Copyright c 2019 by Robert G.

Littlejohn

Physics 221B
Spring 2020
Notes 29
Helium and Helium-like Atoms†

1. Introduction

In these notes we treat helium and helium-like atoms, which are systems consisting of one
nucleus and two electrons. If the nuclear charge is Z = 2, then we are dealing with ordinary helium,
while Z = 3 is the Li+ ion, Z = 4 is the Be++ ion, etc. In recent years there has been some interest
in helium-like uranium, the ion U90+ , with Z = 92, for tests of quantum electrodynamics. A case
not to be overlooked is Z = 1, a system consisting of a hydrogen nucleus plus two electrons. This
system certainly has unbound states, for example, the states in electron-hydrogen scattering. It is
not obvious that it possesses bound states, however, since it is not clear that one proton is capable
of binding two electrons. In fact, a bound state of this system does exist, the H− ion, which can
be thought of as an electron bound to an otherwise neutral hydrogen atom. The second electron
is rather weakly bound, as we shall see. The ion H− plays an important role in applications, for
example, in energy transport in stellar atmospheres. It is also used in proton accelerators to defeat
Liouville’s theorem in the process of filling up accelerating rings. In the following discussion of
helium-like atoms we focus mostly on the bound states, but some attention is also given to the
unbound states.

2. The Basic Hamiltonian

In these notes we work in atomic units, e = m = h̄ = 1. We place the nucleus of charge Z at

the origin of our system of coordinates and we let x1 and x2 be the positions of the two electrons.
Then the basic Hamiltonian describing the helium-like system is

p21 p2 Z Z 1
H= + 2− − + , (1)
2 2 r1 r2 r12

where r1 = |x1 |, r2 = |x2 |, and r12 = |x2 − x1 |.

This Hamiltonian incorporates all the nonrelativistic, electrostatic effects in helium, but it omits
a number of small effects and makes some approximations. The approximation in the use of the
Hamiltonian (1) is numerically quite good, but there is some interesting physics in the neglected

† Links to the other sets of notes can be found at:

http://bohr.physics.berkeley.edu/classes/221/1920/221.html.
2 Notes 29: Helium

terms and consideration of them leads to an improved understanding of some of the steps we take
in solving the approximate system.
First, the frame attached to the nucleus is not really an inertial frame, since the nucleus is pulled
by the electric forces from the electrons and accelerates somewhat in response. The accelerated
nucleus then has an effect back on the electrons, so that the overall effect is to couple the dynamics
of the electrons. However, since the mass of the nucleus is large in comparison to the mass of the
electron, the corrections, which are known as “mass polarization” terms, are small. See Prob. 16.4,
where it is shown that the coupling is proportional to p1 · p2 . Similar corrections result from
ignoring the small difference between the true electron mass and the reduced mass of the electron
and the nucleus. Both these corrections are of order of the electron to nucleus mass ratio, which is
approximately 10−4 in helium.
Second, there are a variety of fine structure terms which, as in hydrogen, can be thought of as
relativistic corrections. These are all of order α2 or (Zα)2 , and are roughly of order 10−3 to 10−4 in
helium. (But as in hydrogen, fine structure effects grow with Z, and are more important in heavier
atoms.) In addition to the relativistic kinetic energy correction, the Darwin term and the spin-orbit
term, all of which are present in hydrogen, there are also terms representing the interaction of one
electron with the magnetic field produced by the orbital motion of the other electron (spin-other-
orbit terms), or produced by the magnetic moment of the other electron (spin-spin terms). There
are also terms taking into account retardation in the communication between the two electrons.
Third, there are interactions with the quantized electromagnetic field, generalizations of the
Lamb shift in hydrogen.
Finally, there may be hyperfine interactions between the electron and the multipole moments
of the nucleus (but not in ordinary helium, since the α-particle has spin 0).
With the neglect of all these small terms, the basic helium-like Hamiltonian (1) is a purely
spatial operator, and does not involve the spin. Thus, it is similar in its level of approximation
to the electrostatic model of hydrogen. Due to the requirements of the symmetrization postulate,
however, spin plays a significant role in the structure of helium, much more so than in hydrogen at
a similar level of approximation.

3. Wave Functions and Hilbert Spaces

As explained in Sec. 18.3, the Hilbert space of a single electron can be written E = Eorb ⊗ Espin ,
where a basis in Eorb is the set of position eigenkets {|xi} and a basis is Espin is the set of eigenkets
{|smi} of the operators S 2 and Sz , where s = 12 and m = ± 21 . The quantum number s is constant
and can be suppressed if it is understood, so sometimes we write simply |mi for the spin eigenkets.
A basis in the total Hilbert space is the product basis {|xmi} where |xmi = |xi|mi. A wave
function in the orbital Hilbert space Eorb is ψ(x) = hx|ψi, while one in the total Hilbert space E is
ψm (x) = hxm|ψi.
 Notes 29: Helium 3

In a system with two electrons the total Hilbert space is the product of two copies of the orbital
and spin Hilbert spaces for the two electrons,
(1) (1) (2) (2)
Etot = Eorb ⊗ Espin ⊗ Eorb ⊗ Espin , (2)

where the numbers in parentheses are the labels of the two electrons. This is the “nominal” Hilbert
space discussed in Notes 28, that is, it consists of the wave functions that we would have if there
were no symmetrization postulate. A basis in Etot is the product of the bases in the constituent
spaces,
|x1 m1 i(1) |x2 m2 i(2) = |x1 x2 m1 m2 i. (3)

If we rearrange the factors we can write the total two-electron Hilbert space as

Etot = Eorb ⊗ Espin , (4)

where
(1) (2)
Eorb = Eorb ⊗ Eorb , (5a)
(1) (2)
Espin = Espin ⊗ Espin (5b)

The factorization (4) is useful for the study of helium and many other systems with two identical
particles. A basis in Eorb is the set {|x1 x2 i}, while there are two obvious bases in Espin , the uncoupled
and coupled,
{|m1 m2 i} or {|SMS i}, (6)
as explained in Sec. 28.8. Here S = 0 is the singlet state and S = 1 indicates the set of triplet states.
Equation (4) means that an arbitrary wave function in Etot is a linear combination of products
of wave functions in Eorb and those in Espin . In the special case that a wave function in Etot consists
of a single product of this type, we can speak of the “spatial part” and “spin part” of the wave
function. Such language is very common in discussions of helium and other two- or multi-particle
systems, but it is important to realize that it only applies to wave functions of a special type. The
reason such wave functions are relevant in the case of helium has to do with the exchange operators
that commute with the Hamiltonian (1).

4. Exchange Operators

The Hamiltonian (1) commutes with three exchange operators that we now define, in terms of
their action on the basis kets |x1 x2 m1 m2 i in Etot . These are
orb
E12 |x1 x2 m1 m2 i = |x2 x1 m1 m2 i, (7)
spin
E12 |x1 x2 m1 m2 i = |x1 x2 m2 m1 i, (8)
E12 |x1 x2 m1 m2 i = |x2 x1 m2 m1 i. (9)
spin
We will denote the eigenvalues of these three operators by eorb
12 , e12 , and e12 , respectively.
4 Notes 29: Helium

orb
The Hamiltonian (1) commutes with E12 because it is a purely orbital operator that is sym-
spin
metrical in the exchange of the two electrons. It commutes with E12 because it is a purely orbital
operator that commutes with any spin operator. And in view of Eq. (10), the Hamiltonian (1) also
commutes with E12 .
But the Hamiltonian (1) is only an approximation to the true Hamiltonian for helium. If we
include the fine structure terms, which as in hydrogen couple the spin and orbital degrees of freedom,
orb spin
then the resulting Hamiltonian no longer commutes with E12 or E12 separately.
It still commutes with E12 , however, regardless of how many physical effects we include or how
many correction terms we add, because the two electrons are identical. All physical Hamiltoni-
ans that describe two electrons commute with E12 ; if this were not true, it would be possible to
distinguish the electrons.
The properties of these exchange operators follow immediately from their definitions. The
operators are all Hermitian and unitary, their square is 1 (the identity operator), and their eigenvalues
are ±1, as in Eqs. (28.3) and (28.4). They also satisfy

