(y pby
(223)
Notice how I have rearranged the commuting factors at my convenience. Now the rest of the calculation
L2y and b
L2z have the same structure as b
L2x with an appropriate permutation of the labels
is simple because b
b2 we make the following replacements in our equation for b
L2x :
x; y; z. Thus to get L
y
y ! z; z ! x, giving
b
L2
y
(z p
bx
z2 pb2x + x2 pb2z
x pbz )
(224)
b2 :
L2z we make the following replacements in our equation for L
and to get b
x
y ! x; z ! y, giving
b
L2
z
=
=
(z p
by
(225)
(y + z ) p
bx + (x + z ) p
by + (x + y ) p
bz
(x + y + z )( p
bx + p
by + p
bz )
2 2
2 2
2 2
(x p
bx + y p
by + z p
bz )
r2 b
p2
2 2
2 2
2 2
(x p
bx + y p
by + z p
bz )
(226)
where we have added and subtracted terms in the first line to make an r2 = x2 + y2 + z2 factor. Notice
also that the cross terms xy, etc contain all possible pairs. The fact that the second term contains some of
p)2 suggests we look at this operator and see what it is; the calculation
the terms that would occur in (r:b
is rather similar to what we have already done:
(r:b
p)
(x p
bx + y p
by + z p
bz )(x p
bx + y p
by + z p
bz )
+2(xy p
bx p
by + xz p
bx p
bz + yz p
by p
bz )
=
x(x pbx
ih) pbz
+2(xy p
bx p
by + xz p
bx p
bz + yz p
by p
bz )
=
2 2
2 2
2 2
bx + y p
by + z p
bz )
(x p
+2(xy p
bx p
by + xz p
bx p
bz + yz p
by p
bz )
ih(r:b
p)
bx p
by + xz p
bx p
bz + yz p
by p
bz )
+2(xy p
ih(r:b
p)
(227)
In the first three terms we have reordered non-commuting factors using the commutation relations as
before, while the factors in the cross terms xy etc could be reordered because they all commute. This is
b 2 , except that the ih term lacks a
very nearly the negative of the terms we found in our expression for L
factor of 2; the result is therefore:
b 2 = r2 b
L
p2
b)
(r:b
p) + ih(r:p
2
(228)
c =
KE
(r:b
p)2
2m
ih
(r:b
p)
2mr2
2mr2
2
b2
pbr
pbr
L
ih +
2m
mr 2mr2
b2
L
2mr2
(229)
(230)
b = rp
br , but we also had
The last step above is trickier than might appear: we have used the fact that r:p
23
to be very careful of the non-commuting factors in the first term:
(r:b
p)
b)
(r:b
p)(r:p
(r p
br )(r p
br ) = r( p
br r) p
br
r(r pbr
r2 pb2r
ih(r:b
p)
ih[r;
] = ih
r
(231)
The additional term coming from the non-classical non-commuting operators is the one with an explicit
a clear indication of its quantum origins. You can now see why we didnt spot it at the
factor of ih,
classical level: the classical limit corresponds to replacing quantum operators by classical variables and
taking the limit h ! 0, which gives the correct expression for the classical kinetic energy; but going
23 In (r:p)2
b b 6 b
b2 occurring in eq. (228) where r2 is to the left of the
and pbr does not commute with r. This is quite different from the term r2 p
p2 and pb2r are very different as we see by comparing eqs. (229) and (230). If you are
momentum operators. Notice also that b
b
2
2
2 2
(r:p) = (r pr ) = r pr (r pr ) = r pr
b2 = (
p
ih)2
2
r2
+ e
e
2
r2
+ e
e
;
where the angular derivatives also act on the unit vectors to their right giving nontrivial additional terms. Here we have found
a way to avoid getting entangled in such calculations; but we cannot do so in Appendix C, where you can see some of the hard
calculations which would be required.
33
the other way, from the classical to the quantum expression we have to take great care: its not a simple
matter of merely replacing variables by operators, as we saw in the subtle manipulations in this appendix.
Finally, in spherical polar coordinates we can use the representation pbr = ih
=r to obtain
c =
KE
b2
p
2m
2
2
+
2
r
r r
h2
2m
h2 1
r2
2m r2 r
r
34
+
+
b2
L
2m r2
b2
L
2m r2
(232)
(233)
b 2 IN SPHERICAL POLARS.
APPENDIX C: AN EXPRESSION FOR L
We begin with the angular momentum operator in spherical polar coordinates,
b=
L
ih e
1
e
sin
(234)
In squaring this we must remember that the unit vectors e and e are not constants, and that the operator
should always be thought of as operating on some function (r; ). Thus,
(
1 b2
L
ih)2
=
=
1
e
e
:
e
e
sin
sin
1
e :
e
e :
e
sin
1
1
1
+ e :
e :
e
e
sin
sin
sin
2
e
1
e
1
(e :e )
+ e :
(e :e )
e :
2
sin
sin
2
2
e
1
1
1
e
1
(e :e )
e :
+ (e :e )
+ e :
2
2
sin
sin
sin2
sin
2 cos
1 2
+
+
(235)
2 sin sin2 2
where we have used the results obtained in Appendix A together with orthogonality of the unit vectors
to evaluate the dot products. We can combine the derivatives into one term to obtain the standard form:
b2
L
2
1
sin
sin
1 2
+
sin2 2
:
(236)
A final comment: by combining the results in Appendix B and C we see that we have computed the delp2 from Appendix B and the expression
squared operator in polar coordinates. Using our expression for b
2
b , we have:
above for L
(
ih)2 2
b
p2
pb2r
r2
=
=
b2
L
r2
2
b2
2
L
2
h
+
+
r2 r r
r2
2
2
1
1 2
2
h
+
+
sin
+
r2 r r sin
sin2 2
2ih
pbr
r
ie.
2 =
2
2
1
+
+
sin
r2 r r sin
+
1 2
sin2 2
(237)
which is the usual expression quoted in mathematics books. That the angle-dependent terms are simply
related to the square of the orbital angular momentum operator cannot of course be deduced from these
texts.
35
h i
Figure 8a. Polar plots of the the first few spherical harmonics.
Here we show projections of the probability distributions
onto the z y plane for ` = 0; 1; 2; 3. The distance from the origin of points on the curve represents this
probability as indicated in the ` = 0 plot. Since the probability density is independent of the azimuthal
angle , the actual distribution is obtained by rotating the plots about the z axis. Figure 8b shows the
resulting 3-dimensional shapes (actually, their square root, jY m j) for ` = 0; 1; 2 obtained by rotating the
plots of Figure 8a about the z-axis.
`
Figure 8b. The square root of the probability distributions, jY m (; )j for ` = 0; 1; 2. The distance
from the origin of a point on the 3-dimensional surface represents jY m (; )j, where (; ) represents the
direction of the vector drawn from the origin to that point.
`
Figure 8.
Figure 8.
Figure 8.
Figure 8.
(238)
(239)
(240)
36