orb spin
E12 = E12 E12 , (10)

and
orb spin
[E12 , E12 ] = 0. (11)
orb spin orb spin
Thus E12 and E12 possess a simultaneous eigenbasis. Since E12 is a function of E12 and E12 , any
orb spin
simultaneous eigenspaces of E12 and E12 is automatically an eigenspace of E12 , with eigenvalue
spin
e12 = eorb
12 e12 . (12)

orb spin
Since the set of operators (H, E12 , E12 ) are mutually commuting, they possess a simultaneous
eigenbasis. Here H is the Hamiltonian (1). Equivalently, we can organize the eigenstates of H so
orb spin
that they are also eigenstates of E12 and E12 .
spin
To construct such eigenstates we start with E12 . Let us use a capital Ψ to refer to states in
spin
Etot and a lower case ψ for states in Eorb . Then an eigenstate of E12 is the product

|Ψi = |ψi|SMS i, (13)

for any |ψi ∈ Eorb , since

spin
E12 |Ψi = σS |Ψi, (14)

where the eigenvalue espin spin

12 = σS is defined by Eq. (28.45). The most general eigenstate of E12 is a
linear combination of such states (13) with the same value of S.
Next, we make |Ψi an eigenstate of H by requiring that the spatial part |ψi of |Ψi be an
eigenfunction of H, regarded as a purely orbital operator. That is, we must solve

H|ψi = E|ψi, (15)

 Notes 29: Helium 5

or, in wave function language

Hψ(x1 , x2 ) = Eψ(x1 , x2 ), (16)

a purely orbital eigenvalue-eigenfunction problem. This is the hard part, something we shall devote
some attention to in these notes.
orb
Then, to obtain an eigenfuction also of E12 , we note that if ψ(x1 , x2 ) is a solution of Eq. (16),
then so is ψ(x2 , x1 ), with the same energy. Thus we can form the wave functions

1  
√ ψ(x1 , x2 ) ± ψ(x2 , x1 ) , (17)
2
orb
which, if they do not vanish, are eigenstates of E12 with eigenvalue eorb
12 = ±1 (the same ± that
appears in Eq. (17)).
Finally, in view of Eq. (12), only those states that are symmetric under spatial exchange and
antisymmetric under spin exchange, or vice versa, are antisymmetric under total exchange E12 , as
required by the symmetrization postulate. The physical subspace of the nominal Hilbert space is
spin
the direct sum of the subspace with eorb
12 = +1 and e12 = −1, and the one with eorb
12 = −1 and
espin
12 = +1.
The situation regarding spatial and spin exchange symmetries is summarized in Table 1, which
also indicates some standard nomenclature. As in Table 28.1, “para” refers to a wave function that
is spatially symmetric, and “ortho” to one that is spatially antisymmetric under exchange. At one
time helium was thought to consist of two different species, parahelium and orthohelium, since the
spectrum of helium indicates two classes of energy levels with no (or only very weak) transitions
between them. Table 1 is almost exactly the same as Table 28.1, the main difference being that
in the case of the hydrogen molecule, it was possible to write down an explicit form for the spatial
part of the energy eigenfunctions, Eq. (28.33), since we had a central force Hamiltonian, whereas
in helium it is much more difficult to solve the spatial eigenfunction equation (15). (And in the
hydrogen molecule we had two identical protons, whereas in helium we have two identical electrons.)

eorb
12 name espin
12 e12 S degen name
+1 para −1 −1 0 1 singlet
−1 ortho +1 −1 1 3 triplet

Table 1. Linkage of spatial and spin exchange symmetry in helium and terminology.

In summary, we have derived the rule about helium eigenfunctions that most students learn in
first courses on quantum mechanics, namely, that these eigenfunctions are products of spatial wave
functions times spin wave functions with opposite symmetry under exchange. We saw the same rule
previously in connection with the hydrogen molecule.
6 Notes 29: Helium

orb
However, the true nature of this rule should be understood. First, the operators E12 and
spin
E12 do not have the same fundamental significance as the overall exchange operator E12 . In
orb spin
particular, there is no symmetrization postulate for E12 or E12 . Next, the factorization of the
energy eigenfunctions into symmetric spatial parts and antisymmetric spin parts, or vice versa, arises
orb spin
because H commutes with E12 and E12 and because commuting operators possess simultaneous
eigenstates. If we had included the small fine structure corrections in the Hamiltonian (1), however,
orb spin
then we would find that H no longer commutes with E12 or E12 separately, although of course it
would still commute with E12 . Thus, the usual rule about the factorization of the wave functions only
applies to the nonrelativistic, electrostatic approximation to the helium atom. This is in contrast to
the symmetrization postulate itself, which is rigorously correct.

5. Good Quantum Numbers

Sometimes an operator that commutes with the Hamiltonian is called a “good quantum num-
ber.” The terminology is old and it confuses an operator with its quantum number, but it is in
common use.
The helium Hamiltonian (1) is not exactly solvable nor even close to any exactly solvable system
(at least for small Z), so it is rather difficult to obtain good approximations to its eigenstates and
eigenvalues. In cases like this it is especially important to pay attention to the exact symmetries
of the Hamiltonian, that is, the operators that commute with it (the “good quantum numbers”).
Commuting operators possess simultaneous eigenbases, and, since symmetry operators are usually
easier to diagonalize than Hamiltonians, it is usually a good idea to diagonalize them first. This
means that we need only search in the eigenspaces of the symmetry operators for the eigenstates of
the Hamiltonian, a significantly easier task than searching in the whole Hilbert space. For example,
if we must diagonalize a matrix to find the energy eigenstates and eigenvalues, it is best to use a
symmetry-adapted basis (one that is already an eigenbasis of the symmetry operators). Even if
we do not attempt to find the eigenstates of the Hamiltonian explicitly, we can at least say that
those eigenstates are labelled by symmetry labels, namely, the quantum numbers of the symmetry
operators. This is a common situation for example in particle physics or nuclear physics, where
no one knows how to calculate the masses of the baryons or the energies of nuclear states very
accurately, but these states are rigorously characterized by their quantum numbers (spin, parity,
etc.)
For the helium Hamiltonian (1) we have already begun this process, by noting that H commutes
orb spin
with E12 and E12 , and by finding the simultaneous eigenbases of these two operators. This served
the dual purpose of narrowing the spaces in which we need to search for energy eigenstates, and of
satisfying the symmetrization postulate (by requiring e12 = −1).
But there are other symmetries of the Hamiltonian (1). Most notably, since H is a scalar
operator, it commutes with rotations. In fact, since it is a purely spatial operator, it commutes with
both spatial and spin rotations independently, and thus with L = L1 + L2 and S = S1 + S2 . It
 Notes 29: Helium 7

does not, however, commute with spatial rotations of particle 1 or of particle 2 separately, because
the term 1/r12 in the Hamiltonian is only invariant when we rotate the positions of both particles
simultaneously. That is, H does not commute with L1 or L2 separately. The Hamiltonian does
commute with S1 and S2 separately, however, in fact, it commutes with any function of spin. But
as we have seen, the total spin S = S1 + S2 is more convenient to work with than the individual
spin
spins because it is physical and commutes with E12 , and because the coupled basis is an eigenbasis
spin
of E12 .
The Hamiltonian (1) also commutes with parity π, a purely spatial operator that acts on the
basis kets of the entire Hilbert space E by

π|x1 x2 m1 m2 i = | − x1 , −x2 , m1 , m2 i. (18)

Parity flips the spatial coordinates of the particles but has no effect on spin (at least in nonrelativistic
quantum mechanics). We will not make much use of parity in the case of helium, but it is more
important in multielectron atoms.

orb spin
Operator E12 E12 E12 L2 Lz S2 Sz
Qu. Num. eorb
12 espin
12 e12 = −1 L ML S MS

Table 2. Operators that commute with the approximate helium Hamiltonian (1) and their corresponding quantum
numbers. The symmetrization postulate forces e12 = −1.

A list of operators that H commutes with along with their quantum numbers is shown in Table 2.
Moreover these operators commute with each other. Thus, without knowing anything else about the
eigenstates of H, we can say that they are characterized by the quantum numbers of the operators in
the table. However, several of the quantum numbers are superfluous. For example, e12 is forced to be
spin
−1 by the symmetrization postulate, so eorb
12 and e12 are opposite one another and not independent.
In fact, by Eq. (28.44) knowledge of S implies knowledge of espin orb
12 and therefore knowledge of e12 .
Thus an independent set of quantum numbers is (LML SMS ). Within a simultaneous eigenspace
of the symmetry operators with the given quantum numbers, we can in principle diagonalize the
Hamiltonian, and label the energy eigenstates within that subspace by a sequencing number N .
Thus, the energy eigenstates can be denoted |N LML SMS i. As for the energy eigenvalues, they are
independent of ML and MS because of the rotational invariance of the Hamiltonian, so we can write

H|N LML SMS i = EN LS |N LML SMS i. (19)

The energies EN LS are (2L + 1)(2S + 1)-fold degenerate, because of the freedom in the magnetic
quantum numbers.
You might suspect that the energies EN LS should also be independent of the total spin S, since
the Hamiltonian (1) is a purely spatial operator. That would be true if there were no symmetrization
8 Notes 29: Helium

postulate, but in that case the quantum number eorb orb

12 would not be superfluous, and, since e12 is a
spatial operator itself, the energies would depend on its eigenvalue. In fact, this is precisely what
happens in real helium where the symmetrization postulate is operative, it is just that the quantum
number eorb
12 is not explicitly stated since it is determined by the value of S. In other words, in real
helium, the spin state specified by S determines the exchange symmetry of the spin part of the wave
function which determines the exchange symmetry of the spatial part of the wave function which
does affect the energy. Thus, the energy does depend on S, in spite of the fact that H is a purely
spatial operator.
In fact, the dependence of the energy on S is typically large, that is, of the order of a Coulomb
(electrostatic) energy, much larger than the (magnetic) energies associated with the explicitly spin-
dependent terms that we have neglected in Hamiltonian (1). This is because wave functions that are
symmetric or antisymmetric under spatial exchange have very different distributions of probability
in three-dimensional space, and therefore correspond to quite different Coulomb energies. We will
see this explicitly below when we analyze helium by perturbation theory. A similar phenomenon
occurs in ferromagnetism, where the energies involved in flipping spins (the magnetization of the
material) are much larger than can be accounted for by the magnetic interaction of arrays of magnetic
dipoles. This discrepancy in energy scales was known before quantum mechanics had matured, and
was regarded as a mystery. The true explanation was given by Heisenberg: when you flip spins,
you change the symmetry of the spin part of the wave function, which by the symmetrization
postulate forces changes in the spatial part of the wave function, which changes the Coulomb energy
of the charge configuration. Ferromagnets are more complicated because there is a large number of
particles, but the basic idea is the same as in helium.
In Table 2 we have listed all the obvious symmetries (except parity) of the basic helium Hamilto-
nian (1). Are there any more (not so obvious) symmetries? Sometimes a system does have additional
symmetries in addition to the obvious ones. Notably this happens in the case of hydrogen, which
has an SO(4) symmetry (only the SO(3) symmetry under spatial rotations is obvious). This extra
symmetry is responsible for the extra degeneracy of hydrogen, in comparison to other central force
problems. As we say, hydrogen has “hidden” symmetry. Another example with hidden symmetry is
the multidimensional, isotropic harmonic oscillator. Apart from these two examples, however, there
are very few systems in the real world with extra or hidden symmetry (it depends partly on your
definition of “hidden,” but you should not expect such symmetry in most problems). In particular,
helium has no extra symmetry beyond what we have listed.
The standard spectroscopic notation for the level EN LS is N 2S+1 L, where L is replaced by
one of the code letters, S, P , D, F , etc., for L = 0, 1, 2, 3, etc. As we did in our discussion of the
hydrogen molecule in Sec. 28.8, we use capital letters are used to represent the quantum numbers of
the collective system (in this case, the whole atom), and reserve lower case letters for the quantum
numbers of a single particle (in this case, an electron). For example, the state 23 P is the state with
N = 2, S = 1 and L = 1. This is really a manifold of (2L + 1)(2S + 1) = 3 × 3 = 9 degenerate states.
 Notes 29: Helium 9

6. Ionization Potentials

So far we have succeeding in labeling the energy levels of helium-like atoms by their good quan-
tum numbers and determining their degeneracies, in the approximation given by Eq. (1), without
knowing much about the system apart from its symmetries. We now look at some experimental and
other facts about such atoms.
First we present some ionization potentials. The ionization potential of an atom is the energy
needed to remove one electron from the atom, assumed to be in its ground state, to infinity. If an
atom has several electrons, there are successive ionization potentials, as more and more electrons
are removed. The final state consists of a nucleus plus the electrons, all of which are at rest and at
an infinite distance from one other.
It is a matter of convention in most areas of physics to decide where on the energy axis E = 0
lies, since only energy differences are physically meaningful. (This is true as long as gravitational
effects are neglected. But in general relativity, the mass corresponding to energy by m = E/c2 has
a gravitational effect, so the absolute measure of energy is physically significant.) But it is common
in electrostatics to define the state of zero potential energy to be the one in which all charges are at
an infinite separation from one another. This is what we have done with the potential energy in the
Hamiltonian (1). If the kinetic energies also vanish, then we have a state of zero total energy. With
this convention for the state E = 0, the sum of all the ionization potentials of the atom (with a
minus sign) is the ground state energy of the atom. (Notice that we have not separated the protons
in the nucleus or transported them to infinity. We are using a convention that is convenient in
atomic physics, but perhaps not nuclear physics.)

0 He++ + 2e− 0 H+ + 2e−

-1 continuum -0.25 continuum

-2 He+ + e− -0.5 H + e−
E0 = −0.528

E0 = −2.904
-3 -0.75
Fig. 1. Important energies of helium obtained from ion- Fig. 2. Same as Fig. 1, but for H− . The energy scale is
ization potentials. Energies are measured in atomic units. down by a factor of 1/4 compared to helium.
10 Notes 29: Helium

In the case of helium, the ionization potentials are

He → He+ + e− − 0.904 a.u.,

(20)
He+ → He++ + e− − 2.0 a.u.

The first ionization potential may be compared to that of hydrogen, which is 0.5 a.u. In fact, helium
has the highest ionization potential of any neutral atom. See Fig. 31.1. The second ionization
potential is easy to understand, since the He+ ion is a hydrogen-like atom, whose ground state
energy is −Z 2 /2 = −2 a.u. To calculate the first ionization potential, however, is a nontrivial
problem. In the case of H− , the ionization potentials are

H− → H + e− − 0.028 a.u.,
(21)
H → H+ + e− − 0.5 a.u.

In H− the extra electron is only weakly bound to the neutral H atom.

Figures 1 and 2 illustrate the information obtained from the ionization potentials. The energy
E = 0 is the state of complete dissociation. The ground state energy of helium is E = −2.904 a.u.,
while the state in which one electron has been removed to infinity has energy E = −2 a.u. Once
this electron has been removed, it may be given any positive kinetic energy, so there is a continuous
spectrum above E = −2 a.u. In H− , the ground state energy is E = −0.528 a.u. and the continuum
begins at E = −0.5 a.u., the ground state energy of a neutral hydrogen atom.

Parahelium Orthohelium
(S = 0) (S = 1)
-2.0

41 S 41 P 41 D 43 S 43 P 43 D
31 S 31 P 31 D 33 P 33 D
33 S
-2.1
21 P
23 P
21 S
23 S
-2.2

-2.9 11 S

-3.0

Fig. 3. Bound states of helium with quantum numbers. Energies are measured in atomic units. Notice the gap in the
energy scale between the ground state and the excited states.
 Notes 29: Helium 11

7. The Bound States

From Fig. 1 it is clear that the bound states of helium must lie between E0 = −2.904 a.u.
and E = −2 a.u. These bound states and their quantum numbers are displayed in Fig. 3 with an
expanded energy scale. The energy levels in the figure are the eigenvalues of the Hamiltonian (1),
which are not exactly the same as the experimental values due to the neglect of small terms in that
Hamiltonian. The difference, however, is too small to see on the scale of the diagram (in fact, at
higher resolution, many of these levels will be seen to have a fine structure splitting).
The levels in Fig. 3 cannot be regarded as purely experimental data, even at a scale where fine
structure is ignored and the Hamiltonian (1) is a good approximation, because spectroscopic data
only gives differences between energy levels, not their absolute values. It takes some disentangling of
the experimental data to produce energy levels. In addition, the levels in Fig. 3 have been assigned
quantum numbers, which requires some theoretical input. In effect, we have skipped ahead to the
answer (the spectrum of helium) in presenting Fig. 3. This may be the best way to understand
helium on a first pass.
In Fig. 3, the para states (spatially symmetric, spin singlet) are on the left, and the ortho
states (spatially antisymmetric, spin triplet) are on the right. Each vertical column contains the
energy levels for a given value of L and S, the good quantum numbers of the Hamiltonian (1). Only
columns out to L = 2 (D-states) are shown in the figure, but in actual helium L ranges from 0 to
∞. The levels in a given column are sequenced in ascending order by N , with certain conventions
for the starting value of N . In each column there is an infinite number of levels that accumulate
just below the continuum limit, much as in hydrogen. The energy levels in a given column are the
eigenvalues of the Hamiltonian (1), restricted to a simultaneous eigenspace of the operators L2 , S 2 ,
Lz and Sz (it being understood that we are within the antisymmetric eigenspace of the exchange
operator E12 ).
Notice the following qualitative features of the energy level diagram for helium. First, the
ground state is a singlet state (11 S), which is considerably below the excited states (notice the gap
in the energy scale). The excited states look roughly the same on the para and ortho sides, but there
is no 13 S state (no state on the ortho side that seems to correspond to the 11 S state). Second, for a
fixed value of N and S, the energies are an increasing function of L (the “staircase effect”) (at least
this is true for N = 2 and most other values of N ). This is the same effect we saw in alkali atoms.
Third, for fixed N and L, the ortho states are lower in energy than the para states (thus, if a 13 S
state existed, we would expect it to be below the actual ground state 11 S). Later we will provide a
theoretical explanation for some of these features.
In the case of H− , the ground state is the only bound state, and it has the quantum numbers
11 S (same as the ground state of helium).
12 Notes 29: Helium

8. Perturbation Analysis. The Unperturbed System

We now turn to the problem of finding the spatial eigenfunctions of the Hamiltonian (1), which
is a purely orbital operator. Denoting such an eigenfunction by ψ(x1 , x2 ), we wish to solve

Hψ(x1 , x2 ) = Eψ(x1 , x2 ). (22)

In addition, we wish these eigenfunctions to be even or odd under orbital exchange,

orb
E12 ψ(x1 , x2 ) = ±ψ(x1 , x2 ), (23)

where the ±1 is eorb orb orb

12 , the eigenvalue of E12 . As explained above, e12 = +1 are called para states, and
eorb
12 = −1 are called ortho states. We know it is possible to construct simultaneous eigenfunctions
orb
of H0 and E12 because they commute. Once we have found these eigenfunctions, it will be easy
to multiply them by spin states (singlet or triplet) of the opposite exchange symmetry to obtain an
overall eigenfunction of the Hamiltonian, including spin, that satisfies the symmetrization postulate.
This part is relatively easy; the hard part will be solving Eq. (22) as a purely orbital problem.
An obvious strategy for finding the eigenfunctions of the Hamiltonian (1) is to use perturbation
theory, in which the inter-electron Coulomb potential 1/r12 is taken as a perturbation. That is, we
write H = H0 + H1 , where
 p2 Z   p22 Z
1
H0 = − + − , (24a)
2 r1 2 r2
1
H1 = . (24b)
r12

We have grouped the terms in H0 to emphasize that H0 is the sum of two hydrogen-like Hamiltonians,
one for each electron, moving in the field of a nuclear charge Z. This decomposition of H into H0
and H1 is not very favorable for perturbation theory in the case of small Z, since the inter-electron
potential is comparable to the electron-nucleus potentials. The comparison is especially unfavorable
for H− (the case Z = 1). It gets better for larger Z. We do the perturbation analysis anyway, since
it is relatively straightforward and there are not many choices.

e− r12 e−

z
r2
r1
γ

nucleus y

Fig. 4. Coordinates for the nucleus and two electrons in space.

 Notes 29: Helium 13

It is obvious from Fig. 4 what the effect of the perturbation 1/r12 is. When it is switched off,
the two electrons no longer repel one another and settle in somewhat closer to the nucleus than
they would otherwise. The wave functions for the two electrons can be thought of as two clouds
that overlap in the neighborhood of the nucleus. Conversely, when we turn the perturbation on, the
electron clouds expand somewhat so the electrons can get away from one another. They also tend
to stay on opposite sides of the nucleus, for the same reason.
As usual in perturbation theory, we must understand the unperturbed system first, its eigen-
functions, degeneracies, and symmetries. Since H0 is a sum of two hydrogen-like Hamiltonians, the
(two-particle) eigenfunctions of H0 are products of single particle eigenfunctions of the hydrogen-like
Hamiltonians in the field of nuclear charge Z. We call these single-particle eigenfunctions orbitals.
They are central force eigenfunctions, and since we are only concerned with the spatial part here,
they are characterized by the usual central force quantum numbers (nℓm). Notice that we are using
lower case letters for the quantum numbers of a single electron. The single particle hydrogen-like
energy is −Z 2 /2n2 . We denote these orbitals in ket language by |nℓmi. Then the two-particle
eigenfunctions of H0 are products of these, which we denote by

|n1 ℓ1 m1 i(1) |n2 ℓ2 m2 i(2) = |n1 ℓ1 m1 n2 ℓ2 m2 i, (25)

where the numbers in parentheses indicate which of the two electrons (1 or 2) is referred to, and
where the right hand side is a shorthand notation for these states. These states depend on six
quantum numbers. They are eigenfunctions of H0 ,

H0 |n1 ℓ1 m1 n2 ℓ2 m2 i = En(0)
1 n2
|n1 ℓ1 m1 n2 ℓ2 m2 i, (26)

where the (0) on the energy indicates that it is the energy in zeroth order of perturbation theory.
This zeroth order energy is a sum of the two hydrogen-like energies,
Z2  1 1
En(0)
1 n2
= − + , (27)
2 n21 n22
and it depends only on the two principal quantum numbers n1 and n2 .
orb
The states (25) are eigenstates of H0 , but not in general of E12 . However it is easy to construct
orb
eigenfunctions of E12 simply by symmetrizing or antisymmetrizing under exchange. There are
two cases. In Case I, the two sets of single particle quantum numbers are equal (all of them),
(n1 ℓ1 m1 ) = (n2 ℓ2 m2 ) ≡ (nℓm), and we write simply

|nℓm nℓmi (28)

for these states. In this case the states are already even under orbital exchange, so they are para
states (eorb
12 = +1). There are no ortho states in Case I.
In Case II, one or more of the two sets of quantum numbers are different, (n1 ℓ1 m1 ) 6= (n2 ℓ2 m2 ).
orb
In this case we can easily write down normalized eigenstates of E12 ,
1

√ |n1 ℓ1 m1 n2 ℓ2 m2 i ± |n2 ℓ2 m2 n1 ℓ1 m1 i , (29)
2
14 Notes 29: Helium

where the eigenvalue is eorb

12 = ±1. In this case we have both para and ortho states. The energy
does not change when we carry out this symmetrization or antisymmetrization because both terms
have the same energy. It is still given by Eq. (27), and it does not depend on the choice of the ±
orb
sign. We now have a set of unperturbed energy eigenstates that are also eigenstates of E12 .
0 He++ + 2e−
para ortho

(2, ∞) (2, ∞)
(2, 3) (2, 3)

−1 (2, 2) (2, 2)

−2 (1, ∞) (1, ∞) He+ + e−

(1, 3) (1, 3)

(1, 2) (1, 2)

−3

−4 (1, 1)

Fig. 5. Unperturbed energy levels of helium, that is, levels of the Hamiltonian (24a). The energy axis is vertical with
energies measured in atomic units, and levels are labeled by principal quantum numbers (n1 , n2 ). The shaded area is
the continuum.

Figure 5 shows the energy levels of H0 , with the para states on the left and ortho states on the
right. The levels are labeled by their principal quantum numbers (n1 , n2 ).
Let us focus first on the levels below the shaded area in Fig. 5, that is, the levels lying below
the continuum threshold at E = −2 (all energies are in atomic units). These levels are called singly
excited states because one of the hydrogen-like orbitals is in the ground state (n1 = 1), while the
other (n2 ) takes on any value. The absolute ground state of H0 is the (1, 1) state with n1 = n2 = 1,
at an energy of −4. The ground state is on the para side; there is no analogous state on the ortho
side because this is Case I, described in the paragraph surrounding Eq. (28).
It is instructive to compare these levels with the exact bound states of helium, shown in Fig. 3.
One difference is that the levels in Fig. 3 are separated by the L quantum number, while those in
Fig. 5 are not. There is no point in separating the levels of Fig. 5 by the orbital angular momentum
quantum numbers because the hydrogen-like energies do not depend on them. Apart from that,
however, the two energy level diagrams are qualitatively similar. In particular, in both cases we
 Notes 29: Helium 15

have an infinite number of levels that accumulate at the continuum threshold E = −2.
Our analysis of the unperturbed system shows already why there is no 13 S state in real helium: it
is because it would have to be a perturbed version of an antisymmetrized ground state wave function,
and the ground state wave function is spatially symmetric. On the other hand, the unperturbed
spectrum shows no evidence of the staircase effect (hydrogen energies do not depend on ℓ), and
there is exact degeneracy between the ortho and para sides (when a level exists on both sides). That
is, the unperturbed energies do not depend on the ± sign in Eq. (29), which represents the spatial
symmetry of the wave function. In addition, the unperturbed ground state is rather badly off in
energy, −4 instead of the exact value of −2.904. In fact, all the unperturbed energies are too low;
this is because the perturbation H1 in Eq. (24b) is a positive operator, and can only raise energies.
These differences between the unperturbed and exact spectra are clarified by perturbation theory.

9. The Doubly Excited States

Figure 5 shows something else, the so-called doubly excited states in which both orbitals have
principal quantum numbers > 1. The doubly excited series with n1 = 2 and n2 ≥ 2 is shown in the
figure (twice, once on the para side and once on the ortho side). There are other series (not shown),
with n1 = 3, 4, . . ..
Insofar as the unperturbed Hamiltonian H0 is concerned, these are discrete states imbedded in
the continuum. That is, a state such as the (2, 2) state is a bound state of H0 with a normalizable
wave function, but it exists at an energy that lies in the continuous spectrum of another set of
states, those with n1 = 1 and the second electron with a positive energy. The wave functions of the
continuum are products of the ground state wave function for the first electron times an unbound
hydrogen-like solution for the second electron (subsequently symmetrized or antisymmetrized to
obtain a definite exchange symmetry). These wave functions are not normalizable, as we expect in
the continuous spectrum.
What happens to these discrete, doubly excited states when the perturbation 1/r12 is turned on?
It turns out that they do not remain discrete states, instead they become coupled to the continuum
states of nearby energy, so the actual spectrum of helium above the continuum threshold E = −2 is
strictly continuous. There are no exact bound states of helium above this threshold.
Nevertheless, the exact system does in effect know about the doubly excited states of the
unperturbed system. Consider, for example, what happens when we direct a beam of photons of
some energy at helium atoms in the ground state. When the photon energy exceeds the ionization
potential of helium, 0.904 a.u., it may be absorbed by the atom and an electron ejected. The
electron emerges with an energy that is the difference between the photon energy and the ionization
potential. The experimental cross section for this reaction is shown in Fig. 6. One can see sharp
peaks or enhancements in the cross section, which correspond to the doubly excited states of helium.
One can view this reaction as a two-step process, γ + He → He∗ → He+ + e− , in which He∗ refers
to a doubly excited state.
16 Notes 29: Helium

Fig. 6. Experimental cross section for the reaction γ + He → He+ + e− , as a function of photon energy. The peaks are
resonances corresponding to the doubly excited states of helium. See R. P. Madden and K. Codling, Astrophys. J. 141,
364(1965); and Robert D. Hudson and Lee J. Kieffer, Atomic Data 2, 205(1971).

This effect can be understood classically. The solar system has eight major planets that move in
approximately circular orbits that are stable on human time scales. Nevertheless, it is energetically
possible for Jupiter to move somewhat closer to the sun, moving into a lower energy state, and eject
all the other seven planets from the solar system. Will this ever happen? Modern chaos theory
indicates that it probably will, if we wait long enough, assuming that the motion of the planets is
governed by Newton’s laws and no other effects (such as the death of the sun or an encounter with
another star) intervene. Nevertheless, the solar system must be mostly stable for very long times,
since the earth has apparently been in approximately the same orbit for about 4 × 109 years.
Whether or not the ejection of the major planets is possible, it certainly happens that Jupiter
and other planets have encounters with other (usually smaller) objects and cause them to be ejected.
This effect is deliberately exploited by spacecraft that are to leave the solar system (such as the
Mariner spacecraft).
In the case of the doubly excited states of helium, we can imagine the two electrons having a
close encounter, which causes one to drop into the ground state and the other to be ejected. Unlike
the case of the solar system, where Jupiter can move into any state of (continuous) lower energy,
the first electron must drop into one of the discrete energy levels allowed to it. This is why the
 Notes 29: Helium 17

actual ground state of helium is stable: the two electrons are interacting strongly all the time, but
there are no lower states for one of them to drop into. Apart from the effects of quantization, the
classical picture gives an idea of the mechanism in helium. In particular, depending on the two sets
of quantum numbers of the doubly excited state, it may take a long time for one electron to have
a close encounter with the other. This is like the long-time stability of the solar system. In this
case we have a state of the system that is nearly a stationary state, that is, an energy eigenstate. It
is not exactly a stationary state, however, instead it is a resonance, a state with a finite life time.
Resonances were examined in homework problem 7.3 in a simple one-dimensional example.
Helium can be raised into a doubly excited state by many methods, not only absorption of
photons but also collisions with electrons or other means. In all cases the doubly excited state He∗
exists for a period of time and then decays into the helium ion He+ and a free electron. This is
called the Auger process.

10. Perturbation Analysis of the Ground State

The ground state wave function of H0 is of the form (28), with (nℓm) = (100), that is, it is
the state |100 100i. It is nondegenerate, so to compute the energy shift due to H1 in first order
perturbation theory we must compute

|ψ100 (x1 )|2 |ψ100 (x2 )|2

Z
∆Egnd = h100 100|H1 |100 100i = d3 x1 d3 x2 , (30)
r12

where we have written out the matrix element explicitly as an integral over the six-dimensional
configuration space and where the wave functions are hydrogen-like orbitals in the field of nuclear
charge Z.
The integral (30) has an interpretation in electrostatics. The squares of the wave functions
seen in that integral are probability densities for the two electrons in physical space, which may be
multiplied by the charges −e (= −1 in atomic units) to obtain charge densities. Then the integral
is the mutual energy of interaction of the two charge clouds. That is, if we imagine the two charge
clouds translated rigidly, starting at an infinite separation, then the integral is the energy required to
bring them into their final position where they overlap completely. Because the charge clouds repel
one another, this energy is positive. It is precisely the energy shift ∆E in first order perturbation
theory.
The hydrogen-like orbital for the ground state is
 Z 3 1/2
ψ100 (x) = e−Zr . (31)
π
To do the integral (30) we expand the Coulomb denominator,

X rℓ ∞
1 1 <
= = P (cos γ),
ℓ+1 ℓ
(32)
r12 |x1 − x2 | r> ℓ=0
18 Notes 29: Helium

where
r< = min(r1 , r2 ), r> = max(r1 , r2 ), (33)

and
γ = 6 (x1 , x2 ). (34)

See Sec. 15.11. The angle γ is illustrated in Fig. 4. In addition we expand the Legendre polynomial
in Eq. (32) in terms of spherical harmonics, using the addition theorem (15.71). For our present
purposes we write this theorem as

4π X ∗ (Ω ).
Pℓ (cos γ) = Yℓm (Ω1 )Yℓm 2 (35)
2ℓ + 1 m

Transforming the integral to spherical coordinates in both x1 and x2 and collecting all the
pieces, we have
∞ ∞
Z6
Z Z Z Z
∆Egnd = r12 dr1 dΩ1 r22 dr2 dΩ2 e−2Z(r1 +r2 )
π2 0 0
∞ ℓ
(36)
X r< 4π X ∗ (Ω )
× ℓ+1 2ℓ + 1
Yℓm (Ω1 )Yℓm 2
r
ℓ=0 > m

√
We do the Ω2 integral first, using the orthonormality of the Yℓm ’s and the fact that Y00 = 1/ 4π is
a constant,
∗ (Ω ) = √4π ∗ (Ω )Y (Ω ) = √4π δ δ .
Z Z
dΩ2 Yℓm 2 dΩ2 Yℓm 2 00 2 ℓ0 m0 (37)

The Kronecker deltas cause both sums to be replaced by the single term ℓ = m = 0. Then the Ω1
integration is easy,
Z √
dΩ1 Y00 (Ω1 ) = 4π. (38)

This leaves us with Z ∞

∆Egnd = 16Z 6 r12 dr1 S(r1 ), (39)
0

where S(r1 ) is the second integral,

∞ r1 ∞
e−2Z(r1 +r2 ) e−2Z(r1 +r2 ) e−2Z(r1 +r2 )
Z Z Z
S(r1 ) = r22 dr2 = r22 dr2 + r22 dr2 . (40)
0 r> 0 r1 r1 r2

These integrals are elementary and after some algebra we find

5
∆Egnd = Z, (41)
8
or
(1) 5
Egnd = −Z 2 + Z, (42)
8
 Notes 29: Helium 19

(0) (1) exact

Atom Z Egnd Egnd Egnd
H− 1 −1 −0.375 −0.528
He 2 −4 −2.75 −2.904

Table 3. Comparison of ground state energies at zeroth and first order of perturbation theory with the exact ground
state energies, for H− and He.

where the (1) means the energy correct to first order of perturbation theory. Evidently perturbation
theory is developing a series in inverse powers of Z.
Table 3 shows a numerical comparison of the ground state energies of helium and H − , as
estimated from the unperturbed system and from first order perturbation theory. It will be seen
that the first order correction does close the gap between the calculated energy and the true energy,
but it overshoots the mark and ends up being too high.
If we did not know the exact ground state energy, what would we make of this estimate from
first order perturbation theory? For example, do we have any idea how far away it might be from
the true ground state energy, or whether it is even too high or too low?
We know that the variational principle always gives us an upper bound to the ground state
energy. It turns out that the estimate from first order perturbation theory can be interpreted as
a variational estimate, and that it is, therefore, an upper bound. The trial wave function that
makes this identification is the unperturbed eigenfunction, which in this case is nondegenerate. See
Prob. 27.3. But perturbation theory alone does not give a lower bound, so we are somewhat in
the dark as to how far away our estimate is from the true ground state. In practice we can get
information about this by applying variational estimates with more and more sophisticated trial
wave functions, and observing the numerical convergence. This is not rigorous, but often it works
well enough.
In this example we do have a lower bound, namely, the ground state of H0 , which is −4. This
is not very close to the true ground state, nor can we improve the estimate as we can with the
variational method, by choosing better trial wave functions. There are methods of getting lower
bounds that can be improved, which we do not consider here.
Even without knowing how far away we are from the true ground state, however, an upper
bound can be useful for some purposes. Consider, for example, the H− ion. It is not obvious
on physical grounds that this exists as a bound state. If it were not known experimentally that
it does exist, how could we prove it theoretically? One way is to find a trial wave function that
gives an upper bound to the ground state energy that lies below the continuum threshold, which
is −0.5 a.u. Unfortunately, our estimate from first order perturbation theory, −0.375 a.u., lies above
this threshold, so we do not have a theoretical proof of the existence of H− . But we should keep
this example in mind later when we apply the variational method to helium-like atoms.
20 Notes 29: Helium

11. Perturbation Analysis of the Excited States

We now apply perturbation theory to the excited states of helium, concentrating on the bound
states. We need consider only the singly excited states, since, as noted, the doubly excited states
are already above the continuum threshold, and the perturbation can only raise the energies. We
will set (n1 , n2 ) = (1, n) for the singly excited states, so the unperturbed energy is

(0) Z2  1 
E1n = − 1+ 2 . (43)
2 n

We will be interested in the case n ≥ 2. We write the unperturbed energy eigenstates as

1
|N LM ±i = √ (|100 nℓmi ± |nℓm 100i), (44)
2

where we use the notation (25) for the products of hydrogen-like orbitals. This is case II, shown
orb
in Eq. (29). The ± sign on either side is the eigenvalue of E12 , the orbital exchange operator. As
for the notation N LM , we are using capital letters for the quantum numbers of all the electrons in
the atom, and lower case letters nℓm for the quantum numbers of a single electron. For example,
L is the quantum number of the operator L2 = (L1 + L2 )2 . But for singly excited states the entire
angular momentum comes from the second electron, since the first electron is in the ground state
for which ℓ = 0. Thus in Eq. (44) it is understood that L = ℓ, M = m and N = n. For example,
the state with N = 2, L = 1 and eorb 3
12 = −1 is the unperturbed version of the 2 P state in helium.
Whether we write nℓm or N LM depends on which aspect of the problem we wish to emphasize. In
any case, notice that the notation |N LM ±i contains quantum numbers of the orbital observables
in Table 2 that commute with the entire Hamiltonian H0 + H1 . That is, we are using a symmetry
adapted basis for studying the perturbation.
The unperturbed eigenstates (44) are degenerate, since the unperturbed energy depends only
on the quantum number N = n, and we should think about degenerate perturbation theory. But in
the basis |N LM ±i we need not diagonalize any matrices, since H1 commutes with all the operators
of the complete set of commuting observables that specify the unperturbed basis vectors inside a
degenerate eigenspace of the unperturbed Hamiltonian, that is, the operators L2 , Lz and E12 orb
.
Therefore the energy shifts are given by the diagonal matrix elements,

∆EN L± = hN LM ±|H1 |N LM ±i. (45)

The energy shifts do not depend on M because of the Wigner-Eckart theorem and the fact that H1
is a scalar operator, but they do depend on all the other quantum numbers N L±, as we shall see.
Substituting Eq. (44) into Eq. (45) we have

1
∆EN L± = [(h100 nℓm|H1 |100 nℓmi + hnℓm 100|H1 |nℓm 100i)
2 (46)
±(h100 nℓm|H1 |nℓm 100i + hnℓm 100|H1 |100 nℓmi)].
 Notes 29: Helium 21

But the pairs of matrix elements inside the round parentheses are equal, as we see by writing them
out as spatial integrals and swapping x1 and x2 , so we have

∆EN L± = Jnℓ ± Knℓ , (47)

where
1
Jnℓ = h100 nℓm| |100 nℓmi, (48)
r12
1
Knℓ = h100 nℓm| |nℓm 100i. (49)
r12
The integrals Jnℓ and Knℓ are called the direct and exchange integrals, respectively. As for the
direct integrals, if we write them out explicitly as a integrals over the configuration space of the two
electrons, using the hydrogen-like orbitals ψnℓm (x), we have

|ψ100 (x1 )|2 |ψnℓm (x2 )|2

Z
Jnℓ = d3 x1 d3 x2 . (50)
|x1 − x2 |
This has the interpretation as the mutual electrostatic energy of the two electron clouds associated
with orbitals (100) and (nℓm), exactly as in the integral (30), which applies to the ground state of
helium. In fact, that integral is a direct integral itself, with (nℓm) = (100). The direct integrals Jnl
are obviously real and positive.
As for the exchange integrals, they are

ψ ∗ (x1 )ψnℓm
∗ (x )ψ
nℓm (x1 )ψ100 (x2 )
Z
2
Knℓ = d3 x1 d3 x2 100 . (51)
|x1 − x2 |
These are also real, as is easily shown by swapping the variables of integration x1 and x2 . It can
also be shown that the exchange integrals are positive. We will omit the proof of this fact since it is
not particularly illuminating, but we note that it is plausible that Knℓ should be positive, since the
denominator makes the integrand large in regions of the six-dimensional configuration space where
x1 ≈ x2 . But in such regions, the numerator is approximately

|ψ100 (x)|2 |ψnℓm (x)|2 , (52)

where x ≈ x1 ≈ x2 . That is, the numerator is positive in such regions. (The rigorous proof that
Knl > 0 is given in an appendix to Slater’s books on atomic physics.)
The fact that Knℓ is positive has a simple physical interpretation, however. Since the energy
shifts (47) are Jnℓ ± Knℓ , it means that the ortho states (with the − sign) are lower in energy than
the para states (with the + sign). This is one of the qualitative features of the exact spectrum of
helium noted in Sec. 7. One can understand why this is so as follows. Let Ψ(x1 , x2 ) be the wave
function of the two electrons. For ortho states, Ψ(x1 , x2 ) = −Ψ(x2 , x1 ). But when x1 = x2 , this
means Ψ = 0. For ortho states, the two-electron wave function vanishes when the two electrons are
on top of each other, that is, the wave function has a node at x1 = x2 . For para states, on the
22 Notes 29: Helium

other hand, the wave function has an anti-node, that is, it is a maximum at x1 = x2 . (In using the
words “node” and “anti-node” we are applying one-dimensional terminology to the six-dimensional
problem at hand, but the terminology does convey the right idea.) But since the electrons repel one
another and do not like to be on top of one another, the electrostatic energy of interaction is larger
in para states than in ortho states.

21 P
2K21
23 P
21 S
2K20
23 S J21

J20

1s2s 1s2p

Fig. 7. Effect of the perturbation 1/r12 on the N = 2 levels of helium. Diagram is schematic and not to scale, but it
does show the correct ordering of the N = 2 levels in real helium, that is (from lowest to highest), 23 S, 21 S, 23 P , 21 P .

A schematic illustration of the effects of the perturbation on the N = 2 levels of helium is

given in Fig. 7. The unperturbed levels are labeled with the quantum numbers of the hydrogen-
like orbitals that enter into the unperturbed wave functions, that is, 1s2s and 1s2p. Such a list
of central force quantum numbers for individual electrons is called an electron configuration of the
atomic state. In the same notation, the ground state is given the configuration 1s2 (both electrons
are in the 1s orbital). Notice, however, that the electron configuration applies rigorously only to
the unperturbed wave function. For example, although we say that the electron configuration in
the ground state of helium is 1s2 , or that that of the N = 2 levels is 1s2s or 1s2p, nevertheless the
exact wave functions are not products of single electron orbitals (in the case of the ground state) or
symmetrized or antisymmetrized products (in the case of the excited states). This is because the
perturbation changes the wave functions.
A qualitative feature shown in Fig. 7 is the fact that the direct integrals Jnℓ are an increasing
function of ℓ. This is responsible for the “staircase effect,” that is, the fact that the P levels are
higher than the S levels. A similar effect can be seen in the alkalis (see Fig. 17.3 for the case of
sodium), and the explanation is the same: as the angular momentum of one electron is increased,
it moves further away from the nucleus, and the nuclear charge is more effectively screened by the
other electron(s) (one, in the case of helium; the core electrons, in the case of the alkalis).
We see that all the qualitative features of the excited states of real helium, discussed in Sec. 7,
are explained by perturbation theory. These are the absence of any 13 S state; the fact that the
ortho states are lower in energy than the singlet states; and the fact that the energy is an increasing
function of L (the staircase effect). From a quantitative standpoint, however, perturbation theory,
as we have outlined it, is less satisfactory, which is why we do not attempt to evaluate the direct and
exchange integrals explicitly. The variational method does better from a quantitative standpoint.
 Notes 29: Helium 23

12. Variational Treatment of Ground State

The unperturbed ground state was represented in ket language in Sec. 10 as |100 100i. In wave
function language it is the product of two hydrogen-like ground state orbitals (see Eq. (31)),

Z 3 −Z(r1 +r2 )
Ψ1s2 (x1 , x2 ) = e . (53)
π

where the subscript 1s2 refers to the electron configuration of the ground state. The unperturbed
energy (−4) is rather badly off. The reason was explained in Sec. 8: the two electrons repel one
another, and the true charge cloud stays somewhat further away from the nucleus than indicated by
Eq. (53).
Equivalently, we can say that each electron partially screens the nucleus from the other, so that
each electron sees an effective nuclear charge Ze that is somewhat less than the true nuclear charge
Z. Since only one electron is doing the screening and the screening is only partial, we would expect
any reasonable measure of Ze to lie between Z − 1 and Z. This suggests that we use as a trial wave
function the product of two hydrogen-like ground-state orbitals in a Coulomb field with effective
charge Ze , that is,
Ze3 −Ze (r1 +r2 )
Ψtrial (x1 , x2 ) = e , (54)
π
where we treat Ze as a variational parameter. In accordance with the variational method we will
minimize the expectation value of the energy with respect to Ze , and thereby determine Ze . This is
an example of how we can use physical reasoning to choose a trial wave function. We will write the
wave function (54) in ket language as |100 100(Ze)i.
The trial wave function (54) is already normalized, so we do not need any Lagrange multipliers
(see Sec. 27.5). We merely need to calculate the expectation value of H. To do this we write H in
the form,
 p2 Ze   p22 Ze  1 1 1
1
H= − + − + (Ze − Z) + + , (55)
2 r1 2 r2 r1 r2 r12
where we have added and subtracted terms to cause the appearance of two single-particle, hydrogen-
like Hamiltonians with nuclear charge Ze .
The expectation value of the first major term in Eq. (55) with respect to the trial wave function
is
 p2 Ze 
1
h100 100(Ze)| − |100 100(Ze)i
2 r1
 p2 Ze 
= h100(Ze )|(1) h100(Ze )|(2) 1 − |100(Ze )i(1) |100(Ze )i(2) , (56)
2 r1

where we have factored the two-particle wave function into a product of two single-particle orbitals,
and where the numbers in parentheses indicate which particle is referred to. But the operator in
the middle depends only on the coordinates of particle 1, so the bra and the ket referring to particle
24 Notes 29: Helium

2 just combine to give their normalization, which is 1. Thus the two-particle matrix element (56)
reduces to a single-particle matrix element,
 p2 Ze  Z2
h100(Ze )| − |100(Ze )i = − e , (57)
2 r 2
where we drop the (1) superscripts and 1 subscripts, since the coordinates x1 of the remaining
particle are just dummy variables of integration. This is the expectation value of a hydrogen-like
Hamiltonian with effective charge Ze with respect to its own ground state, which is just the ground
state energy with the same effective charge, as shown. The expectation value of the second major
term in Eq. (55) proceeds similarly, and gives the same answer, doubling the contribution (57).
As for the third major term in Eq. (55), we will need the expectation value of 1/r1 with respect
to the trial wave function. This is
1 1
h100 100(Ze)| |100 100(Ze)i = h100(Ze )|(1) h100(Ze )|(2) |100(Ze )i(1) |100(Ze )i(2)
r1 r1
1
= h100(Ze )| |100(Ze )i = Ze , (58)
r
where we proceed as in the previous paragraph to reduce the two-particle matrix element to a
one-particle matrix element, and where we use Eq. (24.39) for the final expectation value (with
Z replaced by Ze and n = 1). The expectation value of 1/r2 proceeds similarly, and doubles the
answer. Overall, the expectation value of the third major term in Eq. (55) is

2(Ze − Z)Ze = 2Ze2 − 2ZZe . (59)

As for the final term in Eq. (55), the inter-electron potential 1/r12 , this depends on the coor-
dinates of both particles and so its expectation value cannot be reduced to a single-particle matrix
element as in the case of the other terms. However, the expectation value in question is the same
one we calculated earlier in our perturbation analysis of the ground state (in Sec. 10) except for the
replacement of Z by Ze . Therefore the answer follows immediately from Eq. (41),

1 5
h100 100(Ze)| |100 100(Ze )i = Ze . (60)
r12 8

Adding up the pieces, we have

5
h100 100(Ze )|H|100 100(Ze )i = Ze2 − 2ZZe + Ze ≡ F (Ze ). (61)
8
We wish to minimize F (Ze ) with respect to Ze . We have

dF 5
= 2(Ze − Z) + , (62)
dZe 8
which has the root
5
Ze = Z − . (63)
16
 Notes 29: Helium 25

We see that by the variational criterion of the best trial wave function, each electron screens 5/16
of a nuclear charge from the other electron. The variational estimate for the ground state energy is
F (Ze ) where Ze is given by Eq. (63). This is

var 5 25
Egnd = −Z 2 + Z − . (64)
8 256
The first two terms agree with the results of first order perturbation theory, Eq. (42), but with
correction which is negative. Since the variational result is an upper limit to the ground state
energy, we see that
exact var (1)
Egnd < Egnd < Egnd , (65)

where the (1) superscript refers to the results of first order perturbation theory, as in Eq. (42).

(0) (1) var exact

Atom Z Ze Egnd Egnd Egnd Egnd
11
H− 1 16 −1 −0.375 −0.473 −0.528
11
He 2 1 16 −4 −2.75 −2.848 −2.904

Table 4. Comparision of ground state energies of He and H− , calculated by both perturbation theory and the variational
method, with the exact ground state energies.

Table 4 summarizes the results of both perturbation theory and the variational method for the
ground states of He and H− . In the case of helium we see that the variational method has closed
about two thirds of the distance between the results of first order perturbation theory and the exact
ground state energy. In the case of H− the variational result is disappointing, in that the estimate of
the ground state energy (−0.473) is still above the continuum threshold of −0.5. If we did not know
that H− exists as a bound state, we would not yet have a theoretical proof that it does. To give
that, it would be necessary to find a more complicated trial wave function (with more or different
parameters), such that the expectation value of the energy would be less than −0.5. From a physical
standpoint, it is clear what is wrong with our trial wave function (54): it does not incorporate the
tendency of the electrons to stay on opposite sides of the nucleus, that is, it does not incorporate
any correlation in the positions of the two electrons. With a trial wave function that does take this
effect into account, it is possible to obtain a variational estimate that is less than −0.5, thereby
providing a theoretical proof of the existence of H− as a bound state.
The variational method has been applied to the ground state of helium and helium-like atoms
with much more sophisticated and complicated trial wave functions than we have considered here.
The original calculations along these lines were carried out by Hylleraas, and with modern computers
it is possible to go much further. By such means the ground state energy of the Hamiltonian (1)
for Z = 2 and other values of Z has been obtained to many significant figures. The accuracy
obtained is far beyond that needed to show the difference between the ground state energy of the
26 Notes 29: Helium

Hamiltonian (1) and the experimental ground state energy of helium, which are not equal because
the Hamiltonian (1) omits a number of small effects, as discussed in Sec. 2.

Problems

1. The bound states of helium are shown in Fig. 3.

3
(a) How would this diagram be different if the electron were a spin 0 particle? If it were a spin 2
particle? You do not need to be quantitative, but if some levels go up or down, say which ones and
which way. If some levels appear or disappear, say which ones. What are the degeneracies of the
levels in actual helium? What would they be if the electron were a spin-0 particle?

(b) The electric dipole operator for a multielectron atom is defined by Eq. (23.12), where the charges
are the electrons (with q = −e). Electric dipole transitions between two states are governed by the
matrix element of the electric dipole operator between the states in question. Explain why there are
no electric dipole transitions between ortho and para eigenstates of the Hamiltonian (1).
Remark: The eigenstates of the Hamiltonian (1) are not exactly the eigenstates of real helium
because of the neglect of small terms, notably, the fine structure terms. When these are included,
it turns out that electric dipole transitions are allowed between ortho and para states, but the
amplitude is small and the lines are very weak. For this reason, it was thought at one time on
spectroscopic grounds that parahelium and orthohelium were two different species of helium. The
weak spectral lines connecting the ortho and para states are called “intercombination lines.”

2. This problem was borrowed from Eugene Commins. Consider a simple “one-dimensional” hydro-
gen atom that obeys the equation,

1 ∂2ψ
− − Zδ(x)ψ = Eψ. (66)
2 ∂x2
[Units such that m = h̄ = 1, and the Coulomb potential is replaced by −Zδ(x)].

(a) Find the ground state energy and wave function ψ0 (x), and verify that

1
hT i = −E = − hV i. (67)
2

(b) Now consider the one-dimensional “helium” atom which obeys

1 ∂2ψ 1 ∂2ψ
Hψ(x1 , x2 ) = − − − Zδ(x1 )ψ − Zδ(x2 )ψ + δ(x1 − x2 )ψ = Eψ, (68)
2 ∂x21 2 ∂x22

where x1 , x2 are the coordinates of the two “electrons” on the x-axis. First treat δ(x1 − x2 ) as a
perturbation and find the ground state energy to first order. Compare to 3-dimensional helium.
 Notes 29: Helium 27

(c) Now employ the variational method with a trial wave function analogous to that used in class for
the ground state of 3-dimensional helium. Find the best value of hHi and compare to 3-dimensional
helium.

(d) For one-dimensional “helium,” take a trial wave function of the form

ψ(x1 , x2 ) = u(x1 )u(x2 ), (69)

where u(x) is the variational parameter (that is, the whole function, as in Hartree theory). Assume
u(x) is real; there is no loss of generality in this. Find an equation for u(x). It will be a kind
of Hartree equation, and it will contain a pseudo-energy eigenvalue, call it ǫ. Define ǫ so that the
equation looks like
1
− u′′ (x) + . . . = ǫu(x), (70)
2
where the ellipsis indicates omitted terms.

(e) The desired solution must be normalizable (in fact, we demand that it be normalized), so we
must have
lim u(x) = 0. (71)
x→±∞

Show that this can only be satisfied if ǫ < 0. Hint: To solve the equation, make the change of
notation, u → x, x → t, and interpret it as a one-dimensional problem in classical mechanics.

(f ) Find the normalized solution u(x) and the pseudo-eigenvalue ǫ.

(g) Find the variational estimate of the ground state energy of one-dimensional “helium”. Hint:
Note that
hHi = 2ǫ − hδ(x1 − x2 )i. (72)
Does your new estimate improve on the results of part (a)? Would you expect it to do so?

3. A problem on the excited states of helium.

(a) The exchange integral Knℓ defined in Eq. (51) depends on two hydrogen-like orbitals. It is
easily generalized to any two single-particle orbitals (recall that “orbital” means a single-particle
wave function). Show that if the two orbitals have no spatial overlap, then the exchange integral
vanishes. Then use your knowledge of hydrogen-like orbitals to explain the dependence of the
exchange integral on n for fixed ℓ, and on ℓ for fixed n.

(b) Go to http://physics.nist.gov/PhysRefData/ASD/levels form.html and

check your predictions using the experimental data on the first several excited states of helium.
Enter “He I” for the atom (this is neutral helium). You will see that some of the levels have a fine
structure, which we did not discuss in detail in class. This should not prevent you from checking
your predictions.